Legacy LSAM connectors (UNIX/Linux, Windows, IBM i, SQL, SAP R/3, SAP BW, z/OS, MCP and OS 2200)
- Synopsis
- Full Details
In briefThe legacy LSAM connectors let an OpCon Continuum workflow run work on the UNIX, Windows, IBM i, SQL, SAP R/3 and CRM, SAP BW, z/OS, Unisys MCP and Unisys OS 2200 LSAM machines you already have, through a relay. Each platform offers several job types; the relay builds what the agent needs from the job's fields, and the agent reports the outcome back.
WhereIn the workflow editor: Workflows, open the workflow, add or open a job, then the Job Definition tab › Job Type list. Every legacy LSAM job type is there — UNIX, Windows, IBM i, SQL, SAP R/3, SAP BW, z/OS, MCP and OS 2200. Its fields appear under Job Parameters, the machine under Agent Assignment, and the account under Connection. The result of a run is on the job's Summary tab, in the Termination and Agent Message rows, and what the agent reports while the job runs is in its Agent Feedback section — an SAP R/3 job's children in a Child Processes table above it.
When
- A workflow has to drive work that already runs on an existing UNIX, Windows, IBM i or SQL LSAM machine — a command, a file transfer, a file wait, a T-SQL, Oracle or MySQL script, an SSIS package, SQL through an ODBC driver or OLE DB provider, an IBM i batch job, an existing SAP R/3 or CRM background job, an SAP BW process chain, a z/OS JCL member submitted through JES — held until its datasets, a console message, a job or task, tape units or a REXX check are ready, if you need — a z/OS started task, REXX exec or operator command, a Unisys MCP program, WFL job, EAE/AB Suite report or file copy, rename or removal, or a Unisys OS 2200 ECL runstream.
- A z/OS job is submitted on the mainframe outside OpCon, or held there by the agent, and you want it on its schedule — a z/OS Tracked Job or z/OS Queued Job the agent's report lands on.
- An MCP or OS 2200 job must wait for something on its own system first — a Prerun, or on OS 2200 File Dependencies, that the agent checks before it starts the job.
- A legacy job was refused at dispatch, failed where you expected it to finish, or reported an exit code you need to interpret.
- You are writing exit criteria, or an Exit Description or LSAM Feedback event, against what a legacy job reports.
- A UNIX Command, SAP R/3 Job or SAP BW Process Chain failed part-way and you want to restart it at the failed step — or a z/OS Batch Job or z/OS Started Task, from a start step to an end step.
- You want to read or fix the JCL member a z/OS Batch Job submits, or a z/OS REXX Exec's exec, without leaving Continuum — View / edit JCL.
- Some child jobs of an SAP R/3 Job failed and you want to restart only those, from its Child Processes table.
- You want to pick an SAP job, an application server or a process chain from the SAP system rather than type it — Search… beside SAP Job Name, Exec. Target or Process Chain Name.
- SAP R/3 Jobs may hold SAP job numbers SAP no longer has, and you want to find and correct them
before they fail with
70006— Check SAP Job Numbers in Workflows.
Use something else when the work runs on a Universal Agent — the Run Command, Run Script and Wait for File job types — or when it moves a file between two legacy machines, which is SMAFT File Transfer.
How
- Make sure the relay is current and the LSAM machine is registered with it.
- Add a job to the workflow and choose its job type in the Job Type list.
- Set the machine and the job's parameters, and pick the Connection the job runs as, where the job type takes one.
- Add exit criteria where the job type offers them.
- Save, run, and read the Termination row; a dispatch refusal names the field.
The full steps are in Run commands on legacy LSAM machines.
- Most field values are checked by the relay when the job runs, not when you save it. A refused job fails before anything reaches the machine, and the message names the field but never its value. An OS 2200 Job is checked when you save it too: the editor refuses one the agent could not be sent.
- A blank Connection means something different on each platform; on a UNIX job, on every MCP job, and on MS SQL Job, Oracle and MySQL, it is not allowed.
- The six SQL job types are listed under SQL - in the Job Type list, and their forms show only the fields your choices use — Script Source, Connection Option and Package Location each hide, and clear, the fields that do not apply.
- An Oracle, MySQL or Other DB job needs SQL*Plus, the mysql client, or the ODBC driver or OLE DB provider, installed on the agent machine.
- Fifteen job types take an exit-criteria table, and on the six SQL types, UNIX - Episys: Run JobFile, UNIX - Episys: Find Report from RSJ Output, Windows - WS_FTP Pro and Windows - File Arrival Continuum always applies it itself. IBM i jobs report a message ID instead, which becomes the job's termination description, and an SAP R/3 Job or SAP BW Process Chain reports the agent's result text, whose leading code is its exit code; when it fails, its Agent Message says what that code means. A z/OS job is judged by its agent — against the job's condition-code or return-code range and its Step Control rows, or, for a console command, finished once the command is issued — an MCP job by its agent against the fail codes you set, and an OS 2200 Job by its agent against its Completion Status rules; in each, the agent's text becomes the termination description.
- Newer job types and fields need a current relay. An older one refuses the job at dispatch rather than run it without them — IBM i lists, MCP preruns, z/OS Step Control, pre-runs and Restart options, and OS 2200 preruns and file dependencies included.
- Never put a password in an Other DB connection string, where only common key names are refused, or in a MySQL user variable, which the agent's log shows. The login belongs in the batch user.
- An IBM i File Arrival job checks for the file once unless a
$@RECHKFREQdynamic variable sets a re-check interval. - The machine must be an agent of the job type's own platform. A UNIX job pointed at a Windows agent is refused at dispatch with LSAM_AGENT_TYPE_MISMATCH.
- Kill on an SAP R/3 Job is refused until the agent's Allow Kill is turned on, and on an SAP BW Process Chain it is always refused.
- An SAP BW Process Chain's name is checked against its 25-character limit only when the job runs.
- An MCP job or an OS 2200 Job restarts from the beginning only; a restart that names a step is refused. So does a z/OS REXX Exec or z/OS Console Command.
- Saving JCL needs z/OS JCL Edit and a z/OS user ID assigned to you; viewing it needs Agents Execute under All workspaces and an agent with a JORS Port.
- A z/OS job's pre-runs — dataset, message, job/task, tape and REXX — need a current relay. An older one refuses the job at dispatch, so it goes to Initialization Error instead of waiting.
- An MCP job must name a Connection, and an MCP agent's name must end with a period.
- Kill on an OS 2200 Job is always refused.
- A z/OS or OS 2200 job's output can be listed only when its agent reports where it is, and only once the agent's JORS Port is set. An MCP job's can be listed while it runs, too.
- Search… on an SAP field is there only with a Specific Agent assignment and a Primary Agent chosen, and it and Start Check on Check SAP Job Numbers need Agents Execute granted under All workspaces.
Task walkthrough: Run commands on legacy LSAM machines. This page is the full configuration and troubleshooting reference.
These connectors run commands on existing legacy LSAM agents — the agents many customers already run before moving to the Universal Agent. They let an OpCon Continuum workflow drive UNIX/Linux, Windows, IBM i, SQL, SAP R/3 and CRM, SAP BW, z/OS, Unisys MCP and Unisys OS 2200 LSAM machines without rebuilding that automation.
| Connector | Where it is documented |
|---|---|
Windows LSAM (opcon.lsam.windows) | The Windows tab |
UNIX/Linux LSAM (opcon.lsam.unix) | The UNIX/Linux tab |
IBM i LSAM (opcon.lsam.ibmi) | The IBM i tab |
SQL LSAM (opcon.lsam.sql) | The SQL tab |
SAP R/3 and CRM LSAM (opcon.lsam.sap-r3) | The SAP R/3 tab |
SAP BW LSAM (opcon.lsam.sap-bw) | The SAP BW tab |
z/OS LSAM (opcon.lsam.zos) | The z/OS tab |
MCP LSAM (opcon.lsam.mcp) | The MCP tab |
OS 2200 LSAM (opcon.lsam.os2200) | The OS 2200 tab |
What every legacy job shares, whatever its platform (how a job reaches its machine, the Batch User, the shared parameters, exit criteria), is on the How It Works tab, and outcomes and troubleshooting for all of them on the Troubleshooting tab. Each platform tab opens with the job types its connector offers.
SMAFT File Transfer (opcon.lsam.smaft) also runs on legacy agents through the relay, but it
belongs to none of the per-platform sections below: a single job names two machines — Windows,
UNIX or z/OS — and which of the two runs the job is derived rather than assigned. It has its own page —
SMAFT File Transfer job.
- How It Works
- Windows
- UNIX/Linux
- IBM i
- SQL
- SAP R/3
- SAP BW
- z/OS
- MCP
- OS 2200
- Troubleshooting
How It Works
- Overview
- Connection
- Shared Parameters
- Exit Criteria
- Arrived Files
Overview
How it works
Unlike built-in jobs, these don't run on a Universal Agent. The platform dispatches the job through the relay, which forwards it to the legacy LSAM machine; the LSAM runs the work and reports completion back.
Every hop is the relay's; no Universal Agent appears anywhere in the path. Two consequences matter:
- A working relay and reachable LSAM machines are required. An LSAM command that finds no agent to run on points at the relay or LSAM connectivity, not the Universal Agent.
- The job is routed by its machine name — the name of the agent picked under Agent Assignment — which must match the LSAM agent name registered with the relay. A mismatch means the job can't be placed.
- The agent must be of the job type's own platform. A job whose machine points at an agent of another type — a UNIX job aimed at a Windows agent, say — is refused at dispatch with LSAM_AGENT_TYPE_MISMATCH, rather than sent to an agent that would misread it. See A legacy job aimed at an agent of another type.
- Legacy agents are now registered in-product, on the Agents tab of the Agents page: register, edit, or delete a legacy agent there. The relay hot-reconciles its LSAM connections within one heartbeat of a change — no relay restart needed. Agent names are unique per relay (a duplicate is rejected).
- A legacy agent's status can be Online, Offline, or Unknown. Unknown (yellow) means the relay went stale, so the agent's reachability can't be confirmed — distinct from Offline (the agent itself is down). On relay recovery, agents leave Unknown within a heartbeat. See Agents and pools.
The schedule date every legacy job is sent with
Every legacy job carries the schedule date of the workflow instance it belongs to — a plain calendar date, with no time and no time zone, exactly as Classic sends it. The agent reads it as a local date, and two things on the machine depend on it:
- the File Arrival watch window, whose day offsets are counted from midnight of this date — see Authoring the watch window (on the Windows tab, under File Arrival);
%SMA_MSLSAM_SCHEDULE_DATE%, which a Windows agent substitutes into paths for you, including the paths it reads exit criteria and failure criteria from.
The relay used to stamp this field from its own clock in UTC rather than from the job's schedule
date. For any agent west of UTC, an evening dispatch therefore carried tomorrow's date — so a
File Arrival job watched the wrong day's window, and a path built from
%SMA_MSLSAM_SCHEDULE_DATE% named the wrong day's file. Agents east of UTC were unaffected in the
evening and affected in the early morning instead.
The relay sends the workflow's schedule date now. It needs a current relay for this: an older one still uses its UTC clock — see Keeping a relay current. If the platform cannot supply the date for a particular dispatch, the relay falls back to the UTC date and records a warning in its log rather than refusing the job.
The names the agent is sent
Every legacy job is dispatched with three names alongside its schedule date: the job name, the workflow name (the agent calls it the schedule name) and the frequency name.
These are your real names now. Earlier builds sent an internal identifier in place of the job
name and sent no workflow name at all, so an LSAM machine's own records showed something like
job-0a608a27-0e09-4da6-98cb-0daf05951c85 for every job and nothing at all for the workflow. If
you have scripts or housekeeping on a machine keyed off what the agent reported, this is the change
to check.
Three consequences on the machine itself:
- A UNIX job's script can read them. The UNIX agent exports
SMA_USER_SPECIFIED_JOBNAME,SMA_SCHEDULE_NAMEandSMA_SCHEDULE_FREQUENCYinto the job's environment. Because they are written into a shell script, the characters` $ " ' \ ; & | < > ( ) ~and any tab or line break in a name are replaced with_before the name is sent — a job namedLoad(daily)arrives asLoad_daily_. The name you authored is unchanged; only the copy the agent sees is scrubbed. - Windows output file naming picks the workflow name up. A Windows agent configured with
UseScheduleNameInOutputFile=TRUEnow has a name to use, where before it had none. - IBM i still receives a sanitised name, because IBM i object names are constrained: the name
is upper-cased, anything illegal becomes
_, and a name that does not begin with a letter gains a leadingJ. SoDaily Backupreaches an IBM i agent asDAILY_BACKUP, and2026 reportasJ2026_REPORT.
A name longer than 128 characters is truncated on the wire, on every platform. That is the width the LSAM job monitor can hold, and a longer one would overflow it. Your name is not changed — only what is sent is shortened.
What the agent reports while the job runs
A legacy agent sends more than a status with each update it makes about a running job. Continuum keeps that detail on the job as its agent feedback:
| Feedback | Platforms | What it holds |
|---|---|---|
| Job Status Description | Every legacy platform | The status text the agent sends with its updates. A new value is added each time the text changes, so it reads as the job's status history in the agent's own words. The newest 64 are kept |
| Child Process | SAP R/3 and CRM | One value for each child job the SAP job started, updated as that child's status changes |
| EXEC Target | SAP R/3 and CRM | The application server the SAP job ran on, reported when it ends |
| Job Number | SAP R/3 and CRM | The SAP job number, reported when it ends |
| Process | SAP BW | Detail the SAP BW agent reports about the processes of a running process chain, as the agent words it |
| Process Chain Log | SAP BW | The log ID of the process-chain run, reported once the chain has started |
| User Message | z/OS | A user message the z/OS agent reports for the job, as the agent words it |
| Step Completion | z/OS | What the z/OS agent reports about a step of the job as it completes, as the agent words it |
| Trigger Messages | z/OS | A trigger message the z/OS agent reports for the job, as the agent words it |
| Mix Number | MCP | The task's mix number |
| Process Name | MCP | The task's process name |
| Task Processor Time, Accum Processor Time | MCP | Processor time for the task, and accumulated |
| Task I/O Time, Accum I/O Time | MCP | I/O time for the task, and accumulated |
| Task Print Lines, Accum Print Lines | MCP | Print lines for the task, and accumulated |
| Task Elapsed Time, Accum Elapsed Time | MCP | Elapsed time for the task, and accumulated |
| Task Ready Q Time, Accum Ready Q Time | MCP | Ready-queue time for the task, and accumulated |
| Task Elapsed Wait Time, Accum Elapsed Wait Time | MCP | Elapsed wait time for the task, and accumulated |
You see it on the job while it runs. The job's Summary tab carries an Agent Feedback section, below Completion, whenever the job has any: one row per kind of feedback, its values in the order the agent sent them. An SAP R/3 job's Child Process values are shown instead in a Child Processes table directly above it — see Restarting only the failed child processes. While the job is running its detail is refreshed every 15 seconds, so a SAP job's children appear and change as they run. Each run's feedback is also kept with that run: open it from Job Execution History and the run's Job History Detail shows it, with Child Process in the Agent Feedback section.
The newest status text is also shown on its own. While a legacy job is running, its Summary tab shows a Status Description row in the Execution section, directly after Status, with the agent's latest status text as the agent sent it. For an MCP job that text begins with the task's mix number. The row goes away once the job ends, because Termination then carries the final text, and a restarted job does not show the previous run's text.
An event can fire on it. An LSAM Feedback event fires when a value of the kind you choose appears or changes and matches a pattern, while the job is still running.
It belongs to the run. A restart, a failure retry or another dispatch of the job clears it, so the next run starts with none, with one exception: an SAP R/3 job's Child Process values are kept across a restart and a retry. They are cleared when a recurring job runs again.
A job keeps up to 256 values sent by its agent, besides its Job Status Description, and each value up to 1,024 characters; a longer one is cut there. Agent feedback needs a current relay: an older one passes none of it on.
Connection
The Connection field — the account the job runs as
Every legacy job type has a Connection field (earlier releases labelled it Batch User) except nine: SAP R/3 Job and SAP BW Process Chain, which sign on with the logon configured on the agent, z/OS Batch Job and OS 2200 Job, whose User ID field names the user instead, and z/OS Started Task, z/OS REXX Exec, z/OS Console Command, z/OS Tracked Job and z/OS Queued Job, which name no user. The field names a saved connection of its platform's batch-user type, and that connection is where the account and — on Windows and SQL — its password live. An MCP batch user has no password: it is a usercode, and an access code, when one is needed, is entered on the job. The job definition holds only a reference; no credential is stored in the workflow.
| Platform | Connection type to pick from | Blank means |
|---|---|---|
| Windows | Windows Batch User (batch-user-windows) | The job runs as the agent's own service account. A legitimate, deliberate setting, and the only way to choose it: a Windows Batch User connection is always a named account. |
| UNIX/Linux | UNIX Batch User (batch-user-unix) | Not allowed — a UNIX job must name a batch user. |
| IBM i | IBM i Batch User (batch-user-ibmi) | The job runs under the job description's default user. |
| SQL | SQL Batch User (batch-user-sql) | The job runs as the agent's own service account and does not impersonate — whichever way the job's Windows Authentication setting is set. MS SQL Job, Oracle and MySQL are the exceptions: each requires a batch user, and a job with none is refused at dispatch. |
| MCP | MCP Batch User (batch-user-mcp) | Not allowed — every MCP job must name a batch user. |
A job takes one batch user. (An MCP Start job can name a second MCP Batch User, in its own
Job Summary User field, for the usercode its job summary is kept under, and an MCP job whose
prerun is a Start prerun can name one in Prerun Job Summary User (Start) for the prerun's job summary.) The picker offers only connections of the matching type, and a
connection of the wrong platform's type is rejected at dispatch
(RC_WRONG_CONNECTION_TYPE) rather than degrading to the service account — so a mismatch fails
loudly instead of running under an identity nobody chose.
The field values, character rules, and password rules for each type are in Connections. To vary the credential per environment, transform the job's connection reference — not the Connection parameter, which is a display mirror that no transformation rule may change. See Versions and deployments.
Earlier builds took the run-as identity from the job's own parameters: Windows accepted only the
literal Use Service Account, UNIX carried free-text user and group fields, and IBM i a free-text
user field. Those fields are gone. A UNIX or IBM i job saved before the change keeps its orphaned value
in the stored configuration, but nothing reads it — re-pick the account from the Connection
field, or a UNIX job will fail at dispatch with RC_NO_UID.
Shared Parameters
Shared parameters
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target LSAM machine; must match the agent name registered with the relay. |
| Connection | See Connection | a saved connection | The account the job runs as. Every legacy job type carries this field but the nine that Connection lists. |
| Start Image | Yes | 1–4009 chars (Windows: 1–4000) | Path to the program or script to run. On Windows, quote a path that contains spaces. |
| Parameters (Windows: Arguments) | No | ≤2005 chars (Windows: ≤2000) | Command-line arguments, appended to Start Image with a space. On Windows Command the field is labelled Arguments. |
| Pre Run Image | No | ≤2005 chars (Windows: ≤4000) | A prerun command run before the main job. A prerun that fails does not fail the job — see A prerun that fails is re-attempted. |
| Exit Criteria | No | ≤20 conditions | Conditions tested against the job's exit code. Leave empty for the default: fail on any non-zero exit code. See Exit criteria. |
| Exit Criteria Result | No | Fail / Finish OK | What a match decides for the whole table. Default Fail. See Exit criteria (on the How It Works tab, under Exit Criteria). |
On Windows, the Start Image, Arguments, Working Dir, Pre Run Working Dir and Pre Run Image max-lengths follow the legacy Windows job limits (command line 4000, directories 255). The UNIX limits differ — see the table above.
A prerun that fails is re-attempted
A prerun command runs on the agent before the job itself. When it fails, the job does not fail: it moves to Pre-run Failed, fires the Job Prerun Failed trigger, and is started again from the beginning about three minutes later — as many times as it takes.
That makes a prerun usable as a gate. A prerun that tests whether the job's inputs are ready holds the job until they are, rather than failing it on the first miss.
Because the job never ran, its own outcome is not decided: the prerun's exit code is never put to the job's exit criteria (on the How It Works tab, under Exit Criteria), and nothing downstream is released — a dependency on Failed does not see a failed prerun. The agent's message is kept on the job while it waits, so you can see why the prerun failed.
See A failed prerun is re-attempted, not failed for what an operator can do to a job waiting this way, and the two cases that end it instead.
While the prerun runs, the job reads Pre-run Active. When the agent reports that it is running the job's prerun, on any legacy platform, the job moves to Pre-run Active — from Start Attempted when the prerun is the first thing the agent reports, or from Running — and on to Running when the agent reports the job itself running. It is still a running job: Late to Finish and Max Run Time are checked against it as they are on Running, and an operator can Kill it. The job sends one Job Running notification per run, so a job that was already Running before its prerun sends no second one. This needs a current relay; on an older one the job reads Running throughout. See Job and workflow statuses.
A legacy job reads Running only once the agent reports it. Until the agent first reports the job running, or running its prerun, the job reads Start Attempted, and after about five minutes Still Attempting Start — so a job behind a disconnected relay or a stopped agent no longer reads Running. A relay from before this change does not report the start: its jobs read Running as soon as the relay takes them. See A job is Running only once its agent confirms it.
Exit Criteria
Exit criteria — deciding what an exit code means
Left alone, a legacy job finishes OK on exit code 0 and fails on anything else. Exit
criteria replace that default with a table of your own conditions, so a program whose codes do not
follow that convention still reports the outcome you mean.
Fifteen job types offer the table: UNIX Command, UNIX - Embedded Script, UNIX - Episys: Run JobFile, UNIX - Episys: Find Report from RSJ Output, Windows Command, Windows - Embedded Script, Windows - Web Services, Windows - WS_FTP Pro, Windows - File Arrival, SQL - MS SQL Script, SQL - MS SQL Job, SQL - MS SQL DTExec, SQL - Oracle, SQL - Other DB and SQL - MySQL.
Every other LSAM job type carries none at all — the five IBM i types, the four file operations, Corelation, both Fiserv DNA types, the UNIX File Arrival type, the other five Episys types, SAP R/3 Job, SAP BW Process Chain, the six z/OS job types, the six MCP job types, OS 2200 Job and SMAFT File Transfer. For those the outcome the machine reports decides on its own. A z/OS Batch Job or z/OS Started Task sets the condition-code range its agent judges by instead, and a z/OS REXX Exec the return-code range — see z/OS Batch Job and z/OS REXX Exec. An MCP job's agent judges by the fail codes the job sets, and an OS 2200 Job's by its Completion Status rows.
The table
| Field | Limits | Notes |
|---|---|---|
| Exit Criteria Result | Fail (default) / Finish OK | What a match decides — one setting for the whole table, not one per row. |
| Criteria rows | ≤20 | Each row is an operator, a Value, and — for Range only — an End Value. |
| Operator | Equal To, Not Equal To, Less Than, Greater Than, Less Than or Equal To, Greater Than or Equal To, Range | Range is inclusive at both ends: it matches when the exit code is between Value and End Value. |
| Value, End Value | -2147483648 to 2147483647 | Whole numbers. End Value must be greater than or equal to Value. |
End Value appears only when the row's operator is Range. On every other operator the field is hidden, because it has no meaning there — the row keeps its place in the table rather than reflowing when you change the operator.
The rows are an OR — the exit code matches the table when it matches any one row.
Exit Criteria Result then says what a match decides, and a code matching nothing takes the opposite. There is no third "nothing matched" outcome:
| Exit Criteria Result | A row matched | No row matched |
|---|---|---|
Fail (the default) | Failed | Finished OK |
Finish OK | Finished OK | Failed |
Leave the table empty and the default stands: fail on any non-zero exit code. Leave Exit Criteria
Result unset and it means Fail, which is how every job saved before the setting existed behaves —
so adding the setting changed no existing job.
Reach for Finish OK when the list of good codes is shorter than the list of bad ones. "Finish OK
on 0, on 3, or anywhere from 10 to 20" is three rows and one setting; written as failure
conditions it cannot be expressed at all.
Where the table is evaluated
Two layers can decide a legacy job's outcome, and which one does depends on what the table needs:
- The agent decides when the installed LSAM can enforce the table itself — that is Exit Criteria
Result
Fail, no range row, and no more than five rows. The table travels to the agent with the job, exactly as it always has. - Continuum decides, after the agent reports, for anything else:
Finish OK, a Range row, or rows six through twenty. The agent is sent the platform default instead (fail on a non-zero code), and Continuum then applies your real table to the exit code the agent reported. - All six SQL types are always decided by Continuum. The SQL agent has no way to receive criteria, so central evaluation is the only route to a criteria table on that platform — for MS SQL Script, MS SQL Job, MS SQL DTExec, Oracle, Other DB and MySQL alike, including a plain one-row table.
- UNIX - Episys: Run JobFile and UNIX - Episys: Find Report from RSJ Output are always decided by Continuum too. The relay never sends their table to the agent: the agent always runs the job with the default (fail on a non-zero code), and Continuum applies your table to the code it reports. That is how Classic treats these job types, and it means their tables work on any relay build.
- Windows - WS_FTP Pro is always decided by Continuum as well, for the same reason: the relay never sends its table to the agent, which runs the transfer with the default, and Continuum applies your table to the exit code the agent reports. A WS_FTP Pro table works on any relay build.
- So is Windows - File Arrival. The Windows agent always judges a file-watcher job by the default, whatever it is sent, so the relay sends no table and Continuum applies yours to the watcher's exit code. A File Arrival table works on any relay build.
When Continuum decides, three things are visible on the job:
- The termination text of a Windows, UNIX/Linux or SQL job keeps the exit code — see
The exit code is also the job's termination description.
Finished OK (Exit Criteria) Return Code='N'orFailed (Exit Criteria) Return Code='N'is written there only when the job has no termination text of its own. - The job's Status History tab in Processes records an Exit Criteria Evaluated entry holding the exit code, the Exit Criteria Result in force, which row matched (or that none did), the verdict, and the status the agent had reported.
- The agent's own log reaches the opposite conclusion on a job that finished OK by criteria, because the agent was given the default. The Job History entry is where to see which layer decided.
Two outcomes Continuum never overrides, both deliberate:
- A job an operator killed. A Kill is never undone by a table that finishes everything OK.
- A completion that carried no exit code — an agent that reported none, or a relay-side infrastructure failure. With nothing to test, the reported verdict stands.
A Run JobFile or Find Report from RSJ Output job the UNIX agent could not start at all
stays Failed as well. The agent reports a start failure with a code of its own, 10000 or above,
so Continuum leaves any code from either job type outside the range a UNIX process can return (-128 to 255) unjudged rather than test it
against your table. A WS_FTP Pro job the Windows agent could not start — because wsftppro.exe
was not found, the process could not be created, or the agent could not log on as the job's
account — stays Failed the same way: the agent reports a code of its own for it, 30000 or
31000 to 31017, and Continuum leaves those codes unjudged. A Windows - File Arrival job the
agent could not start — the watcher program was not found, or the agent could not log on as the
job's account — stays Failed for the same reason. The watcher's own codes, -1 included, are
always judged by your table.
Finish OK, a range row, and rows beyond the fifth are all carried by relay changes. A relay on an
older build refuses the dispatch rather than sending a table it cannot represent, so the job does
not start — which is the safe direction, but it does mean the job stays un-run until the relay is
upgraded. Tables of up to five plain conditions with the default Fail are unaffected and dispatch
on any relay build, and so does every Run JobFile, Find Report from RSJ Output,
WS_FTP Pro and Windows File Arrival table, which never travels to the agent.
What is rejected
The table is validated when you save the workflow and again at dispatch. Rejected outright:
- more than 20 conditions — the twenty-first is refused rather than silently dropped;
- a missing or unrecognised operator, or a value that is not a whole number in range;
- a range row with no End Value, or with an End Value below its Value. Either shape can never match anything, so the inverse rule above would quietly hand the job the opposite outcome. The workflow refuses to commit and names the row: Range criterion #2 needs an End Value.
On UNIX, Signal Criteria is a separate table and did not change: still five rows, still the six basic operators, no range operator and no Exit Criteria Result. It tests the signal that stopped the process, not an exit code.
Unless the LSAM's LSAM_0_255 option is enabled — it is off by default — a UNIX exit code of 128 or
more reaches OpCon Continuum as a negative number. Write the criteria against the value the agent
actually reports, which is visible as the job's Return Code.
The exit code is also the job's termination description
A Windows, UNIX/Linux or SQL legacy job that reports an exit code now carries that code as its termination description, whether it finished OK or failed — the same thing Classic records for these jobs.
This is what makes an Exit Description workflow event
usable on a legacy job. The trigger compares the termination description, and until now a legacy job
had nothing to compare: a successful job's description was blank, and a failed one read Job failed.
A trigger written for 0 therefore never fired, on any legacy job. Writing the exit code there fixes
that, so a trigger can key off a specific code.
One case keeps its text instead: a failure with no exit code — the relay could not reach the agent, or the job never produced one. The message describing that failure is more use than a blank, so it stays.
A job whose outcome exit criteria decided keeps its exit code too. Deciding the
verdict centrally used to replace the description with
Finished OK (Exit Criteria) Return Code='N' / Failed (Exit Criteria) Return Code='N', which
overwrote the very code an Exit Description event compares — so such an event could never fire on
a job with an exit-criteria table, including an Equal To 0. The exit code now stays, on either
outcome, and the evaluation is recorded in the job's status history (the Exit Criteria Evaluated entry on the Status History tab) rather
than over the description. Classic behaves the same way.
For a job type that is not one of the three above, exit criteria still writes that line as the description, except on a failure where the agent gave a reason of its own — an OpCon MFT failure message, say — which is kept.
SMAFT file-transfer jobs are not included in the exit-code-as-description rule. Their Classic termination text has not been confirmed, and writing a plausible value would be worse than writing none.
An IBM i job carries its message ID instead. IBM i reports no numeric exit code, so its
termination description is the message ID the agent sent — SMA0036, SMA0037 and so on — on
success and failure alike. See
The message ID is the job's Exit Description (on the IBM i tab, under Batch Jobs).
A z/OS job carries the agent's text. A z/OS job, of any of its six types, has no numeric exit code either, so its termination description is the text the z/OS agent reports when the job ends, on success and failure alike. See What the job reports (on the z/OS tab).
Arrived Files
The file that arrived is available to later jobs
Once a File Arrival job matches — Windows or UNIX — the path it matched is recorded on the job instance. It shows on the job's Summary tab as Arrived File, and five properties carry it and its parts into the events and notifications that job fires, so a downstream job can be handed the actual file name rather than the pattern that found it.
The recorded path belongs to the attempt, not the job. A new attempt clears it when it starts, a job that fails before it is dispatched carries none, and a recurrence or a container's nested child starts each run clean — so an event can never be handed the previous run's file. A job parked between a failure and its retry keeps the failed attempt's file, because a File Arrival job can fail after matching and that attempt's events should still name what it found.
Windows
The Windows LSAM connector (opcon.lsam.windows) has these job types. Each has its own tab below.
| Job type | Agent type | Purpose |
|---|---|---|
Windows Command (windows-command) | Windows LSAM | Run a command on a Windows LSAM machine |
Windows - WS_FTP Pro (windows-wsftp-pro) | Windows LSAM | Run a WS_FTP Pro file transfer on a Windows LSAM machine |
Windows - Command: File Copy / Move / Rename / Delete (windows-file-copy / -move / -rename / -delete) | Windows LSAM | Copy, move, rename, or delete files on a Windows LSAM machine |
Windows - Corelation (windows-corelation) | Windows LSAM | Submit and monitor a Corelation batch job on a Windows LSAM machine |
Windows - Fiserv DNA (windows-fiserv-dna) | Windows LSAM | Run a Fiserv DNA SQT job on a Windows LSAM machine |
Windows - Fiserv DNA (File Loader) (windows-fiserv-dna-fileloader) | Windows LSAM | Load a file into Fiserv DNA from a Windows LSAM machine |
Windows - File Arrival (windows-file-arrival) | Windows LSAM | Wait for a file to arrive on a Windows LSAM machine during a time window |
Windows - Embedded Script (windows-embedded-script) | Windows LSAM | Run a script from the script library on a Windows LSAM machine |
Windows - Web Services (windows-web-services) | Windows LSAM | Run a multi-step HTTP/REST sequence through the SMA Web Services connector on a Windows LSAM machine |
- Command
- Embedded Script
- Web Services
- WS_FTP Pro
- File Operations
- Corelation
- Fiserv DNA
- File Arrival
Command
Windows LSAM — additional parameters
The Windows LSAM job is at functional parity with the legacy basic Windows job. In addition to the shared parameters (on the How It Works tab, under Shared Parameters):
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Connection | No | a saved connection | The Windows account the job runs as — a Windows Batch User connection. Blank runs the job under the LSAM's own service account. See Connection. |
| Working Dir | No | ≤255 chars | Working directory for the job. |
| Pre Run Working Dir | No | ≤255 chars | Working directory for the prerun command. |
| Run In Command Shell | No | boolean (default off) | Run the command inside a command shell. |
| Process Priority | No | Normal (default), High, Idle, Real-Time | The Windows priority the program and its prerun run at. Use Real-Time only on a machine dedicated to this job. A priority other than Normal needs a current relay: on an older one the job goes to Initialization Error with RELAY_LSAM_FEATURE_UNSUPPORTED rather than running at Normal. |
| Environment Variables | No | — | Environment variables (Name=Value) to set for the job. Supported on Windows as on UNIX. |
| Output Parsing | No | ≤5 rules | Scan the job output for text and set an exit code when it matches. Each rule is a Search Operation (Contains / Does Not Contain), a String To Find (1–255 chars), and an Exit Code To Set. |
| Custom Log File Path | No | ≤255 chars | Log file that output parsing scans. Defaults to the job's standard output. It is presented with the Output Parsing rules in the job editor rather than as a field of its own, because it only affects what those rules read. |
On Windows LSAM today:
- A job runs as the Windows Batch User connection attached to it, or as the LSAM's own service account when none is attached — see Connection (on the How It Works tab, under Connection).
- The Windows Command job has no embedded script — there is no embedded-script field on it; use Start Image and Arguments, a script already on the machine, or the separate Windows - Embedded Script job type.
- Windows returns a single integer exit code, so Signal Criteria and Core Dump Flag do not apply.
- Environment Variables are supported on Windows, the same as on UNIX.
Embedded Script
Windows LSAM — Embedded Script
The Windows - Embedded Script job type (windows-embedded-script) runs a script held in the
script library rather than one pasted into the job. On the job you choose the script, the version
to run, the runner that launches it, and any arguments:
| Field | Notes |
|---|---|
| Script | The library script to run. Only scripts whose language has a Windows runner — built-in, or one your tenant added — are listed. See below. |
| Version | Latest (track the newest version) or a specific pinned version number. |
| Runner | The program that starts the script on the agent. The choices depend on the script language, so pick the script first. |
| Arguments | Optional arguments passed to the script, up to 2000 characters. |
| Working Directory | Optional, up to 255 characters. Left blank, the script runs in the agent service's own directory — it does not fall back to the script's directory. |
| Exit Criteria Result / Exit Criteria | Up to 20 conditions tested against the script's exit code. See Exit criteria. |
| Environment Variables | Optional name/value pairs set for the script. |
Because the job holds a reference rather than a copy, fixing the script in the library reaches every job that tracks Latest; a job pinned to a version keeps running that version until you change it. A script cannot be deleted while a job references it. See Script library.
The Script list is filtered to what a Windows agent can actually run. A script whose language has no Windows runner is not offered, because choosing it would leave the required Runner list empty and the job could never be saved. The filter is fixed on the job type, so you do not have to pick an agent first, and it is worked out from the runners your tenant can see — so a Windows runner you add changes what this list offers.
Two consequences for scripts and jobs that predate this:
- A script you expect to see is missing from the list. Its language has no Windows runner — in the built-in catalog that means Shell and Bash. Library scripts carried over before script types were introduced default to Shell, so this can affect a lot of them at once. Either change the script's type in the library, or add a Windows runner for that type, which makes every script of that type available here.
- An existing job already points at such a script. The job still shows the script by name rather than a bare identifier, and the Runner field explains the dead end instead of reading as an empty list. Point the job at a Windows-runnable script, or add the missing runner, to make it saveable.
The script and the runner are checked when the workflow is saved, so a pair that does not go together is named on the save rather than failing on the job's next run — see Script and runner are checked when the workflow is saved. If the script or the runner a job holds has since been deleted, the field says This script is no longer available. Choose another. (or This runner is no longer available. Choose another.) instead of showing an identifier with no explanation.
Web Services
Windows LSAM — Web Services
The Windows - Web Services job type (windows-web-services) runs a sequence of HTTP requests
through the SMA Web Services connector installed on the Windows LSAM machine. A value extracted
from one response is available to every later step, so one job can authenticate, submit work, poll
for completion, and download a result.
Like WS_FTP Pro, this type has no Start Image or Parameters field — the relay assembles the connector command line from the form fields.
Core parameters
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target Windows LSAM machine; must match the registered agent name. |
| Connection | No | a saved connection | The Windows account the job runs as. Blank runs it as the agent's service account. See Connection. |
| Connector Location | Yes | 1–252 chars, no " | Folder containing SMAWSConnector.exe. It is also the job's working directory, which is where the connector looks for config\Connector.config — so it is load-bearing, not cosmetic. Note the cap is 252, one lower than the other Windows types, because this type appends a trailing backslash. |
| Template ID | No | ≤128 chars | A free-text label for your own naming convention. The connector never reads it; it appears in the connector's log. |
| Steps | Yes | 1–50 steps | The requests to run, in order. See below. |
| Variables | No | ≤99 entries | Template-wide name/value pairs substituted into any URL, header, or body. Names conventionally start with @ (for example @Url). |
| Environment Variables | No | — | Environment variables set for the connector process. Any name starting with @ is imported into the connector's variable map and overrides a template variable of the same name. |
| Exit Criteria | No | ≤20 conditions | Conditions tested against the job's exit code — but read the caution below first, because this job type does not exit 0 on success. See Exit criteria. |
| Exit Criteria Result | No | Fail / Finish OK | What a match decides. Default Fail. |
| Property Update on Completion | — | — | Not supported. A job with any Property Update on Completion entry is rejected before dispatch. |
The connector exits with the raw HTTP status of the last step, so a successful call exits 200,
not 0. Leave Exit Criteria empty and the default — fail on any non-zero exit code — fails every
job that worked.
Either of two single-row tables fixes it, and they mean the same thing:
| Exit Criteria Result | Row | Reads as |
|---|---|---|
Fail (the default) | Not Equal To 200 | Fail when the code is not 200 |
Finish OK | Equal To 200 | Finish OK when the code is 200, fail otherwise |
The first is enforced by the agent; the second is enforced by OpCon Continuum after the agent
reports, because Finish OK is not something the legacy wire can carry. See
Exit criteria (on the How It Works tab, under Exit Criteria) for what that difference means on the job.
Credentials
Keep credentials out of the job definition by supplying them as environment variables rather than
template variables. A name starting with @ overrides the template variable of the same name, which
is the supported way to provide @User, @Password, and @Domain. An environment-variable value may
be an <SmaEncrypt> token; the agent decrypts it and masks it in the job log.
The relay's own LSAM Connector log redacts a legacy job's command line, environment values and script body, so a secret passed that way is not readable there — but that was not true of earlier relays. See the connector log before relying on a log written by one.
Authentication mode is chosen from the prefix of a request header's value — Basic, NTLM,
CERT, or SOAP each trigger special handling using @User / @Password / @Domain. Any other
header value is sent verbatim after variable substitution.
Step parameters
Each of the 1–50 steps carries a request and a response definition.
| Parameter | Required | Values / limits | Notes |
|---|---|---|---|
| Method | No | GET (default), POST, PUT, PATCH, DELETE | The HTTP method. Case-sensitive. |
| URL | Yes | 1–2000 chars | May contain @variables. Required — a step with a blank URL is rejected rather than silently discarded. |
| Proxy Server | No | — | Proxy for this step; overrides the connector's configuration file. |
| TLS Version | No | TLS (default), TLSv1.0–TLSv1.3 | The connector as shipped does not enable TLSv1.3, even though it is offered here. |
| Request › Content Type | No | see below | Defaults to application/json. A value outside the list below is refused at dispatch. |
| Request › Body | No | — | Request body; may contain @variables. |
| Request › Body From File | No | — | Read the body from this file on the agent instead. Takes precedence over Body, except for multipart/form-data and application/octet-stream POSTs. |
| Request › File Line Separator | No | CRLF (default) / LF | Line separator used when reading Body From File. |
| Request › Headers | No | ≤50 entries | Name/value pairs. The value's prefix selects the authentication mode — see above. |
| Response › Content Type | No | see below | Expected response content type, used to parse extracted values. |
| Response › Expected Status | Yes | 100–599, default 200 | The comparison is exact equality, not a 2xx class — a 201 where 200 was expected is a failure. |
| Response › Extract Variables | No | ≤50 entries | Values to pull out of this response for later steps. Extract From is a JsonPath (for example $.id) or an XPath; prefix with # to read a response header; or use TEXTSTRING for the whole body. |
| Response › Save Response To | No | — | Write the response body to this file on the agent. An existing file is overwritten. |
| Response › Ignore Result | No | boolean (default off) | Continue to the next step even when this step's status does not match. |
Both content-type lists offer application/json, application/json-patch+json, application/xml,
application/x-www-form-urlencoded, application/octet-stream, multipart/form-data, text/xml,
and text/plain.
Response data check and polling
A step can optionally inspect a value in its response (its Response Data Check), and poll until that value appears. Attribute To Check is the switch — leave it blank and the whole check, including polling, is skipped.
| Parameter | Values | Notes |
|---|---|---|
| Attribute To Check | JsonPath or XPath | The value to inspect. Blank disables the entire check. |
| Good Finish | slash-separated list | Values that mean success, compared case-insensitively (for example COMPLETED/SUCCESS). |
| Bad Finish | slash-separated list | Values that mean failure. |
| Poll | boolean (default off) | Repeat the request until Good Finish or Bad Finish matches. GET steps only — polling on any other method is rejected at dispatch. |
| Poll Delay (seconds) | seconds, default 3 | Wait before the first poll. |
| Poll Interval (seconds) | seconds, default 2 | Wait between polls. |
| Poll Maximum (minutes) | minutes | Give up after this long. Note the unit differs from the two fields above. Leave it blank and the connector polls indefinitely. |
Characters the relay rejects, and why
The connector's job document is single-quoted, and the Windows agent rewrites some sequences on the command line. Rather than let either corrupt a request silently, the relay refuses the job at dispatch. The job saves cleanly and fails when it runs, like every legacy LSAM type.
| Rejected | Where | Why |
|---|---|---|
An apostrophe (') | Any step value, template variable or Template ID — including the URL. Not checked in Connector Location or in environment variables. | It terminates a string in the connector's single-quoted document and corrupts the whole document. The one exception is a body whose content type is application/json or application/json-patch+json: those are the only content types whose quote transform the connector reverses. |
The sequence \* | Any field except environment variables | The Windows agent replaces it with the machine's hostname anywhere on the command line — including deep inside a JSON body — so the corruption would be silent. |
A double quote (") | Connector Location | The working directory is quote-wrapped for transport. |
Quotes in a body are also transformed for transport: for content type exactly
application/xml, every double quote becomes the five characters u0022; for every other content
type, every single quote becomes u0027. A body that legitimately contains that literal text is
therefore altered. Use Body From File when you need the bytes to arrive untouched — it bypasses the
transform entirely.
WS_FTP Pro
Windows LSAM — WS_FTP Pro file transfer
A second Windows LSAM job type, Windows - WS_FTP Pro (windows-wsftp-pro), is registered
alongside windows-command on the same opcon.lsam.windows plugin (Windows LSAM agents only). It
runs a WS_FTP Pro file transfer on the machine.
WS_FTP Pro has no dedicated LSAM field codes. The transfer runs as an
ordinary Windows Run Program job: the relay assembles the form fields into a wsftppro.exe
command line plus a working directory. There is no Start Image or Parameters field
on this job type — the command line is built entirely from the fields below.
Authentication is delegated to WS_FTP Pro site profiles configured on the agent. The job references a source and a destination profile by name only and carries no credentials; the named profiles must already exist in the WS_FTP Pro installation on that machine.
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target Windows LSAM machine; must match the registered agent name. |
| Connection | No | a saved connection | The Windows account the job runs as. Blank runs it as the agent's service account. See Connection. |
| Ws FTPPro Location | Yes | 1–253 chars, no " | Folder containing wsftppro.exe. Capped at 253 so the quote-wrapped working directory stays within its 255 cap. |
| Source Profile | Yes | no " or : | WS_FTP Pro site profile name for the source. |
| Source File | Yes | no " | Source file to transfer. |
| Destination Profile | Yes | no " or : | WS_FTP Pro site profile name for the destination. |
| Destination File | Yes | no " | Destination file to write. |
| File Transfer Options | No | free text | Extra wsftppro.exe command-line switches (e.g. -binary -delete); appended verbatim, only when non-empty. |
| Exit Criteria Result | No | Fail (default) / Finish OK | What a match in Exit Criteria means. See Exit criteria. |
| Exit Criteria | No | ≤20 conditions | Conditions tested against the transfer's exit code, always evaluated by Continuum. Leave it empty to fail on any non-zero exit code. See Exit criteria. |
Plugin job-schema patterns and lengths are
UI render hints only — a job is not checked against them when you save it, so a job with a
bad value saves cleanly and then fails when it runs. The relay is the enforcement point and rejects:
a missing required field (RC_MISSING_WSFTP_FIELD); a double quote in any of the five required text
fields, or a colon in either profile field (RC_INVALID_WSFTP_FIELD — : delimits profile from file
in the assembled command line); a command line over 4000 chars (RC_CMDLINE_TOO_LONG); or a working
directory over 255 chars (RC_WORKDIR_TOO_LONG). A trailing backslash in Source File or Destination
File (e.g. C:\archive\) is preserved losslessly — the relay doubles it so it can't escape the closing
quote and swallow later options.
File Operations
Windows LSAM — file operations
Four more Windows LSAM job types cover routine file housekeeping, registered alongside
windows-command and windows-wsftp-pro on the same opcon.lsam.windows plugin (Windows LSAM
agents only):
| Job type | Name | Underlying command | Purpose |
|---|---|---|---|
windows-file-copy | Windows - Command: File Copy | xcopy | Copy one or more files (wildcards allowed). |
windows-file-move | Windows - Command: File Move | move | Move one or more files, overwriting the destination without prompting. |
windows-file-rename | Windows - Command: File Rename | rename | Rename a single file (cannot overwrite an existing name). |
windows-file-delete | Windows - Command: File Delete | del | Delete one or more files or directories (wildcards allowed). |
As with WS_FTP Pro there is no Start Image or Parameters field on these job types — the relay assembles the whole command line from the fields below and runs it through the Windows command shell. Each of them runs under whichever Windows Batch User connection is attached, or under the LSAM's own service account when the Connection field is left blank.
All four also need a machine, set under Agent Assignment (required, 1–24 chars), whose name must match the registered Windows LSAM agent name.
File Copy — additional parameters
| Parameter | Required | Notes |
|---|---|---|
| Source | Yes | Directory and file(s) to copy. Wildcards copy multiple files. |
| Destination | Yes | Destination directory and/or file name. |
| Destination Type | No | File (default) or Directory — answers the file-vs-directory question xcopy would otherwise prompt for. |
| Verify Destination Files | No | Verify each copied file (default on). |
| Use Short Name Format | No | Use the short (8.3) name when copying a non-8.3 file (default off). |
| Copy Subdirectories | No | Copy subdirectories, skipping empty ones (default off). |
| Other Options | No | Extra xcopy switches. Run xcopy /? on the machine to list them. The switches that would make xcopy stop and wait for an answer — /P, /W and /-Y — are removed before the job runs, because nobody is there to answer. |
File Move — additional parameters
| Parameter | Required | Notes |
|---|---|---|
| Source | Yes | Directory and file(s) to move. |
| Destination | Yes | Destination directory and/or file name. Existing files are overwritten without prompting. |
File Rename — additional parameters
| Parameter | Required | Notes |
|---|---|---|
| Path and Current File Name | Yes | Drive, full path, and file to rename. |
| New File Name | Yes | New name for the file. Rename cannot overwrite an existing name. |
File Delete — additional parameters
| Parameter | Required | Notes |
|---|---|---|
| File To Delete | Yes | Comma-separated list of one or more files or directories; wildcards allowed. Naming a directory deletes every file in it. |
| Force Delete Read Only Files | No | Delete read-only files too (default off). |
| Delete Specified Files from all Subdirectories | No | Delete the named files from all subdirectories (default off). |
| Attribute filters | No | Read Only, Not Content Indexed, Archive, Hidden, System, and Reparse Point — set each to Include (delete only files carrying that attribute) or Exclude (skip them). Leave a filter unset to ignore that attribute. |
| Other Options | No | Extra del switches. Run del /? on the machine to list them. /P, which would make del stop and ask before each file, is removed before the job runs. |
- These are the first relay-assembled command lines that run through a shell, so an operand
containing a space or a shell metacharacter (
&,|,<,>,^) is quoted automatically. A trailing backslash is doubled first so it can't escape the closing quote. - File Delete quotes each entry separately, because File To Delete is a list — quoting the whole value would collapse it into one file name that doesn't exist.
Corelation
Windows LSAM — Corelation batch job
A seventh Windows LSAM job type, Windows - Corelation
(windows-corelation), registered alongside the other Windows types on the same
opcon.lsam.windows plugin (Windows LSAM agents only). It submits and monitors a Corelation
batch job through SMA's Corelation connector (SMARunCorelationJob.exe) on the machine.
Corelation has no dedicated LSAM field codes. As with WS_FTP Pro and the file-operation types,
the relay assembles the form fields into a SMARunCorelationJob.exe command line plus a
working directory — there is no Start Image or Parameters field on this job type. The
SMARunCorelationJob.exe connector, and a configuration file holding the Corelation connection
details, must already be present in the connector folder on that machine. The job runs under
whichever Windows Batch User connection is attached, or under the LSAM's own service account
when the Connection field is left blank.
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target Windows LSAM machine; must match the registered agent name. |
| Connection | No | a saved connection | The Windows account the job runs as. Blank runs it as the agent's service account. See Connection. |
| Connector Location | Yes | 1–253 chars, no " | Folder containing SMARunCorelationJob.exe. The job runs in this folder. |
| Type | No | Job Name (default) / Job Serial | Whether Corelation Job below is a job name or a job serial. |
| Corelation Job | Yes | no " | The Corelation job name or serial to submit. |
| Configuration File | Yes | ≤255 chars, no " | Connector configuration file with the Corelation connection details; a relative path resolves against the connector folder. A new job starts with .\SMARunCorelationJob.ini stored; clear it and saving is refused with Required parameter 'configurationFile' is missing. |
| Batch Server | No | no " | Corelation batch server name. Leave blank to use the connector's default (Batch Server). |
| Use Least Busy Batch Queue | No | boolean (default on) | Send the job to the open queue with the least work. A new job starts with it on. |
| Batch Queue | While Use Least Busy Batch Queue is off | no " | The Corelation batch queue to submit the job to. Shown, and required, only while Use Least Busy Batch Queue is off; turning the switch on clears it. |
| Format As | No | Batch Options (default) / Parameters | How the runtime parameters below are passed to the connector. |
| Property Owner | No | no " | Parent property that wraps the batch options. Greyed out unless Format As is Batch Options. |
| Parameters | No | ≤99 entries | Runtime parameters as Name/Value pairs. Neither name nor value may contain a double quote (") or a pipe (|). |
| Include XMLIn Output | No | boolean (default on) | Log every Corelation request and response XML to the job output. |
| Show View Submit Response | No | boolean (default off) | Log the VIEW_SUBMIT response. |
| Include Job Details In Output | No | boolean (default off) | Retrieve job details only — the Corelation job is not run. |
| Show Job List | No | boolean (default off) | List the available Corelation jobs in the job output. |
| Include Debug Messages | No | boolean (default off) | Verbose connector debug output (also logs the XML). |
The same rule as WS_FTP Pro and the file
operations. A bad value saves cleanly and then fails when the job runs. The relay rejects a
missing required field (RC_MISSING_CORELATION_FIELD); a double quote in any text field, a pipe
in a parameter name/value, an unrecognized Type or Format As value, or a blank parameter name
(RC_INVALID_CORELATION_FIELD); more than 99 parameters (RC_TOO_MANY_CORELATION_PARAMETERS); a
command line over 4000 characters (RC_CMDLINE_TOO_LONG); or a working directory over 255
characters (RC_WORKDIR_TOO_LONG). A trailing backslash in a value (for example C:\Reports\) is
preserved losslessly — the relay doubles it so it can't escape the closing quote and swallow the
switches that follow.
A job saved before the switch existed opens with it set from its queue. A blank queue, or
leastbusy in any case, opens with Use Least Busy Batch Queue on; any other queue — a
property token included — opens with the switch off and the queue kept. Opening the
job changes nothing until you save it, and a job that is never opened in the editor runs exactly as
before.
Fiserv DNA
Windows LSAM — Fiserv DNA jobs
Two more Windows LSAM job types cover for running Fiserv DNA jobs,
registered alongside the other Windows types on the same opcon.lsam.windows plugin (Windows LSAM
agents only):
| Job type | Name | Purpose |
|---|---|---|
windows-fiserv-dna | Windows - Fiserv DNA | Run a DNA SQT job — a DNA application (APPL), optionally scoped by cycle codes. |
windows-fiserv-dna-fileloader | Windows - Fiserv DNA (File Loader) | Load a file into DNA — the PS_FILELOADER application with file batch/record counts and control totals. |
Both run a job through SMA's DNA connector (SMARunDNAJob.exe) on the machine. Like WS_FTP Pro,
the file-operation types, and Corelation, DNA has no dedicated LSAM field codes: the relay
assembles the form fields into a SMARunDNAJob.exe command line plus a working directory,
so there is no Start Image or Parameters field on either job type. The SMARunDNAJob.exe
connector, and a configuration file, must already be present in the connector folder on that
machine. Both types run under whichever Windows Batch User connection is attached, or under the
LSAM's own service account when the Connection field is left blank.
Credentials live in the configuration file, not the job. The DNA/SQT (Oracle) user name and
password are read — and decrypted — by the connector from the [SQRT Parameters] section of the
configuration file. They are deliberately not job parameters, so a DNA job carries no
credentials of its own.
Common parameters (both types)
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target Windows LSAM machine; must match the registered agent name. |
| Connection | No | a saved connection | The Windows account the job runs as. Blank runs it as the agent's service account. See Connection. |
| Connector Location | Yes | 1–253 chars, no " | Folder containing SMARunDNAJob.exe; also the job's working directory. Capped at 253 so the quote-wrapped working directory stays within its 255 cap. |
| Config File | Yes | ≤255 chars, no " | Connector configuration file (holds the [SQRT Parameters] credentials); a relative path resolves against the connector folder. A new job starts with .\SMARunDNAJob.ini stored; clear it and saving is refused with Required parameter 'configFile' is missing. |
| Application Name | Yes | no " | DNA application (APPL) name; also names the connector's log file, so avoid characters that are invalid in a file name. A new File Loader job starts with PS_FILELOADER (the fixed DNA application for file loads) stored; on the SQT job there is no default, so enter one, or saving is refused with Required parameter 'applicationName' is missing. |
| Application Number | No | no " | DNA application (APPL) number. The connector prefers it over the name for the database lookup. |
| Effective Date | No | no " | Run date of the request. The connector requires exactly 10 characters and otherwise doesn't validate the format; leave blank to let it derive the date from the DNA BANKOPTION table. |
| Effective Date Offset | No | -365 to 365 | Day offset applied to the effective date. |
| Parameters | No | ≤99 entries | Runtime parameters as Name/Value pairs (see Runtime parameters below). Neither name nor value may contain a double quote (") or a pipe (|). |
Fiserv DNA (SQT job) — additional parameters
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Cycle Code Not Required | No | boolean (default off) | Skip the connector's check that the application has cycle codes. |
| Cycle Codes | No | ≤99 entries | Cycle codes to run (for example EOM, EOQ). No entry may contain a double quote (") or a pipe (|); a blank entry is rejected. |
Fiserv DNA (File Loader) — additional parameters
The File Loader always runs the PS_FILELOADER application, so it takes the file's identity and
control totals instead of cycle codes. Each value may be a literal or the name of an OpCon
property to resolve at run time.
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| File Name | Yes | no " | Name of the file to load, matched against DNA's EXTFILE table. |
| File Batch Count | Yes | no " | Batch count. |
| File Record Count | Yes | no " | Record count. |
| Credits | Yes | no " | Total credit amount. |
| Debits | Yes | no " | Total debit amount. |
| File Number | Yes | no " | File number. |
Runtime parameters
Parameters is a list of Name/Value pairs (up to 99). Each pair may optionally carry date handling:
- Action — the date type: Business date (
B) or Calendar date (O); required before an offset or a date format can be set. - Offset — a day offset (-365 to 365); requires a date type.
- Date Format — honored only with Calendar date (
O) and an offset set; it is ignored on the Business date (B) path, so setting it there is rejected rather than silently dropped.
Neither the name, the value, nor the date format may contain a double quote (") or a pipe (\|) —
the pipe delimits the parts of a parameter internally. A blank value is allowed (the connector maps
it to a database NULL); a blank name is not.
The same rule as WS_FTP Pro, the file
operations, and Corelation. A bad value saves cleanly and then fails when the job runs. The
relay rejects a missing required field (RC_MISSING_DNA_FIELD); a double quote in any text field,
a pipe in a cycle code or a parameter name/value/date-format, a blank parameter name or cycle
code, an unrecognized date type, an offset or date format without the date type it requires, or a
value of the wrong type (RC_INVALID_DNA_FIELD); more than 99 parameters
(RC_TOO_MANY_DNA_PARAMETERS); more than 99 cycle codes (RC_TOO_MANY_DNA_CYCLE_CODES); a command
line over 4000 characters (RC_CMDLINE_TOO_LONG); or a working directory over 255 characters
(RC_WORKDIR_TOO_LONG). Every value is quoted for you, and a trailing backslash in a value (for
example C:\DNA\) is preserved losslessly — the relay doubles it so it can't escape the closing
quote and swallow the switches that follow.
File Arrival
Windows LSAM — File Arrival
The Windows - File Arrival job type (windows-file-arrival) is registered alongside the other
Windows types on the same opcon.lsam.windows plugin (Windows LSAM agents only). It waits for a
file to appear on the machine during a time window and succeeds once the file has arrived and
stopped changing, so a downstream job can depend on it and only run after the file is really there.
Unlike the other assembled-command-line Windows types (WS_FTP Pro, the file operations, Corelation, and Fiserv DNA), File Arrival needs no connector installed and no command line: the LSAM agent's own built-in file watcher does the watching. You only describe what to watch for and when to watch. The job runs under whichever Windows Batch User connection is attached, or under the LSAM's own service account when the Connection field is left blank.
Parameters
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target Windows LSAM machine; must match the registered agent name. |
| Connection | No | a saved connection | The Windows account the job runs as. Blank runs it as the agent's service account. See Connection. |
| File Name | Yes | 1–4000 chars | Full path and file name to watch for. Wildcards * and ? are allowed in the file name part; the directory part must already exist, or the job fails. UNC paths and %ENVVAR% references are allowed. The path may be wrapped in double quotes, in whole or in part — see Quotes in the file name. |
| Start Offset | No | -142560 to 143999 | When to start watching, relative to midnight of the job's schedule date. Authored as a Day Offset plus a Time — see Authoring the watch window. Defaults to 0. |
| End Offset | No | -142560 to 143999 | When to stop watching, relative to midnight of the job's schedule date. Must not be earlier than Start Offset. To watch across midnight, use a later day rather than an earlier time. Defaults to 0. |
| File Size Stable Time | No | 1 to 999 | Seconds to wait before re-checking the file, to confirm it has been completely written. The file counts as complete once its size, creation time, and last-write time are all unchanged across the interval. Defaults to 5. |
| Include Subdirectories | No | boolean (default off) | Also watch subdirectories of the given path. |
| Exit Criteria Result | No | Fail / Finish OK | What a match decides for the criteria table. Default Fail. |
| Exit Criteria | No | ≤20 conditions | Conditions tested against the watcher's exit code, always evaluated by Continuum. Leave empty to fail on any code other than 0. See Exit criteria. |
Leave both offsets at zero to check whether the file already exists right now and finish immediately, rather than waiting for a window.
A File Arrival job stays running for the whole window, which can be many hours, and finishes as soon as the file arrives and settles. Size the window with the start and end offsets to match when you actually expect the file.
Quotes in the file name
A double quote can never be part of a Windows file name, so a quote in File Name is read as quoting around the path, the way the Windows command line reads it. File Arrival jobs migrated into Continuum often store their name this way, wrapped in quotes because the path has a space in it:
| File name as stored | What is watched for |
|---|---|
"C:\Core Exports\ACH*.txt" | C:\Core Exports\ACH*.txt |
"C:\Core Exports"\ACH*.txt | C:\Core Exports\ACH*.txt |
C:\Core Exports\ACH*.txt (no quotes) | C:\Core Exports\ACH*.txt, unchanged |
You never need to add the quotes yourself: a path with a space and no quotes is watched as typed. Once the value has a quote in it, though, it must still read as one path:
- the quotes must be balanced — an opening quote with no closing one is refused;
- a space outside the quotes ends the path, so anything after it but spaces is refused — quote the part with the space, or the whole path;
- a quote cannot stand for a literal quote character — a doubled
""inside quotes is refused, and so is a quote straight after a single backslash (\"), which escapes it, so never put a quote directly after a backslash; - a value that is only quotes, such as
"", counts as a missing file name.
A refused value fails the job at dispatch, before anything reaches the machine — see Outcome and exit codes.
This needs a current relay. A relay older than this release refuses any double quote in the file name, with File Arrival fileName must not contain a double quote (RC_INVALID_FILEARRIVAL_FIELD), so on such a relay a quoted name still fails at dispatch.
Authoring the watch window
Both offsets are authored as a pair of controls — a Day Offset box and a Time picker — rather than as a raw minute count:
| Control | What it takes |
|---|---|
| Day Offset | Whole days from the job's schedule date, -99 to 99. 0 is the schedule date itself, -1 the day before, 1 the day after. The sign lives here: this is the only one of the two controls that accepts a negative value. |
| Time | A 24-hour clock time, 00:00 to 23:59, on that day. |
So a window of 08:00 on the schedule date to 00:45 the following morning is Day Offset 0 / Time
08:00 to Day Offset 1 / Time 00:45. That is the same stored value the platform has always
held — a single count of minutes from midnight of the schedule date — presented as the day and the
clock time it works out to. Nothing about an existing job's saved window changes, and neither does
what reaches the agent.
Day Offset is days, not minutes. The pair replaces a single box that took the whole offset in
minutes, so 480 meant 08:00. Typing 480 into Day Offset does not give you 08:00: the box
takes each digit as you type it, so it keeps 48 — a window 48 days out, which is still inside the
stored range, so nothing downstream objects. 480 itself is outside -99 to 99, so the box
refuses it and, when you leave it, shows the day it kept, so a rejected entry never looks accepted.
Check the day the box shows before you save.
An offset the pair cannot express — a value outside the range, or a property reference standing in for the number — falls back to the plain numeric box, still in minutes, and is left exactly as stored.
Outcome and exit codes
The job succeeds when a matching file arrives and stops changing within the window, and fails otherwise. The specific reason is preserved as the job's exit code, visible on the job's Completion and Output detail:
| Exit code | Meaning |
|---|---|
0 | A file matched. The job succeeds. |
1 | The window elapsed with no match. |
2 | The monitored directory doesn't exist. |
3 | A candidate file's creation time predates the start of the window. |
4 | A network path was still unreachable at the end of the window. |
-1 | An argument or range error. |
With no criteria table, only 0 is a success — every non-zero code is reported as Failed — and
the number stays on the job, so you can tell why it failed after the fact. Exit criteria change
that, as in Classic: a table that finishes 1 OK, for example, lets a job whose file did not come
today finish OK rather than fail. Continuum always evaluates the table, because the Windows agent
judges every file-watcher job by the default whatever it is sent — see
Where the table is evaluated.
The same rule as the other Windows LSAM job types: a bad value saves cleanly and then fails when
the job runs. The relay rejects a missing file name, including one that is only quotes
(RC_MISSING_FILEARRIVAL_FIELD); and a file name whose quotes are unbalanced, leave more than one
path, or stand for a literal quote (File Arrival fileName has unbalanced or embedded double quotes;
quotes may only wrap the path or a part of it), a file name over 4000 characters, an offset or stable time that isn't a whole
number in range, an End Offset earlier than Start Offset, or a non-boolean Include
Subdirectories (RC_INVALID_FILEARRIVAL_FIELD). A trailing backslash in the file name is preserved losslessly — the
relay doubles it so it can't escape the watcher's closing quote.
UNIX/Linux
The UNIX/Linux LSAM connector (opcon.lsam.unix) has these job types. Each has its own tab below.
| Job type | Agent type | Purpose |
|---|---|---|
UNIX Command (unix-command) | UNIX LSAM | Run a command or script on a UNIX/Linux LSAM machine |
UNIX - File Arrival (unix-file-arrival) | UNIX LSAM | Wait for a file to arrive on a UNIX/Linux LSAM machine during a time window |
UNIX - Embedded Script (unix-embedded-script) | UNIX LSAM | Run a script from the script library on a UNIX/Linux LSAM machine |
UNIX - Episys: … (seven unix-episys-* job types) | UNIX LSAM | Drive Symitar/Episys batch processing through RSJ on a UNIX/Linux LSAM machine — see Episys (RSJ) job types (on the UNIX/Linux tab, under Episys) |
- Command
- File Arrival
- Embedded Script
- Episys
Command
UNIX/Linux LSAM — additional parameters
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Connection | Yes | a saved connection | The UNIX user and group the job runs as — a UNIX Batch User connection. A UNIX job must name one; without it the dispatch fails (RC_NO_UID). See Connection. |
| Nice Value | No | -20 to 19 | Process priority (-20 highest, 19 lowest). |
| Signal Criteria | No | ≤5 conditions | Signal conditions that mark the job failed (same operator/value form as Exit Criteria). |
| Core Dump Flag | No | Y / N | Whether a core dump fails the job. |
| Environment Variables | No | — | Environment variables to set for the job. |
| Output Parsing | No | ≤5 rules | Scan the job output for text and set an exit code when it matches. Each rule is a Search Operation (Contains / Does Not Contain), a String To Find (1–255 chars), and an Exit Code To Set (−999999 to 9999999). See Output parsing on a UNIX job. |
| Custom Log File Path | No | ≤255 chars | A log file to append to the job output before the rules search it. Wildcards allowed. Presented with the Output Parsing rules in the job editor, because it only affects what those rules read. |
The command job used to carry a job sub-type field whose F value meant file watcher. It has been
removed. Waiting for a file on a UNIX agent is the separate
UNIX - File Arrival (on the UNIX/Linux tab, under File Arrival) job type, which has its own fields and needs no
command line.
A UNIX Command job whose script reports its steps to the agent can be restarted on one of those steps instead of from the beginning — see Restart on a step. The other UNIX job types always restart from the beginning.
Output parsing on a UNIX job
Output parsing works on a UNIX job. A job that exits 0 but writes a failure marker to its
output can be failed on that text, the same as on Windows. The rules and the Custom Log File
Path were not carried to the agent at all in earlier builds — the fields did not exist on the UNIX
job, and nothing was sent even if a value reached the job another way — so a job whose only signal of
failure was in its log finished OK. If you have been failing such a job with a wrapper script or
an exit-criteria workaround, you can now express it directly.
How it decides the outcome:
- Up to five rules, searched in order; the first match wins and its Exit Code To Set replaces
the job's own exit code. An exit code of
0reads as no match, so it leaves the job's code alone. - The search is case-insensitive.
- The replacement happens before exit criteria (on the How It Works tab, under Exit Criteria) are applied, so the criteria judge the code a rule set, not the one the command returned.
- A Custom Log File Path is appended to the job's output first, so the rules search both. Leave it empty and they search the job's own output.
The agent runs the search on the machine through a shell, as the account the LSAM daemon runs
under — typically root. A search string is therefore checked before the job is dispatched, and the
job fails without running if it would not be safe to pass on (RC_INVALID_OUTPUT_PARSING). The
error names the rule number and the restriction, and never echoes the value.
A String To Find must not:
- contain
`,$,",\,&, or a line break; - start with
-(a shell would read it as an option) or*.
More than five rules is refused rather than truncated, and Search Operation must be exactly
Contains or Does Not Contain — anything else is rejected rather than being quietly read as
Contains. A Custom Log File Path is checked too, and more strictly: it must not
contain whitespace, a control character, or any of ` $ " \ ; | & < > ( ) '.
These checks are stricter than the legacy product's, and deliberately so: the text reaches a root shell. If a rule you used in the legacy product is refused here, match on a different part of the message — the characters above are rarely the distinguishing part of a log line.
The rules must be contiguous. They are numbered as you order them, and the agent stops reading at the first gap, so a disabled or empty rule in the middle of the list would silently switch off every rule after it. The job editor keeps them contiguous for you; if you build a job through the API, do not leave a hole in the list.
UNIX values reach the agent exactly as you typed them
A UNIX job's values — Start Image, Parameters, Environment Variables — are sent to the agent
unaltered. Quotes, ampersands and angle brackets arrive as the characters you wrote, so
/bin/sh -c "exit 3" runs the command you meant.
This is a correction. The relay used to convert those five characters to XML entities on every
platform, and a UNIX agent does not convert them back — so the shell received the entity text
itself. A command containing a double quote exited 127 with the agent reporting
quot: not found, and the same went for &, <, > and '. Windows and SQL jobs are
unaffected: those agents decode the entities, and their values are still escaped for transport by
default. A Windows, SQL or MCP agent's Requires XML Escape Sequences setting chooses this per
agent — see Check CRC and Requires XML Escape Sequences.
IBM i values are now sent unaltered too — see
IBM i values reach the agent as you typed them (on the IBM i tab, under Dispatch Rules).
Sending a value untouched means two character sequences can no longer be carried, because the agent would read them as part of the message rather than as your text. Both are refused at dispatch, naming the field code, and the job fails without running:
| Rejected in a UNIX value | Why |
|---|---|
</F> | The agent finds where a value ends by looking for the first </F>, so the value would be silently truncated there — you would get a shortened command with no indication anything was lost (RC_FIELD_VALUE_CONTAINS_FIELD_TERMINATOR). |
<F I= | It looks like the start of another field. The agent takes the first match for each field it wants, so text like this in an earlier value could stand in for a later field — overriding, for instance, the account the job runs as (RC_FIELD_VALUE_CONTAINS_FIELD_TAG). |
Both matches are case-sensitive, as the agent's own search is. Neither sequence has a use in a shell command; if you need the characters for something else, put them in a script on the machine and run that.
File Arrival
UNIX/Linux LSAM — File Arrival
The UNIX - File Arrival job type (unix-file-arrival) is registered alongside UNIX
Command on the same opcon.lsam.unix plugin (UNIX/Linux LSAM agents only). Like its Windows
counterpart it waits for a file to arrive during a time window and succeeds once a matching
file has arrived and stopped growing, so a downstream job can depend on it and only run once the
file is really there.
It needs no command line: the LSAM agent watches for the file itself. You describe what to watch for and when to watch.
Parameters
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target UNIX/Linux LSAM machine; must match the registered agent name. |
| Connection | Yes | a saved connection | The UNIX Batch User connection the job runs as. As with every UNIX job type it is required — there is no "run as the agent's own account" option the way there is on Windows — and the account it names has to be able to read the arrived file. See Connection. |
| File Name | Yes | 1–255 chars | Absolute path of the file to watch for. See What the path may contain below. |
| Start Offset | No | -142560 to 143999 | When to start watching, relative to midnight of the job's schedule date. Authored as a Day Offset plus a Time, the same pair as the Windows type — see Authoring the watch window (on the Windows tab, under File Arrival). Defaults to 0. |
| End Offset | No | -142560 to 143999 | When to stop watching, relative to midnight of the job's schedule date. Must not be earlier than Start Offset. 0 means that midnight, not now. To watch past midnight, use a later day rather than an earlier time. Defaults to 0. |
| File Size Stable Time | No | 1 to 999 | Seconds to wait before re-checking the file's size, to confirm it has been written completely. Defaults to 5. |
| Include Subdirectories | No | boolean (default off) | Also search every subdirectory of the given directory. |
Leave both offsets at zero to ask whether a matching file exists right now and finish immediately, rather than waiting for a window.
A file counts only if it was last modified inside the window. A file that keeps growing keeps the job waiting — even past the end of the window — until its size holds steady across File Size Stable Time.
The Windows type treats a file as complete once its size, creation time and last-write time are all unchanged across the interval. The UNIX agent compares size only. A file whose timestamps move but whose length does not will satisfy a UNIX watch and not a Windows one.
What the path may contain
The file-name part may use the wildcards *, ? and [...]. Matching is case-sensitive,
and files whose names begin with a dot are ignored.
The directory part is taken literally — no wildcards — must already exist, and must not
contain spaces. A directory that doesn't exist is the job's exit code 2 rather than a wait that
never matches.
These characters are refused anywhere in the path, as are control characters:
` $ " \ ; | & < > ( ) '
Classic accepts those characters; Continuum does not, on any agent. Older UNIX agents match files by handing the name to a shell, so a crafted file name could run commands as the account the agent runs under. Newer agents never reach a shell, but an agent does not reliably report its version, so the restriction applies to every agent rather than only the ones that need it. If you need one of these characters in a watched path, you cannot use this job type for it.
Outcome and exit codes
A matching file makes the job succeed; anything else makes it fail. The reason survives as the job's exit code:
| Exit code | Meaning |
|---|---|
0 | A file matched. The job succeeds. |
1 | The window elapsed with no match, or the file could not be read. |
2 | The directory being watched doesn't exist. |
3 | A candidate file was last modified outside the window. |
-1 | The end offset was earlier than the start offset. |
Unlike Windows - File Arrival, the UNIX type offers no exit criteria (on the How It Works tab, under Exit Criteria) table. The watcher's own exit
status is the verdict — only 0 is a success. The exit code still stays on the job, so you can
tell which failure you had.
As with the other legacy job types, a bad value saves cleanly and then fails when the job runs.
The relay refuses a missing file name (RC_MISSING_FILEARRIVAL_FIELD); and a file name over 255
characters, one holding a refused character, whitespace in the directory part, an offset or stable
time that isn't a whole number in range, or an End Offset earlier than Start Offset
(RC_INVALID_FILEARRIVAL_FIELD).
Embedded Script
UNIX/Linux LSAM — Embedded Script
The UNIX - Embedded Script job type (unix-embedded-script) runs a script held in the script
library rather than one pasted into the job — the UNIX counterpart of
Windows - Embedded Script. On the job you choose the script, the
version to run, the runner that starts it, and any arguments:
| Field | Notes |
|---|---|
| Connection | Required. It is also the account that owns the script file the agent writes for the run. |
| Script | The library script to run. Only scripts whose language has a UNIX runner are listed — Shell, Bash, Perl and Python in the built-in catalog, plus any script type your tenant has added a UNIX runner for. |
| Version | Leave empty to always run the latest version, or pin a specific version number. |
| Runner | The program that runs the script on the agent. The choices depend on the script's language, so pick the script first. |
| Arguments | Optional, up to 2000 characters. |
It carries the ordinary UNIX job settings as well — exit criteria (on the How It Works tab, under Exit Criteria), Signal Criteria, Core Dump Flag and Environment Variables.
Arguments are inserted unquoted into the runner's command line, so the agent's shell interprets
quotes, $VAR references and metacharacters such as ; | & and >. Quote them yourself if you
mean them literally. This is the same rule as
UNIX values reach the agent exactly as you typed them (on the UNIX/Linux tab, under Command).
As on Windows, the job holds a reference rather than a copy: fixing the script in the library reaches every job that tracks the latest version, and a script cannot be deleted while a job references it. The script and the runner are also checked when the workflow is saved, and a job whose script or runner has been deleted says so on the field rather than showing a bare identifier. See Script library.
The built-in UNIX runners name their interpreter by absolute path — /bin/sh, /bin/bash,
/usr/bin/perl, /usr/bin/python3 — and a host without the interpreter at that path fails when the
script starts. If your machines keep an interpreter somewhere else, add a runner that names the path
they actually use; a UNIX command format is limited to 127 bytes and $FILE has to be its own
space-separated word. See Runners.
Episys
UNIX/Linux LSAM — Episys (RSJ) job types
Seven job types drive Symitar/Episys core banking on a UNIX LSAM machine through the RSJ utilities installed there. Each one is a form that builds the RSJ command line for you, so there is no command-line field to fill in.
| Job type | What it does |
|---|---|
UNIX - Episys: Run JobFile (unix-episys-run-jobfile) | Runs a batch job file from the partition's BATCH directory. |
UNIX - Episys: Answer Prompts (unix-episys-answer-prompts) | Fills in the prompts in a batch job's control file. |
UNIX - Episys: Compare ACH Totals (unix-episys-compare-ach-totals) | Compares a FED ACH file's totals against an Episys batch output. |
UNIX - Episys: Find Batch Output Sequence Number (unix-episys-find-batch-output-sequence) | Finds a job's batch output sequence number and writes it to a property. |
UNIX - Episys: Find Report from RSJ Output (unix-episys-find-report-rsj-output) | Finds a report by title in a job's RSJ output and writes its sequence number to a property. |
UNIX - Episys: Find Report from Episys Reports (unix-episys-find-report-episys-reports) | Finds a report by title in the partition's report directory and writes its sequence number to a property. |
UNIX - Episys: FTP all Reports in List (unix-episys-ftp-reports) | Sends every report listed in a file to an FTP destination. |
Fields every Episys type shares
| Field | Notes |
|---|---|
| Connection | Required. RSJ and the Episys utilities run as this UNIX user. |
| RSJ Path | Where RSJ is installed on the machine. It defaults to [[RSJPATH]], which reads a global property of that name — the job fails before it starts if that property does not exist. Create RSJPATH once and every Episys job picks it up, or type a path on the job to override it. |
| SYM Number | The Symitar partition number, such as 001. The SYM prefix is added for you. |
Episys Job, where the type takes one, names the batch job file. Run JobFile allows up to 30 characters, because that is RSJ's own limit; the others allow 32.
Per-type fields worth knowing
- Run JobFile takes an optional Restart Point (a label in the job file to resume from), an optional Edit File, a Multi-thread switch that runs without RSJ's partition lock, and Delete edit file to remove the edit file when the job succeeds. Edit File and Multi-thread cannot be combined. It takes exit criteria: Exit Criteria Result and Exit Criteria follow Delete edit file, and Continuum always evaluates the table itself. Leave it empty and the job fails on any non-zero exit code, as before.
- Answer Prompts takes one row per prompt, each a Prompt and a Response; at least one row
is required. A prompt may not contain a colon or a double quote, and a response may not contain a
double quote. A response may end in
:Nto change the Nth occurrence instead of the first. Update First Match Only is on by default. Extract Daily/Monthly Group List in Use converts^to,in each response before it is applied. - Compare ACH Totals takes the FED File and the Batch Output File. A bare name is looked up in the partition's own directories rather than requiring a full path.
- The three Find types write their answer to the Property Name you give, and take a MSGIN Directory used when the utility cannot reach the agent directly. The two report-finding types take a Report Name and an optional Occurrence To Search For (1–99; the first, if empty).
- Find Report from RSJ Output is the other Episys type with exit criteria: Exit Criteria Result and Exit Criteria follow Occurrence To Search For, and Continuum always evaluates the table itself. Leave it empty and the job fails on any non-zero exit code.
- FTP all Reports in List takes the list file as Report Name, a Destination Host Name, a Port Number (21 by default), a Destination Folder, and an optional Extension added to each remote file name.
On the two report-finding types, Report Name is passed to the utility without quoting. For a
title containing spaces, type the quotes yourself — "ACH REPORT".
Episys credentials
Four of the seven types need a credential of their own, on top of the batch user, and it is held as a saved connection rather than typed into the job:
| Connection type | Used by | Holds |
|---|---|---|
Episys External Event Credential (episys-event-credential) | The three Find types | The OpCon user and External Event Password the utility sends the property update with. |
Episys FTP Credential (episys-ftp-credential) | FTP all Reports in List | The FTP user and password to log in with. |
On both, leave the password blank when editing to keep the one already stored.
The External Event Password is sent in double quotes, so it may not contain a double quote.
The FTP user and password
are passed unquoted and may not contain spaces; on an agent that runs jobs through the shell —
the default — characters such as $ ` \ & and ; in the FTP password are interpreted by
the shell.
The UNIX agent prints the full command line at the top of every job's output, and for these four job types that command line contains the password. Anyone who can read the job's output can read the credential, and it stays in the stored output until the job is rerun or the instance is deleted — refreshing the output re-reads the same text rather than clearing it.
Treat these credentials as visible to everyone who can see the job, and give them the narrowest rights that let the utility do its work. There is also no per-connection permission on picking one: anyone who can author an Episys job can select any Episys credential saved in your tenant.
IBM i
The IBM i LSAM connector (opcon.lsam.ibmi) has these job types.
| Job type | Agent type | Purpose |
|---|---|---|
IBM i Command (ibmi-command) | IBM i LSAM | Submit a batch job on an IBM i (AS/400) LSAM machine — see below |
IBM i - Operator Replay (ibmi-operator-replay) / IBM i - Restricted Mode (ibmi-restricted-mode) | IBM i LSAM | Run an Operator Replay or Restricted Mode script on an IBM i LSAM machine |
IBM i - FTP (ibmi-ftp) | IBM i LSAM | Send, receive or delete files over FTP from an IBM i LSAM machine |
IBM i - File Arrival (ibmi-file-arrival) | IBM i LSAM | Wait for a DB2 file or an IFS file to arrive on an IBM i LSAM machine |
- Batch Jobs
- Batch Environment
- Lists
- FTP
- File Arrival
- Dispatch Rules
Batch Jobs
IBM i LSAM
The five IBM i job types are registered and appear in the job type catalog, but their end-to-end run path has not been verified on a real IBM i system. Confirm availability with your Continuous contact before planning work around them.
One blocker has been removed. Every IBM i job dispatched to a real LSAM used to fail immediately with Initialization Error: the name Continuum sent contained a hyphen, which an IBM i job name cannot hold, so the agent refused the submission. The name is now built from letters, digits and underscores only. This needs a current relay — see Keeping a relay current.
The job runs on the IBM i under a name taken from its own job name. The relay takes the first
12 characters of the job's name, replaces anything other than a letter, digit or underscore with
_, uppercases it, and puts J in front if it does not start with a letter — Daily Backup
becomes DAILY_BACKUP, 2026 report becomes J2026_REPORT. The agent shortens that to the
10 characters an IBM i job name allows. Earlier relays named every IBM i job JOB_ followed by
characters of an internal identifier, so the job could not be recognised by its name on the IBM i. This needs a current relay.
The IBM i LSAM connector (opcon.lsam.ibmi) runs work on an existing IBM i (AS/400) LSAM
machine, dispatched through the relay like the UNIX and Windows LSAM job types. It brings IBM i
work that already runs on an LSAM into an OpCon Continuum workflow without rebuilding it. It
provides five job types:
| Job type | What it runs | What you give it |
|---|---|---|
IBM i Command (ibmi-command) | A batch job | The Call — the CL command or program to run |
IBM i - Operator Replay (ibmi-operator-replay) | An Operator Replay script | The Script Name |
IBM i - Restricted Mode (ibmi-restricted-mode) | A Restricted Mode script | The Script Name |
IBM i - FTP (ibmi-ftp) | A file transfer to or from an FTP server, or a remote delete | The FTP fields (on the IBM i tab, under FTP) — there is no Call and no script |
IBM i - File Arrival (ibmi-file-arrival) | A wait for a DB2 file or an IFS file to arrive | The File Arrival fields (on the IBM i tab, under File Arrival) — there is no Call and no script |
The first four take the same batch user, prerun and batch environment (on the IBM i tab, under Batch Environment) fields, and three of them take the IBM i lists (on the IBM i tab, under Lists) — a message monitor, spool-file actions and dynamic variables. IBM i - Restricted Mode takes dynamic variables only. IBM i - File Arrival takes a smaller set: no prerun, no library or job description fields, and no spool-file list.
One difference matters for how you detect failure: these job types have no exit-criteria setting. IBM i does not expose an exit-code field for the relay to translate, so the Exit Criteria you would set on a UNIX or Windows LSAM job are not available here, and OpCon Continuum's own central evaluation excludes IBM i deliberately — the agent's reported status is the authority on this platform. What IBM i offers instead is the message monitor (on the IBM i tab, under Lists): the agent watches the job's messages and can answer them, send an event, or end the job. Check the IBM i job log on the machine when you need the reason a job ended as it did.
An IBM i job's output no longer reports a signal it never received. An IBM i agent reports its
outcome as an IBM i message ID — SMA0036, or SMA0023-CPF0006 — where a UNIX agent reports a
numeric exit code. That field used to be read as though it were the UNIX one, so digits inside the
message ID were reported as a Signal: line in the job's output: a job that failed with
CPF0006 claimed to have been stopped by signal 6. The field is no longer parsed on this platform,
and the message ID is what the job records.
The message ID is the job's Exit Description
The message ID an IBM i agent reports when a job ends is now the job's termination description — the Termination row on the job's Summary tab, and the value an Exit Description event compares. It is set on success and on failure alike, exactly as the agent sent it:
| The agent reports | Typically when |
|---|---|
SMA0036 | The job finished OK |
SMA0037 | The job failed |
A pair such as SMA0023-CPF0006 | The job could not be submitted — here, because of errors in the command |
Earlier builds dropped this field on its way to Continuum, so an IBM i job's description was blank when it finished OK and held the agent's failure text when it failed.
- The agent's failure text moves to Agent Message. When the agent also sent a reason for a failure, it appears in the Agent Message row below Termination — unless it is the same text, which is then shown once.
- A prerun failure is unchanged. A job whose prerun failed (on the How It Works tab, under Shared Parameters) is waiting to be re-attempted, not ended, and keeps the agent's message as before.
- It needs a current relay. An older relay does not pass the field on, and the job keeps the earlier behavior — see Keeping a relay current.
An Exit Description event's Value takes text, so Equal To SMA0037 fires on a job that
ended with that message ID, and on no other. The comparison ignores letter case and trailing
spaces. Contains 0037 also matches SMA0037 — and any other message ID that holds those
characters, such as SMA0023-CPF0037. See
Exit-description events.
IBM i — core parameters
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target IBM i LSAM machine. Comes from the job's agent assignment, and must match the agent name registered with the relay. |
| Call | Yes on IBM i Command | 1–2000 chars | The CL command or program to run. |
| Script Name | Yes on Operator Replay and Restricted Mode | 1–2000 chars | The name of the Operator Replay or Restricted Mode script the LSAM runs. |
| Connection | No | a saved connection | The IBM i user profile the job runs as — an IBM i Batch User connection. Blank uses the job description's default user. See Connection. |
| Prerun | No | ≤2000 chars | A command run before the job itself — typically a precondition test such as CHKDTAARA or CHKFILE. A prerun that fails does not fail the job — see A prerun that fails is re-attempted (on the How It Works tab, under Shared Parameters). |
Each type reads only its own field. A Call left over on an Operator Replay job, or a Script Name on an IBM i Command job, is never sent. A Call or Script Name that holds nothing but spaces or control characters counts as missing, and the job is refused at dispatch rather than sent with an empty one.
Batch Environment
IBM i — batch environment
These map to the IBM i batch attributes you would set on the machine. Leave them alone unless the job needs a specific batch environment — the values a new job starts with reproduce what the job description already does.
| Parameter | Default | A new job starts with | Offered in the list |
|---|---|---|---|
| Current Library | * | the default | *, *CRTDFT, *USRPRF |
| Initial Library List | * | *JOBD | *JOBD, *, *NONE, *SYSVAL, *CURRENT |
| Job Description | * | the default | *, *USRPRF, SMALSAJ00 |
| Job Description Library | * | the default | *, *LIBL, *CURLIB, SMADTA |
| Job Queue | * | *JOBD | *JOBD, * |
| Job Queue Library | * | *LIBL | *LIBL, *, SMADTA, *CURLIB |
| Output Queue | * | *JOBD | *JOBD, *, *CURRENT, *USRPRF, *DEV |
| Output Queue Library | * | *LIBL | *LIBL, *, SMADTA, *CURLIB |
| Job Date | *JOBD | the default | *JOBD, *SYSVAL, *SCHEDULE DATE |
| Job Queue Priority | * | the default | *, 1–9 |
| Accounting Code | — | blank | Any value up to 15 characters |
| Message Logging Level | * | *JOBD | *JOBD, *, 0–4 |
| Message Logging Severity | * | *JOBD | *JOBD, *, 00–99 |
| Message Logging Text | * | *JOBD | *JOBD, *, *MSG, *SECLVL, *NOLIST |
| Log CL Commands | * | *JOBD | *JOBD, *, *NO, *YES |
| Inquiry Message Reply | * | *JOBD | *JOBD, *, *RQD, *DFT, *SYSRPYL |
| Job Log Retention Occurrences | 0 | the default | 0–999 |
| Job Log Retention Days | 0 | the default | 0–999 |
A new job starts with preselected values. When you add an IBM i job, the editor fills in the values in the third column once, as the job is created. A job that is already saved is never changed: it keeps the values it holds, and a field it has no value for shows the default. Opening a job on a running workflow instance fills in nothing either.
IBM i - File Arrival takes every one of these except the four library and job description fields — Current Library, Initial Library List, Job Description and Job Description Library. Its form does not offer them, and a value left in the job from another job type is never sent.
An attribute with a default is still sent with it when you never touch it. The job editor doesn't
store an untouched default, so a value you leave blank is filled in on the way to the machine rather
than being sent empty — an IBM i batch job would reject an empty attribute. A blank
Accounting Code, and a retention value of 0, are not sent at all.
Six of these are pickers in the job editor: Job Queue Priority, the three message-logging fields, Log CL Commands and Inquiry Message Reply. Those are the ones IBM i treats as a closed set of values, so there is nothing to mistype. A job saved earlier with a value outside the list still shows the value it holds rather than reading as blank.
The library, queue, job-description and job date fields offer suggested values and still accept your own. Open the field to pick one of the values in the last column, or type a library, queue or date — or a property token — and move on: what you typed is kept, without picking it from the list. The list also offers it as Use this value. A value the job already holds that is not in the list is listed as itself.
- Escape discards what you typed and leaves the field as it was. It never empties the field.
- The X in the field empties it. Emptying the text by deleting it does not; the field goes back to its value when you leave it.
How the form is laid out
The IBM i fields are grouped in sections you can collapse, in the order you work through them. Collapsing a section changes nothing in the job.
| Job type | Sections, in order |
|---|---|
| IBM i Command, Operator Replay, Restricted Mode | Job Information (Connection, job description, libraries and job queue) · Call Information (Prerun, then the Call or Script Name) · Additional Information |
| IBM i - FTP | Job Information · Call Information (Prerun) · Transfer Information (Action, FTP User, Transfer Type) · Remote Information · Local Information · Additional Information |
| IBM i - File Arrival | Job Information (job queue and its library) · File Arrival Information (Connection, File Name, window and checks) · Additional Information |
Additional Information — the output queue, job queue priority, job date, accounting code, message logging and job log retention — starts closed, because none of its fields is required. Every other section starts open. The lists follow the sections.
Lists
IBM i lists
Three repeating lists sit at the end of the IBM i job form. Each holds up to 99 rows, and every one is optional.
| List | IBM i Command | Operator Replay | Restricted Mode | FTP | File Arrival |
|---|---|---|---|---|---|
| Message Monitor | Yes | Yes | — | Yes | Yes |
| Spool Files | Yes | Yes | — | Yes | — |
| Dynamic Variables | Yes | Yes | Yes | Yes | Yes |
Restricted Mode has no message or spool list. The editor does not offer them; a Restricted Mode job that carries either anyway — from the API or an import — is refused at dispatch rather than having the rows dropped, because a silently dropped reply monitor would leave the job waiting on its inquiry message. File Arrival has no spool list, and a File Arrival job that carries one is refused at dispatch the same way.
Message Monitor
Each row is an IBM i message the LSAM watches for while the job runs, and what it does when the message appears.
| Column | Limits | Notes |
|---|---|---|
| Severity (at least) | 00, 10 … 90; default 00 | 00 matches by Message ID. Above 00, the row matches any message of that severity or higher, and Action must be Send event. |
| Message ID | 7 chars | Used only when Severity is 00, and then required. A seven-character IBM i message identifier such as CPA0701: a letter, two letters or digits, then four hexadecimal digits. Upper-cased for you. Above 00 it is greyed out, and choosing another severity clears it. |
| Compare Data | ≤30 chars | Optional text the message data must contain for the row to match. |
| Position | 0–999 | Where Compare Data starts in the message data. Sent only with Compare Data, and greyed out until Compare Data holds text; emptying Compare Data clears it. |
| Action | Reply (default), Send event, Reply and send event | What the LSAM does when the message appears. |
| Reply | ≤6 chars | The reply to the inquiry message. Required for Reply and Reply and send event; greyed out, and cleared, for Send event. |
| End Job | No (default) / Yes | End the job when the message appears. |
| Event | ≤738 chars | The OpCon event command the LSAM sends, for example $CONSOLE:DISPLAY,Message received. Passed to the agent exactly as written. Required for Send event and Reply and send event; greyed out, and cleared, for Reply. A new row starts with Event and Position greyed out. |
Whether an event a real IBM i agent sends this way reaches Continuum has not been verified yet. Rely on the reply and End Job columns, not on the event, until it has.
Spool Files
Each row names spooled files from this job and what the LSAM does with them.
| Column | Limits | Notes |
|---|---|---|
| File Name | ≤10 chars; default *ALL | Required. A spooled file name, or *ALL. |
| User | ≤10 chars | The user profile that owns the spooled files. |
| Output Queue, Output Queue Library | ≤10 chars each | The output queue that holds them, with its library. |
| Total Copies | 1–255; default 1 | Number of copies. |
| Hold, Save | LSAM default (*) (default), Yes, No | Whether to hold the spooled files, and whether to save them after they print. |
Give either a User or an Output Queue with its Output Queue Library — exactly one
of the two. An Output Queue of *USRPRF needs a library of *USRPRF too. The name columns are
upper-cased for you.
The name columns suggest values you can pick or type over: *ALL for File Name, * for
User, *USRPRF for Output Queue, and *LIBL or *USRPRF for Output Queue Library.
Dynamic Variables
Each row is an IBM i LSAM dynamic variable the agent sets before the job starts, so the Call or script can use it. These are the LSAM's own variables, not Continuum properties.
| Column | Limits | Notes |
|---|---|---|
| Variable Name | ≤12 chars | Required. Letters, digits, $, @ and # only, and each name once per job. Upper-cased for you, so myvar and MYVAR are the same name. |
| Value | ≤128 chars | Required. May contain [[property]] tokens, which are resolved when the job is dispatched. |
An older relay builds every IBM i job as a batch job. Given the new job types it fails them loudly, because they carry no Call. A job with message monitors, spool files or dynamic variables is refused at dispatch by a relay that has not reported it can send them, so it never runs without its lists: it goes to Initialization Error with RELAY_LSAM_FEATURE_UNSUPPORTED: relay '…' (agent pool '…') has not reported support for … Upgrade the relay. A relay older than this release has not reported it, even one that used to send the lists. Upgrade the relay before you add lists to a job — see Keeping a relay current.
FTP
IBM i - FTP
IBM i - FTP sends, receives or deletes files on an FTP server from the IBM i machine. It takes no Call and no script; you describe the transfer in fields instead, and the Action decides which of them it uses.
| Field | Limits | Used by | Notes |
|---|---|---|---|
| Action | see below; default PUT | every action | What the job does. |
| Remote System | ≤255 chars | every action | Required. The FTP server — a host name from the IBM i host table, or an IP address in single quotes. It may be followed by the LSAM's job-parameters separator character and extra IBM i FTP command keywords. |
| Remote Library or Directory | ≤512 chars | every action | Required. The library or directory on the remote system. |
| Remote File Name | ≤128 chars; default *LCLFILNAM | every action except MPUT | The file on the remote system. *LCLFILNAM, offered in the field's list, uses the local file name. For MGET and MDEL, a pattern that selects several files. |
| Transfer Type | BIN (default), ASCII, EBCDIC | PUT, MPUT, GET, MGET | How the file is transferred. |
| Local File Name | ≤128 chars | PUT, MPUT, GET | Required for those actions. The file on the IBM i; for MPUT, a pattern that selects several files. |
| Local Library or Directory | ≤512 chars | PUT, MPUT, GET, MGET | Required for those actions. The IBM i library or IFS directory that holds or receives the files. |
| FTP User | ≤128 chars | every action | The user that signs on to the FTP server. Leave it empty to use the LSAM's default FTP user. |
| Action | Does |
|---|---|
| PUT | Send a file |
| MPUT | Send multiple files |
| GET | Receive a file |
| MGET | Receive multiple files |
| DEL | Delete a remote file |
| MDEL | Delete multiple remote files |
A field the Action does not use is greyed out. The Used by column above is the rule. When you change the Action, every field the new one does not use is greyed out and its value cleared, in the same change — PUT to DEL clears Transfer Type and both local fields, and PUT to MPUT clears Remote File Name. A move between two actions that use the same fields, such as PUT to GET, keeps every value. Opening a saved job clears nothing, and a value a job still holds in a field its Action does not use — saved before this change, or written through the API — is not read and never sent.
No password is sent from Continuum. The IBM i LSAM supplies the FTP user's password itself, as it does for an FTP job defined on the machine.
A quoted IP address in Remote System reaches the agent as you typed it, quotes included — see IBM i values reach the agent as you typed them (on the IBM i tab, under Dispatch Rules).
An FTP job carries the same batch environment fields and lists as an IBM i Command job.
File Arrival
IBM i - File Arrival
IBM i - File Arrival waits for a file to arrive on the IBM i machine — a DB2 file or member, or a
file in the IFS — and succeeds once it is there and its size has stopped changing. The IBM i LSAM
does the watching itself, so there is no Call, no script and no prerun. A nightly core job can wait
on PAYLIB/DAILYTRN this way before the jobs that load it are released.
| Field | Limits | Default | Notes |
|---|---|---|---|
| File Name | 1–4000 chars | — | Required. A DB2 file as LIBRARY/FILE or LIBRARY/FILE(MEMBER), or an IFS path starting with /. The wildcards * and ? may be used in the file-name part. A tab, a line break or any other control character inside the name is not allowed. Spaces and line breaks at either end are trimmed. |
| Creation Window Start | −99 to 99 days, with a time | 0 | The earliest creation time a matching file may have, as a Day Offset and a Time from midnight at the start of the job's schedule date (on the How It Works tab, under Overview). 0 means no start limit. |
| Creation Window End | −99 to 99 days, with a time | 0 | The latest creation time a matching file may have, on the same scale. A job that keeps checking also stops here, unless a dynamic variable sets another end time. 0 means no end limit. When both ends are set, the end must not be earlier than the start. |
| File Size Stable Duration (seconds) | 0–999 | 5 | How long the file's size must stay unchanged before it counts as arrived. 0 accepts the file as soon as it is found. |
| Failure Condition | see below | Fail on error | Which outcomes fail the job. |
| Check Lock on DB2 File | on / off | off | Fail with completion code CKF0003 if the DB2 file is locked. Ignored for an IFS path. |
| Check Read Authority, Check Write Authority, Check Execute Authority | on / off | off | Check that the Connection's user profile has that authority to the file. Each one needs a connection. |
| File Name Variable | ≤12 chars | — | An IBM i LSAM dynamic variable that receives the found file's name when the job succeeds. |
| Record Count Variable | ≤12 chars | — | An IBM i LSAM dynamic variable that receives the found file's record count (DB2) or byte count (IFS) when the job succeeds. |
| Connection | a saved connection | — | The profile whose authority to the file is checked. Leave it empty when no authority check is set — the picker reads None — no authority check. |
The Failure Condition list offers Fail on error (the default), Always succeed, Fail on error, except when the end time is reached, and Fail if the file contains zero records or zero bytes.
Whichever you choose, the agent's completion code is the job's Termination — see The message ID is the job's Exit Description (on the IBM i tab, under Batch Jobs).
Without a re-check interval the file is checked once. To keep watching, add the agent's own
$@ keywords as dynamic variables (on the IBM i tab, under Lists): $@RECHKFREQ sets the seconds between checks,
$@JOBENDTIME and $@JOBREFDATE set when to stop watching, $@CRTREFDATE sets the date the
creation window is counted from, and $@AUT and $@FAILIFZERO adjust the checks. These are read by
the IBM i LSAM, not by Continuum.
A File Arrival job also carries the batch environment (on the IBM i tab, under Batch Environment) fields other than the library and job description ones, a message monitor (on the IBM i tab, under Lists) and dynamic variables — but no spool-file list. Like every IBM i job type it has no exit-criteria setting: the agent's completion code and the Failure Condition decide the outcome.
An older relay fails a File Arrival job at dispatch, so nothing waits on the machine: it refuses it as an unsupported IBM i job type, or — on a relay that predates the IBM i job types altogether — builds it as a batch job and finds no Call. See Keeping a relay current.
Dispatch Rules
What is refused at dispatch
The job form checks only that the top-level required fields are filled in. The rest is checked by the relay at dispatch, so a job with a bad value saves cleanly and fails when it runs — without anything reaching the machine. Every message names the list, the row and the column, or the field, and never the value:
- a missing Call (IBM i Command) or Script Name (Operator Replay, Restricted Mode);
- for File Arrival, a missing File Name, or one that holds a control character such as a tab or line break; a Creation Window End earlier than its start; a window value, stable duration or Failure Condition outside its range; or an authority check with no Connection;
- for FTP, a field the chosen Action requires and the job lacks — IBM i FTP PUT requires a Local
File Name — or an Action or Transfer Type outside its list. The list is exact and
case-sensitive:
putis refused; - a list with more than 99 rows — the whole job is refused, rather than the extra rows being dropped;
- a row that breaks its column rules on the Lists tab — IBM i message monitor row 1: Reply is required when Action is Reply or Both;
- a message or spool list on a Restricted Mode job, or a spool list on a File Arrival job;
- a dynamic-variable name used twice;
- the text
</F>or<F I=in any value — see IBM i values reach the agent as you typed them.
A list is checked in full before anything is sent, so one bad row fails the job; a partial monitor never reaches the machine.
The event command and the dynamic-variable values are redacted from the relay's connector log.
The agent still receives them; the log shows *** in their place. The message IDs, spool-file names
and variable names stay visible, so you can still tell which row was sent.
IBM i values reach the agent as you typed them
An IBM i job's values — the Call, the FTP fields, list rows, every other field — are sent to the
agent unaltered. Quotes, ampersands and angle brackets arrive as the characters you wrote, so a
Call such as CALL PGM(NIGHTLY) PARM('A&B'), or an FTP Remote System given as an IP address in
single quotes, reaches the machine intact.
This is a correction. The relay used to convert those characters to XML entities before sending
them, so PARM('A&B') reached the machine as PARM('A&B'). For the
same reason, an IBM i - File Arrival job used to refuse & ' < > and " in its
File Name; it no longer does. It needs a current relay.
As on UNIX, two character sequences cannot be carried in a value sent unaltered, and are refused at
dispatch naming the field code, so the job fails without running: </F>, which the agent would read
as the end of your value, and <F I=, which it could read as the start of another field. Both
matches are case-sensitive. See UNIX values reach the agent exactly as you typed them (on the UNIX/Linux tab, under Command).
SQL
The SQL LSAM connector (opcon.lsam.sql) has these job types. Each has its own tab below.
| Job type | Agent type | Purpose |
|---|---|---|
SQL - MS SQL Script (sql-script) | SQL LSAM | Run a T-SQL script, or inline statements, on a SQL LSAM machine |
SQL - MS SQL Job (sql-agent-job) | SQL LSAM | Start a SQL Server Agent job on a SQL LSAM machine and report its outcome, or monitor one already running |
SQL - MS SQL DTExec (sql-dtexec) | SQL LSAM | Run an SSIS package with DTExec on a SQL LSAM machine, from msdb, the file system or the SSIS catalog |
SQL - Oracle (sql-oracle) | SQL LSAM | Run an Oracle SQL*Plus script on a SQL LSAM machine |
SQL - Other DB (sql-other-db) | SQL LSAM | Run SQL through an ODBC driver or OLE DB provider on a SQL LSAM machine |
SQL - MySQL (sql-mysql) | SQL LSAM | Run a MySQL script file with the mysql command-line client on a SQL LSAM machine |
Each name starts with SQL - in the Job Type list and on the canvas, so the six sort together, the way the Windows and IBM i types do under their own prefix. This page uses the short names after this table — MS SQL Script, MS SQL Job, MS SQL DTExec, Oracle, Other DB and MySQL. Saved jobs keep their job type, and what reaches the agent is unchanged.
- MS SQL Script
- MS SQL Job
- MS SQL DTExec
- Oracle
- Other DB
- MySQL
MS SQL Script
SQL LSAM — MS SQL Script
The SQL LSAM connector (opcon.lsam.sql) runs a T-SQL script on an existing SQL LSAM machine,
dispatched through the relay like the other legacy job types. It brings SQL Server work that
already runs on an LSAM into an OpCon Continuum workflow without rebuilding it.
The machine runs the script with sqlcmd, and most parameters below map directly onto a
sqlcmd switch — which is what makes Other Options usable.
SQL — core parameters
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target SQL LSAM machine. Comes from the job's agent assignment, and must match the agent name registered with the relay. |
| Connection | No | a saved connection | The account the job runs as — a SQL Batch User connection. Blank runs it as the agent's own service account, with no impersonation. See Connection. |
| Server Name\Instance | Yes | 1–255 chars | The SQL Server host, optionally HOST\INSTANCE for a named instance. |
| Database Name | No | ≤255 chars | The initial database. Left blank, the server's default for the login applies. |
| Windows Authentication | No | default off | On, the job authenticates to SQL Server as the impersonated Windows account rather than with a SQL Server login and password. |
| Encrypt Connection | No | default on | Encrypts the connection to SQL Server. |
SQL — the script
Script Source says where the SQL comes from, and the form shows only the field that choice uses:
| Parameter | Limits | Notes |
|---|---|---|
| Script Source | SQL statements / Script file | A new job starts on SQL statements. Choosing Script file hides SQL Script Statements and clears what was typed there; choosing SQL statements hides and clears Script File Path. |
| SQL Script Statements | ≤4000 chars | Inline T-SQL, typed into the job. Shown unless Script Source is Script file. |
| Script File Path | ≤255 chars | Path to a .sql file on the agent machine — for example D:\scripts\nightly.sql. This is not a script library reference, unlike Windows - Embedded Script. Shown unless Script Source is SQL statements. |
A job saved before Script Source existed has none set, so it shows both fields, as it always
did, and runs exactly as before. Either way, the job must carry exactly one of the two. Neither,
or both, is rejected at dispatch — the form will let you save it. A Script Source that
disagrees with the field that is filled in — Script file with only statements, say, as a job
written outside the editor could hold — is rejected at dispatch too, naming scriptSource.
SQL — output and options
| Parameter | Limits | Notes |
|---|---|---|
| Use Exit Code From Script Result | default off | Returns the script's own result as the job's exit code. Applies to inline statements only — it has no effect when Script File Path is used, and it is hidden while Script Source is Script file (its setting is kept). |
| Exit Criteria Result | Fail / Finish OK | What a match decides for the criteria table. Default Fail. |
| Exit Criteria | ≤20 conditions | Conditions tested against the job's exit code. See Exit criteria. |
| Redirect File Path | ≤255 chars | Writes the script's output to this file on the agent. |
| Other Options | ≤255 chars | Additional sqlcmd switches, appended verbatim — for example -t 300. |
| Environment Variables | name/value pairs | Environment variables set for the sqlcmd process. |
Two things about the parameters above are worth setting deliberately:
- Exit criteria are available here, and OpCon Continuum always evaluates them. The SQL agent has no way to receive a criteria table, so the table is applied centrally to the exit code the agent reports rather than on the machine — see Exit criteria (on the How It Works tab, under Exit Criteria). This is the only legacy platform where that is true of every table, including a plain one-row table.
- Know which code you are writing conditions against. With Use Exit Code From Script Result
off,
sqlcmdexits0on success and1on an error, so that is the whole range a table has to work with. Turn it on and the script's own result becomes the exit code, which is what makes a criteria table worth having on this type. - An environment variable with an empty value is not sent. Both halves of a pair have to carry a value to reach the machine, so a name with a blank value is dropped as a pair rather than arriving with an empty value or shifting the variables after it.
With Encrypt Connection on, sqlcmd requires SQL Server to present a certificate the agent
trusts. Against a server that has no trusted certificate the job fails when it tries to reach the
database — after a clean dispatch, so it reads as a job failure rather than a routing problem.
Either install a trusted certificate on SQL Server or turn Encrypt Connection off.
- A SQL machine cannot be either end of a file transfer. It is the one legacy platform with no file-transfer capability at all, so it has no file-transfer endpoint and never appears in a SMAFT File Transfer job's machine pickers.
- A legacy agent group can be of type SQL, and routes like any other group. Adding an agent of a different type to it is rejected.
- Job output files are retrievable for a SQL job the same way as for the other legacy job types — what the script printed is available from the job's output.
MS SQL Job
SQL LSAM — MS SQL Job
The second SQL LSAM job type starts an existing SQL Server Agent job on a SQL Server
instance and reports its outcome — a nightly ETL kick-off, a replication monitor, a maintenance plan
you already have defined in msdb. It can also be pointed at a run that is already in progress
and only watch it.
Unlike MS SQL Script (on the SQL tab, under MS SQL Script), this type is not a sqlcmd wrapper: the agent
connects to SQL Server itself and drives the SQL Server Agent job in process. So there is no
database name, no script, and no Encrypt Connection setting on it.
MS SQL Job — parameters
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target SQL LSAM machine. Comes from the job's agent assignment, and must match the agent name registered with the relay. |
| Connection | Yes | a saved connection | The account the job runs as — a SQL Batch User connection. Required on this type, unlike MS SQL Script: a job with none is refused at dispatch. With SQL Server authentication it is the SQL login; with Windows Authentication it is the Windows account the agent impersonates. |
| Server Name\Instance | Yes | 1–255 chars | The SQL Server instance that owns the job — HOST\INSTANCE for a named instance, or HOST,1433 to name a port. |
| SQL Job Name | Yes | 1–128 chars | The name of the SQL Server Agent job to start or monitor. |
| Windows Authentication | No | default off | On, the job connects to SQL Server as the batch user's Windows account rather than as a SQL login. |
| Monitor Only | No | default off | Don't start the job — watch a run that is already in progress. See below. |
| Monitoring End Time | No | 0 to 143999 | When to stop watching, as a Day Offset and a Time from midnight of the job's schedule date on the agent — the same pair the File Arrival (on the Windows tab, under File Arrival) window uses, without the negative days. Day 0 / 00:00 means watch until the job completes. |
| Connection Retry Attempts | No | 0 to 99 | How many times to retry the connection to SQL Server, 5 minutes apart. |
| Exit Criteria Result | No | Fail / Finish OK | What a match decides for the criteria table. Default Fail. |
| Exit Criteria | No | ≤20 conditions | Conditions tested against the job's return code, always evaluated by Continuum. See Exit criteria. |
The batch user needs rights in msdb to see and start the job. For a job it does not own, that
means SQLAgentOperatorRole.
The return codes you write criteria against
This type does not return a program's exit code — it returns the SQL Server Agent job's outcome, as a small fixed set:
| Code | Means |
|---|---|
0 | Succeeded |
1 | Failed |
2 | Retry |
3 | Cancelled |
4 | In progress, or a monitoring error |
5 | Unknown — also what Monitor Only reports for a job that has never run |
-1 | A monitoring error before the first status check |
With no criteria table, only 0 finishes the job OK and every other code fails it. A table is
how you change that — to treat 2 (Retry) as a success you want to see, for instance. The SQL agent
cannot receive a criteria table at all, so Continuum always evaluates this type's table centrally,
exactly as it does for MS SQL Script.
Monitor Only, and when monitoring stops
When Monitoring End Time is reached, the job is reported Finished OK even though the SQL Server Agent job is still running — and the SQL job is not stopped. Finished OK here means stopped watching, not succeeded.
If the end time has already passed when the job starts, the SQL Server Agent job is still started (unless Monitor Only is on), left running unmonitored, and the Continuum job reports Finished OK immediately. A Day Offset and Time in the past is therefore a way to fire a SQL job and forget it, not a way to make the job fail.
Leave Monitoring End Time at day 0 / 00:00 if you want the job's outcome to be the SQL
Server Agent job's real outcome.
Monitor Only does not start anything. Two cases are worth knowing:
- If the SQL Server Agent job is not running when monitoring begins, the outcome of its
previous run is reported. A job that has never run reports
5(Unknown). - Cancelling the Continuum job stops the SQL Server Agent job, even with Monitor Only on — so cancelling a watch-only job stops work it did not start.
MS SQL DTExec
SQL LSAM — MS SQL DTExec
The third SQL LSAM job type runs a SQL Server Integration Services (SSIS) package with
DTExec on an existing SQL LSAM machine — a package stored in SQL Server (msdb), a .dtsx file
on the machine, or a package in the SSIS catalog. It brings SSIS work that already runs on an
LSAM into an OpCon Continuum workflow without rebuilding it.
The agent builds the DTExec command line from the fields below, which is why a few characters are refused in them.
MS SQL DTExec — parameters
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target SQL LSAM machine. Comes from the job's agent assignment, and must match the agent name registered with the relay. |
| Package Location | Yes | SQL Server (msdb) (default), File system, SSIS catalog | Where the package is stored. A new job starts with SQL Server (msdb) stored. It comes first, because it decides which of the fields below are shown. |
| Package Path | Yes | 1–255 chars | The package to run. In msdb, its package path — \Maintenance\Nightly. On the file system, the full path to the .dtsx file on the agent machine. In the SSIS catalog, \SSISDB\<folder>\<project>\<package>.dtsx. |
| Configuration File | No | ≤255 chars | File system only. The full path to a package configuration file (.dtsConfig) on the agent machine, such as C:\Packages\Nightly.dtsConfig, applied when the package runs. See The configuration file. |
| Server Name\Instance | For msdb and the SSIS catalog | ≤255 chars | The SQL Server instance that holds the package, such as HOST\INSTANCE or HOST,1433. Hidden for the file system. |
| Connection | No | a saved connection | The account the job runs as — a SQL Batch User connection. With SQL Server authentication it is the SQL login; with Windows Authentication or the SSIS catalog, the Windows account DTExec runs as. Blank runs DTExec as the agent's service account. Hidden for the file system — see below. |
| Windows Authentication | No | default off | On, DTExec connects to SQL Server as the batch user's Windows account rather than as a SQL login. The SSIS catalog always uses Windows authentication, whatever this says. Hidden for the file system. |
| Other Options | No | ≤255 chars | Additional DTExec switches, appended as typed — /Rep EW, for example. |
| Exit Criteria Result | No | Fail / Finish OK | What a match decides for the criteria table. Default Fail. |
| Exit Criteria | No | ≤20 conditions | Conditions tested against DTExec's return code, always evaluated by Continuum. See Exit criteria. |
Where the package is stored decides who runs it:
- On the file system, DTExec always runs as the agent's service account. Choosing File system hides Connection, Server Name\Instance and Windows Authentication and clears them, so a connection picked earlier is dropped from the job too. A job that still names a batch user — one written outside the editor — is refused at dispatch rather than quietly running as an account other than the one you chose.
- In
msdb, it runs with the batch user's SQL login, or with its Windows account when Windows Authentication is on. With no batch user, it runs as the agent's service account. - In the SSIS catalog, it always uses Windows authentication — the batch user's Windows account, or the agent's service account with none.
With a SQL Server login, DTExec receives the login's password on its command line on the agent machine. Windows Authentication avoids that. For the same reason, do not type a password into Other Options in clear text.
The return codes you write criteria against
This type returns DTExec's own return code:
| Code | Means |
|---|---|
0 | Success |
1 | Failure |
3 | Cancelled |
4 | Package not found |
5 | Package could not be loaded |
6 | Syntax or internal error |
With no criteria table, only 0 finishes the job OK and every other code fails it. A package
that cannot be found is an ordinary failure with code 4, not an initialization error: DTExec is
started, and it is DTExec that exits 4. The SQL agent cannot receive a criteria table at all, so
Continuum always evaluates this type's table centrally, as it does for MS SQL Script and MS SQL
Job.
The job's output is DTExec's own output. A /Rep switch in Other Options sets how much of
it DTExec reports.
The configuration file
Configuration File is offered only while Package Location is File system; choosing
another location hides it and clears it. The relay sends it to DTExec as /ConfigFile "<path>",
ahead of whatever Other Options holds, so you type the path only — the relay adds the quotes.
For a package in msdb or the SSIS catalog, add a configuration-file switch to Other Options
yourself, as before.
It needs a current relay. When the relay serving the machine has
not reported support for configuration files, the job is not sent to it: it goes to
Initialization Error with
RELAY_LSAM_FEATURE_UNSUPPORTED, rather than running the package without its configuration. To run
it on such a relay, clear Configuration File and put the path in Other Options as a
/ConfigFile switch.
What is refused at dispatch
The form will save these; the relay refuses them when the job runs, naming the field:
- a Package Path that is missing, contains a double quote, or ends with a backslash — the agent wraps the path in quotes itself, so either would break the command line;
- a Package Path with a character outside the Windows-1252 code page the agent connection uses,
which would otherwise arrive as
?and name a different package; - a missing Server Name for
msdbor the SSIS catalog, or one that contains a space or a double quote, because the agent passes it to DTExec unquoted; - a Package Location other than the three above;
- a batch user on a file-system package;
- a Configuration File on a package that is not on the file system; one over 255 characters,
containing a double quote, ending with a backslash, or holding a character outside the
Windows-1252 code page; one set while Other Options already carries a configuration-file switch
(any switch starting
/Confor-Conf, such as/ConfigFile) — use one or the other; or a configuration file and Other Options that together come to more than 255 characters.
Oracle
SQL LSAM — Oracle
The fourth SQL LSAM job type runs an Oracle SQL*Plus script on an existing SQL LSAM machine — a nightly GL extract or an end-of-day balancing script you already run against Oracle. SQL*Plus must be installed on the agent machine. The script is always a file on that machine: this type has no inline statements.
An Oracle job always signs on with database authentication: the batch user's login and password are the Oracle user name and password SQL*Plus connects with. There is no Windows Authentication setting on this type.
Oracle, Other DB and MySQL need a current relay. An older relay refuses their dispatch as an unsupported job type, so the job never starts — see Keeping a relay current.
Oracle — parameters
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target SQL LSAM machine. Comes from the job's agent assignment, and must match the agent name registered with the relay. |
| Connection | Yes | a saved connection | A SQL Batch User connection holding the Oracle user name and password. Required, and it must have a password: a job with no batch user, or one whose batch user has no password, is refused at dispatch. |
| Server Name | Yes | 1–255 chars | The host name or IP address of the Oracle database server, such as dbhost.example.com. |
| Port | No | 1–65535 | The Oracle listener port. Leave it empty to use the default, 1521. |
| Service Name | Yes | 1–255 chars | The Oracle service name of the database, such as ORCLPDB1. |
| Script File Path | Yes | 1–255 chars | The full path to the SQL*Plus script on the agent machine, such as C:\Scripts\nightly.sql. |
| Parameters | No | ≤255 chars | Values passed to the script as &1, &2 and so on, separated by spaces. Put a value that contains spaces in single quotes: 2026-10-14 'NORTH REGION'. |
| Redirect Result File Path | No | ≤255 chars | Writes the SQL*Plus output to this file on the agent machine instead of the job output. Overwritten on each run. |
| Other Options | No | ≤255 chars | Additional SQL*Plus command-line options, such as -S for silent mode. |
| Exit Criteria Result | No | Fail / Finish OK | What a match decides for the criteria table. Default Fail. |
| Exit Criteria | No | ≤20 conditions | Conditions tested against the SQL*Plus return code, always evaluated by Continuum. See Exit criteria. |
The return codes you write criteria against
| Code | Means |
|---|---|
0 | Success |
| The ORA error number | An Oracle error ended the script — 942 for ORA-00942, or 1017 for an invalid user name or password |
1 | The script file is missing, or another operating-system error |
With no criteria table, only 0 finishes the job OK and every other code fails it. The SQL agent
cannot receive a criteria table, so Continuum always evaluates this type's table centrally, as it
does for the other SQL types. Because the code is the ORA number itself, a row such as Equal To
1017 singles out a sign-on failure.
The script parameters and Other Options are redacted from the relay's connector log. The agent
still receives them; the log shows *** in their place.
What is refused at dispatch
The agent places these fields on a Windows command line to start SQL*Plus, so the relay refuses any character that the command line could not pass on as plain text. The form will save them; the job fails when it runs, before anything reaches the machine, with a message that names the field — or the batch user — and never the value:
- inline SQL statements on the job — left over, for example, from a job switched over from MS SQL Script. Move them into a script file;
- a missing Script File Path, Server Name or Service Name, a missing batch user, or a batch user with no password;
- a Port that is not a whole number from
1to65535; - a value longer than 255 characters, or one with a character outside the Windows-1252 code page the agent connection uses (Other Options is exempt from the code-page check);
- a tab, line break or other control character in any of these fields;
- in Server Name, and in the batch user's login and password:
&|<>^%()", a space,@or/; - in Service Name: the same characters, except that
/is allowed; - in Parameters:
&|<>^%()or"— spaces and single quotes are fine; - in Script File Path and Redirect Result File Path:
",%,<or>. Only%is legal in a Windows file name, so a real path is refused only if it contains one; - in Other Options:
&|<>^%, an odd number of", or the text-$.
A batch user refused for its login or password is fixed on the connection, not on the job. Because the connection is shared with the other SQL types — where all of these characters are allowed — the check is applied only when an Oracle job uses it.
Other DB
SQL LSAM — Other DB
The fifth SQL LSAM job type runs SQL through an ODBC driver or an OLE DB provider on an
existing SQL LSAM machine — the way to reach a database the other SQL types do not cover, such as a
reporting warehouse with its own ODBC driver. The driver or provider must be installed on the
agent machine. The agent runs the SQL itself over the driver, with no command line, so quotes,
<, & and % are ordinary characters in the SQL here. Like Oracle, it needs a
current relay.
Other DB — parameters
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target SQL LSAM machine. Comes from the job's agent assignment, and must match the agent name registered with the relay. |
| Connection | No | a saved connection | A SQL Batch User connection. With Windows Authentication on, the job connects as this Windows account. With it off, this is the database login and password, used for an ODBC connection string or a DSN only. Leave it empty to connect as the agent's service account. |
| Windows Authentication | No | default off | Connect as the batch user's Windows account. Works with drivers that accept Windows authentication, such as SQL Server's. |
| Connection Option | No | ODBC connection string (default), OLE DB connection string, ODBC data source (DSN) | How the agent connects. It decides which of the next two fields is shown. |
| DSN Name | For a DSN | ≤255 chars | The name of an ODBC system data source on the agent machine, such as PayrollDSN. Shown only with ODBC data source (DSN). |
| Connection String | For a connection string | ≤255 chars | The ODBC or OLE DB connection string, such as Driver={ODBC Driver 18 for SQL Server};Server=db01;Database=hr;. Do not include a login or password — put the login in a batch user. Shown only with the two connection-string options. |
| Script Source | No | SQL statements / Script file | Where the SQL comes from. A new job starts on SQL statements. It decides which of the next two fields is shown. |
| SQL Statements | One of the two | ≤4000 chars | The SQL to run, typed into the job. Shown unless Script Source is Script file. |
| Script File Path | One of the two | ≤255 chars | The path to a .sql file on the agent machine, read as UTF-8. Shown unless Script Source is SQL statements. |
| Substitution Variables | No | name/value pairs | Each $(NAME) in the SQL is replaced with its value before the SQL runs. Names are case-sensitive. These are not environment variables. |
| Redirect Result File Path | No | ≤255 chars | The file on the agent machine that receives the query results instead of the job output. Overwritten on each run. |
| Exit Criteria Result | No | Fail / Finish OK | What a match decides for the criteria table. Default Fail. |
| Exit Criteria | No | ≤20 conditions | Conditions tested against the job's return code, always evaluated by Continuum. See Exit criteria. |
Changing Connection Option or Script Source clears the field it hides, so a job never carries a DSN and a connection string, or statements and a script file, at once. Connection stays shown whichever option you choose. A job saved before Script Source existed has none set and shows both script fields, as it always did. Supply exactly one of SQL Statements and Script File Path: neither, or both, is refused at dispatch, and so is a Script Source that disagrees with the field that is filled in.
Which account the job connects as:
| Connection Option | Windows Authentication | Batch User | Connects as |
|---|---|---|---|
| ODBC string or DSN | off | set | The batch user's database login and password |
| Any | on | set | The batch user's Windows account |
| Any | either | empty | The agent's service account |
| OLE DB string | off | set | Refused at dispatch — the agent cannot pass a database login to an OLE DB provider. Turn on Windows Authentication or remove the batch user |
The agent waits for the SQL however long it runs. To stop one that runs too long, give the job a
Max Run Time and a $JOB:KILL command on an
Exceeded Max Runtime job-status event. Only the
first result set reaches the job output, so return only the rows you need.
The return codes you write criteria against
| Code | Means |
|---|---|
0 | Success |
-2146232009 | An ODBC error |
| The provider's error code | An OLE DB error — for example -2147217865 |
-2146233079 | The script file is missing |
The driver's own message is in the job output. A connection that fails to open fails the job with
no return code, so no criteria row can match it. With no criteria table only 0 finishes the job
OK; the table is always evaluated by Continuum, as for the other SQL types.
The connection string must carry no login or secret
The connection string is stored with the job in plain text, so it is checked at dispatch and refused
if it names a credential. The database login comes from the batch user instead. Values are read as
the mode quotes them — in ODBC mode a value in braces {…} counts as one value, in OLE DB mode a
value in double or single quotes does — and the job is refused for:
- a key that contains
password,passwd,secretortoken, or ends inpwd—PWD,ProxyPwd,priv_key_file_pwd. Case and spaces in the key do not matter; - a login key:
UID,User ID,UserorUserName. The job connects as the batch user, or the agent's service account when there is none; - any of those keys nested inside a value too —
Extended Properties="UID=…;PWD=…"is refused, because a provider can read the login from it; - an encrypted token anywhere in the string;
- a brace or quote that is never closed.
The message names the key as you wrote it — or, for a key found inside a quoted or braced value, says
only that — and never the value. Trusted_Connection, Integrated Security and DSN= are allowed.
A driver that takes a secret under some other key name is not caught. Never put a password or key in the connection string: keep it in the batch user. The connection string is also redacted from the relay's connector log, as are substitution-variable values — but the agent's own log on the machine shows the substitution values, so do not put a secret there either.
What else is refused at dispatch
The form will save these; the job fails when it runs, before anything reaches the machine, naming the field:
- a missing Connection String (for the two connection-string options) or DSN Name (for a DSN), or a Connection Option other than the three above;
- a DSN Name containing
[]{}(),;?*=!@or\, which an ODBC data source name cannot contain; - both SQL Statements and a Script File Path, or neither, or a Script Source that does not match the one that is set;
- a substitution variable whose name is empty or contains
=or a control character. A variable with an empty value is not refused — it is simply not sent; - a value longer than its limit, or with a character outside the Windows-1252 code page the agent connection uses. Tabs and line breaks are allowed in SQL Statements and in substitution values, and nowhere else;
- an OLE DB job with a batch user and Windows Authentication off;
- an ODBC or DSN job with Windows Authentication off whose batch user's password contains
;or begins with{, which the driver would read as part of the connection string. Change the credential.
MySQL
SQL LSAM — MySQL
The sixth SQL LSAM job type runs a MySQL script file with the mysql command-line client
on an existing SQL LSAM machine — a nightly load into a MySQL reporting database, or an end-of-day
reconciliation script you already run against MySQL. The mysql client must be installed on the
agent machine, and on its PATH. The script is always a file on that machine: this type has no
inline statements.
A MySQL job always signs on with MySQL authentication: the batch user's login and password are the MySQL user name and password the mysql client connects with. There is no Windows Authentication setting on this type. Like Oracle and Other DB, it needs a current relay.
MySQL — parameters
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target SQL LSAM machine. Comes from the job's agent assignment, and must match the agent name registered with the relay. |
| Connection | Yes | a saved connection | A SQL Batch User connection holding the MySQL user name and password. Required, and it must have a password: a job with no batch user, or one whose batch user has no password, is refused at dispatch. |
| Server Name | Yes | 1–255 chars | The host name or IP address of the MySQL server, such as dbhost.example.com. |
| Port | No | 1–65535 | The MySQL server port. Leave it empty to use the mysql client's default, 3306. |
| Database Name | Yes | 1–255 chars | The database the script runs against, such as sales. |
| Script File Path | Yes | 1–255 chars | The full path to the MySQL script on the agent machine, such as C:\Scripts\nightly.sql. A UNC path such as \\fileserver\share\nightly.sql is accepted. |
| User Variables | No | name/value pairs | MySQL user variables set before the script runs. See User variables. |
| Redirect Result File Path | No | ≤255 chars | Writes the mysql client's results to this file on the agent machine instead of the job output. Error messages still appear in the job output. Overwritten on each run. |
| Other Options | No | ≤255 chars | Additional mysql command-line options, such as --table. |
| Exit Criteria Result | No | Fail / Finish OK | What a match decides for the criteria table. Default Fail. |
| Exit Criteria | No | ≤20 conditions | Conditions tested against the mysql client's return code, always evaluated by Continuum. See Exit criteria. |
User variables
Each name/value pair becomes a MySQL SET @name = value; before the script runs, so the script reads
the value as @name. These are MySQL user variables, not environment variables.
- The value is a SQL expression, not plain text. Put text in single quotes —
'NORTH'— and write a number as it is —20261016. A bare word such asNORTHis not read as text. - Names use letters, digits,
_,.and$, up to 64 characters. - Values are up to 255 characters and cannot contain
();\&|<>^%#,--,/*or a double quote. Single quotes, and backticks, must come in pairs; a doubled''inside a quoted value is fine. So a function call such asNOW()cannot be a value. - A variable with an empty value is not sent.
- Setting any variable also restricts the Script File Path — see What is refused at dispatch.
User-variable values, like Other Options, are redacted from the relay's connector log, but the agent's own log on the machine shows them. Never put a password or key in a user variable.
The return codes you write criteria against
| Code | Means |
|---|---|
0 | Success |
1 | A MySQL error, a failed login or connection, or a missing script file. Also returned when the mysql client is not installed or not on the agent machine's PATH and no user variables are set |
255 | The mysql client is not installed or not on the agent machine's PATH, on a job with user variables set |
Every MySQL error returns 1, so exit criteria can tell success from failure but not one error
from another. The mysql client's own error message is in the job output. With no criteria table,
only 0 finishes the job OK. The SQL agent cannot receive a criteria table, so Continuum
always evaluates this type's table centrally, as it does for the other SQL types.
What is refused at dispatch
The agent places these fields on a Windows command line to start the mysql client, so the relay refuses any character that the command line could not pass on as plain text. The form will save them; the job fails when it runs, before anything reaches the machine, with a message that names the field — or the batch user — and never the value:
- inline SQL statements on the job — left over, for example, from a job switched over from MS SQL Script. Move them into a script file;
- a missing Script File Path, Server Name or Database Name, a missing batch user, or a batch user with no password;
- a Port that is not a whole number from
1to65535; - a value longer than 255 characters, or one with a character outside the Windows-1252 code page the agent connection uses (Other Options is exempt from the code-page check);
- a tab, line break or other control character in any of these fields;
- in Server Name, Database Name, and the batch user's login and password:
&|<>^%()"or a space.@,/,:and other punctuation are allowed, so a MySQL password can use them; - in Script File Path and Redirect Result File Path:
",%,<or>; - in Script File Path, when any user variable is set: also
&|^()or;, and any folder, file or server name in the path that starts with a lowercaseg—\\gfs\share\x.sqlorC:\scripts\gl.sql. Rename it, or leave the job without user variables; - in Other Options:
&|<>^%, an odd number of", or the text-$; - a user variable whose name or value breaks the rules in User variables.
A batch user refused for its login or password is fixed on the connection, not on the job. Because the connection is shared with the other SQL types — where all of these characters are allowed — the check is applied only when a MySQL job uses it.
The relay's checks keep a job's fields from breaking out of the Windows command line. They do not
limit what the mysql client itself does: the script, and the options in Other Options, can use
the client's system command to run operating-system commands on the agent machine, under the
agent's account. Control who can edit MySQL jobs and the scripts they name.
The agent also passes the batch user's password to the mysql client on its command line, where it can be seen on the agent machine while the job runs. Use a MySQL account that has only the rights the script needs.
SAP R/3
The SAP R/3 and CRM LSAM connector (opcon.lsam.sap-r3) has one job type.
| Job type | Agent type | Purpose |
|---|---|---|
SAP R/3 Job (sap-r3-job) | SAP R/3 and CRM | Start an existing SAP R/3 or CRM background job by name and number — see SAP R/3 Job |
SAP R/3 and CRM LSAM — SAP R/3 Job
The SAP R/3 Job type is new, and its run path has not yet been verified against a real SAP system. Confirm availability with your Continuous contact before planning work around it.
The SAP R/3 and CRM LSAM connector (opcon.lsam.sap-r3) provides one job type, SAP R/3 Job
(sap-r3-job). It starts an existing SAP R/3 or CRM background job — the job you already see in
transaction SM37, such as a nightly general-ledger posting run — on an SAP R/3 and CRM agent, by
its job name and job number. It does not define the SAP job: the job, its steps and its variants
stay in SAP.
The SAP agent decides whether the job succeeded. There is no exit-criteria setting and no batch user on this type. The agent signs on to SAP with the SAP user configured on the agent itself, and the job needs a current relay — one that reports it can run SAP R/3 jobs. An older relay refuses the job at dispatch, with RELAY_LSAM_PLATFORM_UNSUPPORTED naming SAP R/3 and CRM jobs.
SAP R/3 Job — parameters
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target SAP R/3 and CRM agent. Comes from the job's agent assignment, and must match the agent name registered with the relay. |
| SAP Job Name | Yes | 1–32 chars | The name of the existing SAP background job, as shown in SM37, such as ZFIN_DAILY. Wildcards (* and ?) are not allowed. |
| SAP Job Number | Yes | 1–8 digits | The job number of that SAP job, as shown in SM37, such as 12345601. A shorter number is padded with leading zeros to eight digits. |
| Start | No | A.S.A.P. / Immediately | A.S.A.P. (the default) starts the job as soon as a background work process is free. Immediately starts it now, and fails the job if no background work process is free. |
| Exec. Target | No | ≤64 chars | The SAP application server to run the job on, such as sapapp01_PRD_00. Leave it empty to let SAP choose. |
| Skip Spool Retrieval | No | check box | Turn it on to have the agent skip retrieving the job's spool output after it ends. Off by default. |
Ampersands, quotes and angle brackets in these fields reach the agent intact.
Looking up the job and the server in SAP
SAP Job Name and Exec. Target each have a Search… button beside them, so you can pick the value from the SAP system instead of typing it. The button is there only when the job's Agent Assignment is Specific Agent with a Primary Agent chosen, and only in an editor you can change: the job editor in the workflow editor, and Edit Job on a job in a running workflow. With a pool assignment, or no primary agent, the fields are plain text boxes.
Search… opens a panel under the field. Nothing is asked of SAP until you start the search there:
| Field | In the panel | The row you pick fills in |
|---|---|---|
| SAP Job Name | Job Name Pattern, pre-filled with the field's value (such as ZFIN_*), then Search. The results show Job Name, Job Number, Status and Class. | SAP Job Name and SAP Job Number, together |
| Exec. Target | Load Servers, with no pattern. The results show Server, Host and Free Background Processes (free of total). | Exec. Target |
Select a row and then Use Selected (or press Enter on it); the arrow keys move between rows.
- The search asks the job's primary agent. It goes through the agent's relay, and the agent signs on to SAP with its own SAP user and the agent's SAP system details. An agent that is not online can still be searched; the panel warns This agent is not online. The search may fail.
- It needs a tenant-wide permission: Agents Execute, under All workspaces. Without it, Search… is unavailable and says You don't have permission to run agent lookups. Ask an administrator. See SAP lookup permission.
- A pattern is 1 to 32 printable ASCII characters, and
*and?are wildcards. A pattern made only of*,?and spaces is refused: Enter at least one character other than *, ? and space. - Each Search is sent once. While it runs, the panel reads Searching… with the seconds counted, and Cancel stops it. A failed search is never sent again on its own. If the agent has not answered after 3 minutes, the panel reads The agent has not answered yet. Search again later.
- A search that completed is remembered while the editor is open. The same search again shows that result without asking the agent; Search Again asks the agent afresh.
- Closing the panel stops the search, with Close or Esc, and so do closing the editor and changing the primary agent.
- At most 5,000 rows are shown, with Showing the first 5000 jobs. Narrow the pattern to see the rest. (or Showing the first 5000 application servers.). With none, the panel reads No SAP jobs match this pattern. or The agent reported no application servers.
Some rows are shown but cannot be picked, each with its reason under the name: a job whose name
is longer than 32 characters or contains * or ?, a job whose number is not 1 to 8 digits, and a
server whose name is longer than 64 characters — the values the job would be
refused at dispatch for. A job in a status other than Scheduled or
Ready can be picked, but reads Only Scheduled or Ready jobs can be started., because the job
would fail with 70006 as it stands. Status shows Scheduled, Released, Ready,
Released (suspended), Active, Finished, Cancelled, or Unknown for any other.
A search that fails says why in the panel — see the SAP R/3 troubleshooting rows (on the Troubleshooting tab).
What the job reports
When the job ends, the agent reports a line of text that starts with a result code, and that text is the job's termination description — the Termination row on the job's Summary tab, and the value an Exit Description event compares. The leading code is the job's exit code:
| The agent reports | Exit code | Means |
|---|---|---|
0 - and the SAP job number, such as 0 - 12345601 | 0 | The SAP job finished |
1 - and the SAP job number | 1 | The SAP job did not finish successfully: SAP reports it cancelled, or it no longer exists |
An agent error code, such as 70006: followed by the agent's text | The error code | The agent could not run the job — see the table below |
The agent's own error codes:
| Code | Means |
|---|---|
70002 | The job number is missing |
70003 | The job name is missing |
70004 | No SAP user is configured on the agent |
70005 | The SAP logon failed |
70006 | The job was not found, or is not in Scheduled or Ready status |
70007 | The job could not be copied |
70009 | The job could not be started as soon as possible |
70017 | The job could not be started immediately — for example, no background work process was free |
If the agent's text carries no leading code, the exit code is recorded as 0 and the text is kept
as the termination description. The job's outcome is still the agent's.
A failed job's Agent Message explains its exit code. When the job fails, its
Agent Message row reads what the exit code means, in plain
words — for exit code 1, The SAP job did not finish successfully: SAP reports it cancelled, or
it no longer exists., and for 70006, The SAP job was not found, or is not in Scheduled or Ready
status. When the agent's own text says more than the bare code, it follows the explanation after
a |. Every code the agent declares has an explanation, including the ones not listed above.
Termination, the exit code and the job's outcome are not changed by it.
Its output
Once the job has ended, the job's Output tab lists the output files the agent kept for that run. The request for the list is completed from what the job reports when it finishes, so while the job is still running the tab says JORS LIST failed: This SAP R/3 job's output can't be listed: OpCon has no record of where the agent filed it., followed by why: the job may still be running, it may have run before the relay was upgraded, or its exit description held no job count.
The output is found on the day the agent filed it. The agent files a job's output under a day folder named from its own clock. Continuum asks for that day by the agent's own time zone, which the agent reports in its status replies, and if the answer comes back empty it asks once more for the day before and then the day after. A job that ends late in the evening in the agent's time zone still shows its output.
A job whose SAP Job Name contains a comma cannot have its output listed. The agent splits the request for the list on commas, so it would list other jobs' output as this one's; the request is refused instead. The job itself runs normally.
What is refused at dispatch
The form saves these; the job fails when it runs, before anything reaches the agent, with a message that names the field and never the value:
- a missing SAP Job Name or SAP Job Number;
- a SAP Job Name over 32 characters, or one containing
*or?; - a SAP Job Number that is not 1 to 8 digits;
- an Exec. Target over 64 characters;
- in SAP Job Name or Exec. Target, a character outside the Windows-1252 code page the agent connection uses.
Restarting it on a step
The SAP agent reports the job's steps as it runs, each one a value that starts with STEP and a
three-digit number and goes on to name the step's program, such as STEP002:ZPROG2:Failed. A failed SAP R/3 Job can be restarted on one of them, so SAP runs the job from
that step: see Restart on a step. While it runs,
its child jobs appear in the job's Child Processes table, and as Child Process
agent feedback (on the How It Works tab), where an
LSAM Feedback event can watch them.
Restarting only the failed child processes
An SAP R/3 job can start child jobs in SAP. When one of them fails, you can restart only the failed children: the agent runs those child jobs again and does not run the parent job, or the children that already finished, a second time.
The Child Processes table. On the job's page, the Summary tab shows a Child Processes table whenever the agent has reported any children, directly above Agent Feedback. Each row is one child job, in the order the agent reported it:
| Column | Shows |
|---|---|
| Name | The SAP job name of the child |
| Job Number | Its SAP job number |
| Status | The child's status as the agent last reported it, such as Finished, Active, Terminated or Does Not Exist. A dash means the agent did not name a status |
Choosing the children to restart. A child can be selected only when its status is Terminated or Does Not Exist and the job itself can be restarted (see Which actions a job status accepts). The check box of any other row is disabled and says why when you point at it:
| The row | What it says |
|---|---|
| A child that has not failed | Only failed child processes can be selected |
| A failed child, while the job cannot be restarted | Restart is not available for this job's current status |
| A child already marked Restarting | Already marked for restart; it runs with the next restart. It is shown ticked and is always part of the restart |
Select all failed ticks every child you can select. Restart selected (N) opens the Restart dialog for the children you chose, plus any already marked: it reads Restart N child processes. Only the selected child processes run; the parent job does not run again. Any restart step chosen earlier is cleared. and lists their names (the first 10, then and N more). Restart on Hold and Reset Retry Count work as usual. Confirming marks those children Restarting and restarts the job in one action; the agent then runs only them. The selection is cleared whenever the job's feedback changes.
The table is on the job's page only. A run opened from Job Execution History shows its children as Child Process values in its Agent Feedback section, without a selection.
A restart of the whole job after a children restart. While children are marked Restarting, the job's next restart runs only them, whichever way it is started. The Restart dialog says so: its step control is disabled and reads The marked child processes (N), with Only marked child processes run again.
What refuses a children restart. Nothing about the job changes when it is refused:
| Message | When |
|---|---|
| Only failed child processes, or ones already marked for restart, can be restarted. | A chosen child has not failed |
| Some selected child processes are no longer on this job. Reload and try again. | A chosen child is no longer among the job's children |
| The child processes changed while you were restarting. Reload and try again. | The agent reported a change to a chosen child while the restart was being made |
| Restart either from a step or with selected child processes, not both. | A request named both a restart step and children |
| Restarting selected child processes is available only for SAP R/3 jobs. | The job is not an SAP R/3 Job |
A children restart needs the Job Instances Execute permission, like every restart.
A kill is refused by default. An SAP R/3 and CRM agent's Allow Kill setting is off unless you turn it on, so Kill on one of these jobs is refused and the job keeps running. See Allow Kill.
Checking every stored SAP job number
An SAP R/3 Job names its SAP job by name and number, and fails with 70006 when SAP no longer
has a Scheduled or Ready job under that number. Check SAP Job Numbers finds those jobs
across your workflows before they run: it compares the number stored in each SAP R/3 Job with the
SAP system, and saves the numbers you choose to correct.
Where. In Workflows, the workflow toolbar's Check SAP Job Numbers action. It is among the workflow actions when a workflow is open — under More actions (⋮) when the toolbar is narrow — and under More actions when none is open. Back to Workflows returns to the workflow editor.
What is listed. Every SAP R/3 Job in the current version of each workflow in the workspaces where you hold Workflows View, by workflow name and then by the job's place in the workflow. Up to 5,000 jobs are listed; past that the page says Only the first 5,000 SAP R/3 jobs are listed. Listing asks nothing of SAP.
| Column | Shows |
|---|---|
| Workflow | The workflow, as a link that opens it in a new browser tab |
| Job | The job's name in the workflow |
| SAP Job Name | The stored SAP Job Name |
| Stored Number | The stored SAP Job Number, or (empty) |
| Agent | The job's primary agent |
| Status | The verdict below, or the state of a fix |
| New Number | The number SAP has, for a Stale job |
| Details | Why the job has its verdict, the deployments a fix will not reach, and the result of a fix |
Start Check checks the listed jobs. Before you start, the page says Each check asks the job's primary agent for the SAP jobs with that name, and the agent confirms the jobs it finds in SAP. Jobs that share a primary agent and an SAP job name are checked with one question to that agent. At most two questions are out at once, and only one per agent; Checked N of M names counts them. Stop starts no more of them — the ones already asked finish — and Start Check carries on with the rest. Each question waits up to 3 minutes for its answer. The Status filter shows the jobs with one verdict, and the line beside it counts every verdict.
| Status | Means |
|---|---|
| Not Checked | Not checked yet |
| In Sync | SAP has exactly one Scheduled or Ready job with this name, and its number is the stored one. Numbers are compared as the job sends them, padded with leading zeros to eight digits |
| Stale | SAP has exactly one Scheduled or Ready job with this name, under another number — or the stored number is empty or not 1 to 8 digits. New Number shows SAP's number, and the job can be fixed |
| Ambiguous | SAP has two or more Scheduled or Ready jobs with this name, listed in Details. Pick the right one in the job editor |
| Not Found | SAP has no job with this name for the agent's SAP user. The stored number is kept |
| Not Startable | SAP has jobs with this name, but none is Scheduled or Ready; Details lists their statuses |
| Error | The check could not be completed — Details says why. This includes an agent that returned too many jobs (The agent returned too many jobs to check this name.) and a number from SAP that the job could not be sent with |
| Cannot Check | The job is not checked, for the reason in Details (below) |
SAP job names are compared without regard to case or surrounding spaces.
What is not checked. These jobs are Cannot Check at once, with nothing asked of SAP:
| Details | When |
|---|---|
| No primary agent: only jobs that run on a specific agent can be checked. | The job's Agent Assignment is a pool, or it has no primary agent |
| The job's agent and its machine name differ, so the check can't tell which SAP system runs it. Open the job and re-select its agent. | The machine name stored in the job is not exactly its primary agent's name, so the check would ask a different SAP system from the one the job runs on |
| Uses a property: the SAP job name or number comes from a property, so it can't be checked here. | The SAP Job Name or SAP Job Number holds a property token |
| Invalid SAP job name: it is empty, longer than 32 characters, or contains * or ?. | A name the job would be refused at dispatch for |
| This SAP job name can't be searched: use printable ASCII characters only. | The name holds a character outside printable ASCII |
| The primary agent no longer exists. or The primary agent can't answer SAP queries. | Found when the agent is asked |
Who can do what. Listing needs Workflows View. Start Check needs Agents Execute granted under All workspaces; without it the page says Checking SAP job numbers needs permission to run agent queries in all workspaces. Ask an administrator. and Start Check is unavailable. Fixing a job needs Workflows Edit in its workflow's workspace; a Stale job you cannot fix reads You can't edit this workflow.
Fixing the Stale jobs. Select the check box of each Stale job to fix, or Select All Stale, then Fix Selected. Save SAP Job Number Fixes lists every change by workflow — Job, SAP Job Name, Old Number and New Number — with that workflow's warnings, and reads Each workflow below is saved as a new version in which only these SAP job numbers change. Nothing is saved until Save Fixes. Then, one workflow at a time:
- The workflow is read again. If its version has moved on since the list was read, or a chosen job's SAP job name or number has changed, nothing is saved and the jobs read Not Fixed: Changed Since Check.
- Otherwise it is saved once, as a new version by you in which only those SAP Job Number values
change. The version's change description reads SAP job number re-sync: followed by each job's
name and its old and new number, such as
SAP job number re-sync: GL Post 12345601 → 12345688(shortened to and N more past 500 characters). The jobs read Fixed, with Fixed in version N. - A save that is refused reads Not Fixed with the reason; one that got no answer reads The save may or may not have completed — reopen the workflow to check. Nothing is retried.
The list is not read again after a save, so the verdicts stay; Refresh List reads it again and clears every verdict. The results exist only on this page: leaving it loses them, and while a check or a save is running the browser asks before you leave.
What a fix does not reach
A fix is a new version of the workflow, so it reaches what runs that version:
- A Pinned deployment on an older version keeps running that version. Details and the dialog say Pinned in the environment at version N — the fix will not reach it until the deployment is updated. See Deployments.
- A transform rule in an environment that names the agent assignment, the machine name, the SAP job name or the SAP job number may change them there: A transformation rule in the environment may change the agent or job number there. See Transform rules.
- Schedules already built keep their own copy of each job: Schedules already built keep their own copy of each job. Rebuild them, or edit the job in each run, to pick up a fix.
Only the deployments you can view are warned about. A warning does not prevent the fix.
SAP BW
The SAP BW LSAM connector (opcon.lsam.sap-bw) has one job type.
| Job type | Agent type | Purpose |
|---|---|---|
SAP BW Process Chain (sap-bw-process-chain) | SAP BW | Start an SAP BW process chain by its technical name — see SAP BW Process Chain |
SAP BW LSAM — SAP BW Process Chain
The SAP BW Process Chain type is new, and its run path has not yet been verified against a real SAP BW system. Confirm availability with your Continuous contact before planning work around it.
The SAP BW LSAM connector (opcon.lsam.sap-bw) provides one job type, SAP BW Process Chain
(sap-bw-process-chain). It starts an existing SAP BW process chain — the chain you see in
transaction RSPC, such as a nightly data load — on an SAP BW agent, by the chain's technical
name. It does not define the chain: the chain and its processes stay in SAP.
The SAP BW agent decides whether the chain succeeded. There is no exit-criteria setting and no batch user on this type: the agent signs on to SAP with the logon configured on the agent itself. The job needs a current relay — one that reports it can run SAP BW jobs. An older relay refuses the job at dispatch, with RELAY_LSAM_PLATFORM_UNSUPPORTED naming SAP BW jobs.
SAP BW Process Chain — parameters
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target SAP BW agent. Comes from the job's agent assignment, and must match the agent name registered with the relay. |
| Process Chain Name | Yes | 1–25 chars when the job runs | The technical name of the process chain, as shown in RSPC, such as ZPC_DAILY_LOAD. The agent upper-cases it. A property token can supply it; the value it resolves to is what is checked. |
Ampersands, quotes and angle brackets in the name reach the agent intact.
Looking up the process chain in SAP
Process Chain Name has a Search… button beside it, on the same terms as an SAP R/3 Job's
lookups (on the SAP R/3 tab): it is there only with a Specific Agent
assignment and a Primary Agent chosen, asks that agent through its relay only when you search,
and needs Agents Execute granted under All workspaces. Enter a Process Chain Pattern of
up to 30 characters, such as ZPC_*, and Search. The results show Process Chain and
Description, and the row you pick fills in Process Chain Name. A chain whose name is longer
than 25 characters is shown but cannot be picked — This process chain name is longer than 25
characters, so it cannot be run. With no match, the panel reads No process chains match this
pattern.; past 5,000 chains, Showing the first 5000 process chains. Narrow the pattern to see the
rest.
What the job reports
When the chain ends, the agent reports a line of text that starts with a result code, and that text is the job's termination description — the Termination row on the job's Summary tab, and the value an Exit Description event compares. The leading code is the job's exit code:
| The agent reports | Exit code | Means |
|---|---|---|
0 - and the chain's log ID | 0 | The process chain finished |
1 - and the chain's log ID, or 1 - NOT FOUND | 1 | The process chain did not finish successfully, or SAP did not find it |
An agent error code, such as 70001 | The error code | The agent could not run or follow the chain — see the table below |
The agent's own error codes:
| Code | Means |
|---|---|
70001 | The agent could not start or restart the process chain |
70002 | The agent could not read the log of the process chain |
70003 | The agent could not interrupt the process chain |
70004 | The agent could not get the status of the process chain |
70005 | The agent could not restart the process chain |
70006 | The agent could not get the processes of the process chain |
70007 | The agent could not read the log of a process in the chain |
70008 | The SAP system did not respond in time |
These codes are the SAP BW agent's own, and several mean something different from the same number on an SAP R/3 Job.
A failed job's Agent Message explains its exit code. When the job fails, its
Agent Message row reads the meaning above — for exit code
1, The process chain did not finish successfully. — followed after a | by the agent's own text
when that says more than the bare code. Termination, the exit code and the job's outcome are not
changed by it.
While the chain runs, the agent's status text appears as the job's Job Status Description in its Agent Feedback (on the How It Works tab), with the Process values the agent reports and, once the chain has started, its log ID as Process Chain Log. An LSAM Feedback event can watch any of them.
Its output
Once the job has ended, the job's Output tab lists the process-chain log the agent kept for that run. The request for the list is completed from the chain name and the log ID in the job's termination description, so while the job is still running the tab says JORS LIST failed: This SAP BW job's output can't be listed: OpCon has no record of where the agent filed it., followed by why. The same is shown when the termination description held no log ID. As on an SAP R/3 Job, the output is asked for on the day the agent filed it, by the agent's own time zone, with one more try on the day before and the day after.
What is refused at dispatch
The form saves these; the job fails when it runs, before anything reaches the agent, with a message that names the field and never the value:
- a missing or blank Process Chain Name — Missing required field: processChainName in taskDefinition (an SAP BW job needs the process chain name);
- a Process Chain Name over 25 characters once any property token in it is resolved — SAP BW field processChainName must be at most 25 characters, got …. The form accepts a longer value, so this is caught only when the job runs;
- in Process Chain Name, a character outside the Windows-1252 code page the agent connection uses, which would otherwise reach the agent as a different chain name.
A job aimed at an agent that is not an SAP BW agent is refused at dispatch too, with LSAM_AGENT_TYPE_MISMATCH — see A legacy job aimed at an agent of another type.
A name that SAP does not know is not refused here: the job runs, and the agent reports
1 - NOT FOUND or 70001.
Restarting it on a step
When the chain starts, the agent reports the chain's log ID as the job's one step. A failed SAP BW Process Chain job can be restarted on it, so the agent asks SAP to restart that chain run rather than start a new one: see Restart on a step. Restart on a step has not yet been verified against a real SAP BW agent; after one, check the chain's run in RSPC as well as the job's Termination.
A kill is always refused. Jobs on an SAP BW agent can never be killed, so Kill on one is refused with Cannot kill: jobs on SAP BW agents cannot be killed. Let the chain end, or stop it in SAP. See Allow Kill.
z/OS
The z/OS LSAM connector (opcon.lsam.zos) has six job types.
| Job type | Agent type | Purpose |
|---|---|---|
z/OS Batch Job (zos-batch) | z/OS | Submit a JCL member to JES on a z/OS LSAM agent — see z/OS Batch Job |
z/OS Started Task (zos-started-task) | z/OS | Start a procedure, such as a CICS or IMS region, as a started task — see z/OS Started Task |
z/OS REXX Exec (zos-rexx) | z/OS | Run a REXX exec in its own address space — see z/OS REXX Exec |
z/OS Console Command (zos-command) | z/OS | Issue a z/OS operator command — see z/OS Console Command |
z/OS Tracked Job (zos-tracked) | z/OS | Put a job submitted on the mainframe outside OpCon on its schedule when the agent reports it — see z/OS Tracked Job and z/OS Queued Job |
z/OS Queued Job (zos-queued) | z/OS | Hold a job the agent intercepted until its dependencies are met — see z/OS Tracked Job and z/OS Queued Job |
The JCL member a z/OS Batch Job submits, and the exec a z/OS REXX Exec runs, can be opened, edited and saved back to the mainframe from Continuum — see Viewing and editing JCL.
z/OS LSAM — z/OS Batch Job
The z/OS Batch Job type is new, and its run path has not yet been verified against a real z/OS LSAM. Confirm availability with your Continuous contact before planning work around it.
z/OS Batch Job (zos-batch) has the z/OS LSAM submit a JCL member through JES — a nightly core posting job, for
example — from the JCL library on the DD you name, with the JCL substitutions you give it. The member
and its JCL stay on the mainframe; the job names them.
The z/OS agent decides whether the job succeeded, from the condition codes the job's steps end with: the job fails when any step ends with a condition code outside the job's Minimum Condition Code to Maximum Condition Code range, or abends. Step Control rows can change that for a step, and Dataset Pre-run rows hold the job until the datasets it depends on are ready. There is no exit-criteria table and no Connection field on this type: User ID names the z/OS user the job is submitted as. The job needs a current relay — one that reports it can run z/OS jobs. An older relay refuses the job at dispatch, with RELAY_LSAM_PLATFORM_UNSUPPORTED naming z/OS jobs.
z/OS Batch Job — parameters
The fields appear in this order under Job Parameters.
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target z/OS agent. Comes from the job's agent assignment, and must match the agent name registered with the relay. A z/OS agent's name is saved in upper case, at most 8 characters — see z/OS agent names. |
| Member | No | a z/OS name, when the job runs | The JCL member to submit, such as IVPJOB13. Leave it blank to use the start of the job's name — see The member a blank Member submits. |
| DD Name | No | a z/OS name, when the job runs | The DD on the LSAM task that holds the JCL library. Leave it blank to use the agent's own default, usually XPSJCL; Continuum never sends a default of its own. Enter $DUMMY to submit no JCL at all — see A file-watcher job. |
| User ID | No | a z/OS name, when the job runs | The z/OS user the job is submitted as. Leave it blank to submit it under the agent's own authority. |
| Minimum Condition Code | No | 0–4095 | The lowest condition code a step may end with and still succeed. Default 0. |
| Maximum Condition Code | No | 0–4095 | The highest condition code a step may end with and still succeed. Default 0, and it must not be lower than the minimum. |
| JCL Substitutions | No | ≤99 lines, ≤3400 chars in total | Symbolic and keyword substitutions for the JCL, such as \\&HLQ=PROD.DATA\\ or @KEY=VALUE, separated by double backslashes as the agent expects. A multi-line box: use one line per group for readability. Blank lines are dropped, and line breaks do not count toward the total. |
| Temporary Member | No | a z/OS name, when the job runs | A temporary member to submit instead of Member, from the override DD. |
| Override DD Name | No | a z/OS name, when the job runs | The DD that holds the temporary member. |
| Pre-run System | No | a z/OS name, when the job runs | The sysplex system that evaluates the job's pre-run conditions. Leave it blank for the agent's own system. |
| Submit From | No | Primary LSAM only / Any system in the sysplex | Which LSAM in the sysplex may submit the job. Default Primary LSAM only. |
| Step Control | No | ≤80 rows | Rules the agent applies when a step ends — see Step Control. |
| Dataset Pre-run | No | ≤80 rows | Dataset conditions the agent must see before it submits the job — see Dataset Pre-run. |
| Message Pre-run | No | ≤80 rows | Console messages the agent must see before it releases the job — see Message, Job/Task, Tape and REXX Pre-run. |
| Job/Task Pre-run | No | ≤80 rows | Jobs, started tasks or TSO users that must be running, or must not be — see Message, Job/Task, Tape and REXX Pre-run. |
| Tape Pre-run | No | one group | Hold the job until enough tape units are free — see Message, Job/Task, Tape and REXX Pre-run. |
| REXX Pre-run | No | one group | A REXX procedure whose return code of 0 releases the job — see Message, Job/Task, Tape and REXX Pre-run. |
| Restart | No | three options, each LSAM default when blank | How the agent's restart processing treats datasets that already exist and generation data groups — see Restart options. |
A z/OS name is 1 to 8 characters — letters, digits, @, # or $ — and does not start with a
digit. Lowercase letters are upper-cased before the name is checked and sent, so ivpjob13 reaches
the agent as IVPJOB13. The form takes a longer value so that a property token fits; the value the
token resolves to is what is checked.
A blank condition code is sent as 0, so both condition codes always reach the agent. Property
tokens such as [[JI.NAME]] in JCL Substitutions, or in any other field, are resolved before
the job is sent.
Values reach the agent exactly as you typed them, as on UNIX: ampersands, angle brackets and
backslashes in JCL Substitutions arrive as written, which is what JCL symbolics need. For the
same reason </F> and <F I= cannot be carried in any value and are refused at dispatch — see
UNIX values reach the agent exactly as you typed them (on the
UNIX/Linux tab, under Command).
The member a blank Member submits
With Member blank, the relay takes the member from the job's name: it upper-cases the name, keeps
its first eight characters, and stops at the first space, underscore or dot. A job named
Payroll_Nightly submits PAYROLL, and one named GLPOST.US submits GLPOST.
Every character it takes must be a letter, a digit, @, # or $, and the result must not start
with a digit. When the name gives no such member — 2026 posting, or Pay-roll — the job fails at
dispatch with z/OS field memberName is blank and the job name does not give a valid z/OS member
name; set Member. Set Member on the job.
Step Control
Step Control is a list of rules the agent applies when a step ends. Each row names a step, a condition-code range and an action. Add, beside Step Control, adds a row, and the rows can be moved up and down: the agent checks them in order and uses the first row that matches, so list them by priority. Add is unavailable once the list holds 80 rows.
| Column | Required | Limits | Notes |
|---|---|---|---|
| Step Name | Yes | ≤17 chars | The step, as STEP or STEP.PROCSTEP, such as STEP1.PROCSTEP. Each part is 1 to 8 characters — letters, digits, @, #, $, % or ? — and does not start with a digit. % or ? matches any one character, and a * at the end of a part matches the rest of it. A step name alone matches every step of the procedure it calls. Upper-cased before it is sent. |
| Minimum Condition Code | No | 0–4095 | The lowest condition code this row matches. Default 0. |
| Maximum Condition Code | No | 0–4095 | The highest condition code this row matches. Default 0, and it must not be lower than the row's minimum. |
| Action | Yes | one of six | What the agent does when the step ends with a condition code in this range. No default. |
| Message | Depends on Action | ≤20 chars | Required for Send trigger message. For Set restart step, the step to restart from. A leading space is not kept. |
The six actions, as the Action list offers them:
- Abend job at step termination
- Send job completed
- Set restart step
- Send trigger message
- Post error but allow job to continue
- OK to continue
Two rows with the same step name and the same condition-code range are refused, as is a row whose Message holds a character outside the Windows-1252 code page the agent connection uses.
Dataset Pre-run
Dataset Pre-run is a list of dataset conditions the z/OS agent must see before it submits the job — wait for the nightly transmission file to be cataloged before the posting job runs, for example. Every row must be satisfied before the job runs. Add, beside Dataset Pre-run, adds a row, and Add is unavailable once the list holds 80 rows. A collapsed row shows its dataset name.
| Column | Required | Limits | Notes |
|---|---|---|---|
| Dataset Name | Yes | ≤44 chars, dots included | Qualifiers of 1 to 8 characters — letters, digits, @, # or $ — separated by dots, none starting with a digit, such as PROD.BANK.DRAFT. % matches any one character, and a * at the end of a qualifier matches the rest of it. A name ending .G0000V00 matches any generation of a GDG. Upper-cased before it is checked and sent. |
| Generations | No | 1–99 | How many times the condition must happen before the job is released. Default 1. It is a count, not a GDG generation. |
| Condition | Yes | one of seven | What must happen to the dataset. No default. |
| When | Yes | one of two | Whether the trigger is kept between runs. No default. |
| Job Name | No | a z/OS name, when the job runs | The job that must cause the condition. Leave it blank for any job. |
The seven conditions, as the Condition list offers them: Exists, Created, Updated, Deleted, Referenced, Cataloged and Uncataloged. Exists checks the catalog once; the others wait for the event to happen.
The two choices of When:
- Continuous Monitoring keeps the trigger between runs, so a later run may start at once.
- While/As Scheduled Only removes the trigger once it fires, so the next run waits for a new event.
Exists needs a fully qualified name. On an Exists row, the dataset name must hold no % or
* and must not end in .G0000V00, and Generations must be 1. Two rows with the same
dataset name, generations, condition and job name are refused, whatever their When.
While the agent waits, the job says so. A job held by its pre-runs reads Pre-run Active while the agent reports that it is evaluating them, and moves on to Running when the JCL is submitted. When the agent instead reports that the pre-runs are not met, the job is re-attempted after a delay, with its definition sent again — see A prerun that fails is re-attempted (on the How It Works tab, under Shared Parameters).
The rows need a current relay. A job with Dataset Pre-run rows is refused at dispatch by a relay that has not reported it can send them — any relay older than this release — so it never submits its JCL without waiting: it goes to Initialization Error with RELAY_LSAM_FEATURE_UNSUPPORTED: relay '…' (agent pool '…') has not reported support for … Upgrade the relay. The same holds for Step Control rows and the Restart options. Upgrade the relay before you rely on them — see Keeping a relay current.
A file-watcher job
With DD Name set to $DUMMY, the agent submits no JCL: the job completes as soon as its
dataset pre-runs are met. That makes it a file watcher — a job a workflow can wait on until a
dataset arrives, without running anything on the mainframe.
Message, Job/Task, Tape and REXX Pre-run
Four more kinds of pre-run follow Dataset Pre-run on the form, in this order. Each holds the job on the agent until its condition is met, and every pre-run the job has — dataset rows included — must be satisfied before it runs. Each is optional: a job with none of them reaches the agent exactly as before.
Message Pre-run waits for console messages: hold the posting job until the transmission job
writes IEF404I PAYJOB, for example. Add adds a row, and is unavailable at 80 rows. A
collapsed row shows its message key.
| Column | Required | Limits | Notes |
|---|---|---|---|
| Message Key | Yes | ≤44 chars, when the job runs | The text the console message must contain. Case-sensitive, and sent as typed. A part in braces, such as {JOBNAME}, is a variable key; put a - before the braces to exclude messages that contain it. |
| Generations | No | 1–99 | How many times the message must appear before the job is released. Default 1. |
| When | Yes | one of two | Continuous Monitoring or While/As Scheduled Only, as for a dataset row. No default. |
| Job Name | No | a z/OS name, when the job runs | The job that must issue the message. Leave it blank for any job. |
| Offset | No | 0–120 | Where the key starts in the message text, counted from 0. Default 0. |
| Length | No | 1–44 | The length of the key's fixed part, before any part in braces. Default 1. |
Two message rows with the same key, generations, job name, offset and length are refused, whatever their When. Keys that differ only in case are different keys.
Job/Task Pre-run holds the job until a job, started task or TSO user is running, or is not — wait until the CICS region is down before a file reorganization, for example. Add adds a row, up to 80. A collapsed row shows the name.
| Column | Required | Limits | Notes |
|---|---|---|---|
| Job/Task Name | Yes | a z/OS name, when the job runs | The job, started task or TSO user, such as CICSPROD. |
| Condition | Yes | Executing / Not Executing | Whether it must be running or must not be. No default. |
Two identical job/task rows are allowed.
Tape Pre-run holds the job until enough tape units of a device are free. Leave Device blank for no tape pre-run.
| Field | Required | Limits | Notes |
|---|---|---|---|
| Device | No | ≤8 chars, when the job runs | A generic or esoteric unit name, such as 3490: letters, digits, @, #, $, / or -. Upper-cased before it is checked and sent. |
| Units | No | 0–99 | How many units of the device must be free. Blank is 0. Units other than 0 need a Device. |
REXX Pre-run runs a REXX procedure before the job, and only a return code of 0 releases it. Leave Procedure Name blank for no REXX pre-run.
| Field | Required | Limits | Notes |
|---|---|---|---|
| Procedure Name | No | a z/OS name, when the job runs | The REXX procedure to run. |
| DD Name | No | a z/OS name, when the job runs | The DD the agent reads the procedure from. Leave it blank for the agent's default, SYSEXEC; nothing is sent. Needs a Procedure Name. |
| Parameters | No | ≤4000 chars, when the job runs | Passed to the procedure as typed, case kept. On one line: a line break or tab is refused. Needs a Procedure Name. |
While the agent waits, the job says so, as for dataset rows: it reads Pre-run Active while the agent reports that it is evaluating its pre-runs, and is re-attempted after a delay, with the same pre-runs sent again, when the agent reports that they are not met — see A prerun that fails is re-attempted (on the How It Works tab, under Shared Parameters).
These pre-runs need a current relay too. A job with one is refused at dispatch, with Initialization Error, by a relay that has not reported it can send them, rather than running at once without waiting. Upgrade the relay before you rely on them — see Keeping a relay current.
Restart options
Restart, the last group under Job Parameters, holds three options that tell the agent how its restart processing treats a dataset that already exists, and the job's generation data groups (GDGs). Each is optional: left blank it reads LSAM default, nothing is sent, and the agent's own default applies. Clear an option to return it to the default.
| Option | Choices | What it sets |
|---|---|---|
| Duplicate Dataset Action — Normal | None, Scratch, Reuse | What the agent does with a dataset that already exists when the job runs normally. |
| Duplicate Dataset Action — Restart | None, Scratch, Reuse | What the agent does with a dataset that already exists when the job is restarted. |
| GDG Regression | None, Absolute, Relative, Catalog Resync | How the agent resets generation data groups when the job is restarted. Absolute is not compatible with GDGBIAS=STEP. |
The options go to the agent with every run of the job, a normal run and a restart alike, and the agent decides which applies: the Normal action on a normal run, the other two on a restart. A z/OS Batch Job has them, and so do a z/OS Tracked Job and a z/OS Queued Job, where they are the only fields. To restart a job on a step, see Restarting it.
What the job reports
When the job ends, the text the agent reports is the job's termination description — the Termination row on the job's Summary tab, and the value an Exit Description event compares — on success and on failure alike, exactly as the agent sent it. Continuum does not read a numeric exit code out of it, and does not decide the outcome from it: the outcome is the agent's.
Because the description is text, an Exit Description event against it behaves as it does on IBM i: Equal To with the text matches it exactly, and Contains finds part of it — see Match a message ID exactly (on the IBM i tab, under Batch Jobs).
While the job runs, the agent's status text appears as its Job Status Description, and what the agent reports as User Message, Step Completion and Trigger Messages appears beside it, in the job's Agent Feedback (on the How It Works tab). An LSAM Feedback event can watch any of them. While the agent reports that it is running the job's prerun, the job reads Pre-run Active — see A prerun that fails is re-attempted (on the How It Works tab, under Shared Parameters).
Its output
The job's Output tab lists the job's
SYSOUT as the z/OS agent reports it, such as JESMSGLG, JESJCL and JESYSMSG, in the order the
agent sent them. It is shown as it is laid out on the mainframe: each row of the list keeps its
spacing and any trailing *, and a file's contents do not wrap — a wide line scrolls sideways
instead. A file you download keeps its spacing too.
The tab can list the output only when the z/OS agent itself reports where it filed it, and Continuum has no other way to find a z/OS job's output. So:
- for a run whose agent did not report it, the tab says JORS LIST failed: This z/OS job's output can't be listed: its agent did not report where the job's output is for this run… — which is also what a job that ran before your relay was upgraded shows;
- while the job is still running and the agent has not reported it yet, the tab says This z/OS job is still running and its agent has not yet reported where its output is…. The agent reports it at the latest when the job ends, so try again later;
- when the agent has no JORS Port set, the tab says JORS port not configured for agent: …, naming the agent. A z/OS agent starts with JORS Port empty, so set it to the port the agent's JORS listener uses — see Agents and agent pools.
What is refused at dispatch
The form saves these; the job fails when it runs, before anything reaches the agent, with a message that names the field and never the value, such as z/OS field ddName must be 1-8 characters (A-Z, 0-9, @, # or $) and must not start with a digit:
- a Member, DD Name, User ID, Temporary Member, Override DD Name or Pre-run System that is not a z/OS name once any property token in it is resolved;
- a blank Member on a job whose name gives no valid member — see The member a blank Member submits;
- a condition code that is not a whole number from
0to4095, or a Minimum Condition Code greater than its Maximum Condition Code, on the job or on a Step Control row; - JCL Substitutions of more than 99 lines or more than 3400 characters once resolved, or holding a character outside the Windows-1252 code page the agent connection uses;
- more than 80 Step Control rows, a row with no Step Name or Action, a step name that breaks the rules above, a Message over 20 characters, a Send trigger message row with no Message, or two rows with the same step name and condition-code range;
- more than 80 Dataset Pre-run rows, a row with no Dataset Name, Condition or When,
a dataset name that breaks the rules above, Generations outside
1–99, a Job Name that is not a z/OS name, an Exists row with a wildcard, a.G0000V00ending or Generations other than1, or two rows with the same dataset name, generations, condition and job name; - more than 80 Message Pre-run rows; a row with no Message Key or When, a key over 44
characters or holding a character outside the Windows-1252 code page, Generations outside
1–99, an Offset outside0–120, a Length outside1–44, or a Job Name that is not a z/OS name; or two rows with the same key, generations, job name, offset and length, such as z/OS field messagePrerun[4] repeats row 2 (same key, generations, job name, offset and length); - more than 80 Job/Task Pre-run rows, or a row with no Job/Task Name or Condition, or a name that is not a z/OS name;
- a tape Device over 8 characters or with another character, Units outside
0–99, or units with no device, z/OS field tapePrerun.units needs a device; - a REXX Procedure Name or DD Name that is not a z/OS name, Parameters over 4000 characters or holding a line break, a tab or another character outside Windows-1252, or a DD Name or Parameters with no procedure name, such as z/OS field rexxPrerun.ddName needs a procedure name;
- a Restart option that is not one of its choices — possible only in a job edited on the Code tab or written through the API;
</F>or<F I=in any value.
A job aimed at an agent that is not a z/OS agent is refused at dispatch too, with LSAM_AGENT_TYPE_MISMATCH — see A legacy job aimed at an agent of another type.
Restarting it
A z/OS Batch Job or z/OS Started Task can be restarted on a step, or on a range of steps, so the agent re-runs only those steps instead of the whole job. While the job runs, the z/OS agent reports its steps. The Restart dialog then offers two choices in place of the single step list other platforms show:
- Start Step — From the beginning, selected by default, or one of the reported steps;
- End Step — To the end of the job, selected by default, or a reported step at or after the start step. It is unavailable until a start step is chosen, and choosing another start step puts it back to To the end of the job.
The agent's restart processing then applies the job's Restart options. See Restart on a step for what the dialog shows when the job has reported no steps, and how long the step stays with the job.
A $JOB:RESTART or $JOB:RESTARTHLD event names the range in its Restart Step as
start|end, such as STEP2|STEP4. STEP2 alone, or STEP2|, runs from STEP2 to the end of the
job. Each step must be one the job reported, and the end step must not come before the start step;
otherwise the restart is refused, and nothing about the job changes. Restart on a step has not yet
been verified against a real z/OS agent; after one, check the job's steps on the mainframe as well as
its Termination.
A z/OS REXX Exec or z/OS Console Command restarts from the beginning only: an exec is a single step, and a command has none. So does a z/OS Tracked Job or z/OS Queued Job. A restart that names a step is refused with Cannot restart on a step: restart on step is not available for this kind of z/OS job. A Set restart step row in Step Control is the agent's own rule and is not affected by any of this.
A relay that predates z/OS restart on a step refuses a z/OS job restarted on a step: the job does not run, and it is never run from the beginning in its place. Upgrade the relay — see Keeping a relay current.
A kill reaches the agent by default, on every z/OS job type. A z/OS agent's Allow Kill setting is on unless you turn it off — see Allow Kill.
z/OS LSAM — z/OS Started Task, z/OS REXX Exec and z/OS Console Command
The z/OS Started Task, z/OS REXX Exec and z/OS Console Command types are new, and their run paths have not yet been verified against a real z/OS LSAM. Confirm availability with your Continuous contact before planning work around them.
Three more job types run on the same z/OS agents as the z/OS Batch Job. Each has its own fields, and all three share the z/OS Batch Job's Pre-run System, Submit From, Dataset Pre-run, Message Pre-run, Job/Task Pre-run, Tape Pre-run and REXX Pre-run fields, which work exactly as they do there — see Dataset Pre-run and Message, Job/Task, Tape and REXX Pre-run (under z/OS LSAM — z/OS Batch Job). Like it, none has an exit-criteria table or a Batch User: the z/OS agent decides whether the job succeeded. What the job reports when it ends, the Agent Feedback it shows while it runs, its output and its kill are as for a z/OS Batch Job — see What the job reports and Its output. A relay that predates these types refuses them at dispatch with Unsupported z/OS task type for TX1 translation, naming the job type, and nothing reaches the agent.
z/OS Started Task
A z/OS Started Task starts a procedure from the PROCLIB as a started task — a CICS or IMS region, for example. The job stays Running until the task ends, for as long as the region is up; Continuum does not end it on a timer of its own. The agent decides whether it succeeded as it does for a z/OS Batch Job: the job fails when any step ends with a condition code outside its Minimum Condition Code to Maximum Condition Code range, or abends, and Step Control rows can change that for a step — see Step Control. It can be restarted on a step — see Restarting it.
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target z/OS agent, as for a z/OS Batch Job. |
| Started Task Name | No | a z/OS name, when the job runs | The procedure (PROCLIB member) to start, such as CICSPROD. Leave it blank to use the start of the job's name, by the rule a blank Member follows — see The member a blank Member submits. |
| Execution Parameters | No | ≤256 chars, when the job runs | Parameters passed to the started task after its name, such as SIT=6$,START=AUTO. |
| Minimum Condition Code | No | 0–4095 | The lowest condition code a step may end with and still succeed. Default 0. |
| Maximum Condition Code | No | 0–4095 | The highest condition code a step may end with and still succeed. Default 0, and it must not be lower than the minimum. |
| Pre-run System, Submit From | No | as for a z/OS Batch Job | |
| Step Control | No | ≤80 rows | As for a z/OS Batch Job — see Step Control. |
| Dataset Pre-run | No | ≤80 rows | Dataset conditions the agent must see before it starts the task — see Dataset Pre-run. |
| Message Pre-run, Job/Task Pre-run, Tape Pre-run, REXX Pre-run | No | as for a z/OS Batch Job | See Message, Job/Task, Tape and REXX Pre-run. |
z/OS REXX Exec
A z/OS REXX Exec runs a REXX exec in its own address space. The agent checks the exec's return code against the job's Minimum Return Code to Maximum Return Code range: a return code outside it fails the job. There is no Step Control, because an exec is a single step.
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target z/OS agent, as for a z/OS Batch Job. |
| Exec Name | No | a z/OS name, when the job runs | The REXX exec to run, such as RXDAILY. Leave it blank to use the start of the job's name, by the rule a blank Member follows — see The member a blank Member submits. |
| DD Name | No | a z/OS name, when the job runs | The DD on the LSAM task that holds the REXX library. Leave it blank to use the agent's own default, usually SYSEXEC; Continuum never sends a default of its own. |
| Execution Parameters | No | ≤256 chars, when the job runs | The argument string passed to the exec, such as ARG1 ARG2. |
| Minimum Return Code | No | 0–4095 | The lowest return code the exec may end with and still succeed. Default 0. |
| Maximum Return Code | No | 0–4095 | The highest return code the exec may end with and still succeed. Default 0, and it must not be lower than the minimum. |
| Pre-run System, Submit From | No | as for a z/OS Batch Job | |
| Dataset Pre-run | No | ≤80 rows | Dataset conditions the agent must see before it runs the exec — see Dataset Pre-run. |
| Message Pre-run, Job/Task Pre-run, Tape Pre-run, REXX Pre-run | No | as for a z/OS Batch Job | See Message, Job/Task, Tape and REXX Pre-run. |
z/OS Console Command
A z/OS Console Command issues a z/OS operator command, such as D A,L. A command has no
completion code: the agent issues it as it is and reports the job finished once it has, and that is
the job's outcome. There are no condition-code fields and no Step Control. A job that is
refused before the agent issues the command — at dispatch, say — fails as any other job does.
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target z/OS agent, as for a z/OS Batch Job. |
| Operator Command | Yes | ≤256 chars, when the job runs | The operator command to issue, such as D A,L. |
| Host Job Name | No | a z/OS name, when the job runs | The name of the started task that issues the command, such as CMDHOST. Leave it blank to use the start of the job's name, by the rule a blank Member follows — see The member a blank Member submits. |
| Pre-run System, Submit From | No | as for a z/OS Batch Job | |
| Dataset Pre-run | No | ≤80 rows | Dataset conditions the agent must see before it issues the command — see Dataset Pre-run. |
| Message Pre-run, Job/Task Pre-run, Tape Pre-run, REXX Pre-run | No | as for a z/OS Batch Job | See Message, Job/Task, Tape and REXX Pre-run. |
What all three share
The parameter text reaches the agent exactly as you typed it. Execution Parameters and
Operator Command keep their case, and characters such as &, ', <, >, , and = arrive
as written. Property tokens such as [[JI.NAME]] are resolved first, and leading and trailing
spaces are dropped. The form takes a longer value so that a token fits; the resolved value must be
at most 256 characters, and a longer one is refused rather than cut. Left blank, Execution
Parameters is not sent at all. The names — Started Task Name, Exec Name, DD Name,
Host Job Name and Pre-run System — follow the z/OS name rule and are upper-cased, as on a
z/OS Batch Job.
What is refused at dispatch. The form saves these; the job fails when it runs, before anything reaches the agent, with a message that names the field and never the value:
- a name that is not a z/OS name once any property token in it is resolved;
- a blank name on a job whose own name gives none, such as z/OS field startedTaskName is blank and the job name does not give a valid z/OS name; set Started Task Name (or execName and Exec Name, hostJobName and Host Job Name);
- a blank Operator Command once resolved, with z/OS field operatorCommand is required;
- Execution Parameters or an Operator Command of more than 256 characters once resolved, or holding a character outside the Windows-1252 code page the agent connection uses;
- on a Started Task or REXX Exec, a condition or return code outside
0–4095, or a minimum greater than its maximum; - a Step Control row, or a pre-run of any kind, that breaks the rules a z/OS Batch Job's would — see What is refused at dispatch;
- a field another z/OS job type uses, such as z/OS field stepControl is not used by a zos-rexx job; remove it. Changing a job's type in the job editor clears its parameters, so this happens only to a job edited on the Code tab or written through the API. A z/OS Batch Job refuses another type's fields the same way;
</F>or<F I=in any value.
z/OS LSAM — z/OS Tracked Job and z/OS Queued Job
The z/OS Tracked Job and z/OS Queued Job types are new, and their run paths have not yet been verified against a real z/OS LSAM. Confirm availability with your Continuous contact before planning work around them.
z/OS Tracked Job and z/OS Queued Job
These two types bring a job that the z/OS agent itself reports under the control of its schedule. Neither holds any JCL in Continuum: there is no Member, DD Name or JCL field, and no View / edit JCL. The job is defined on the workflow so that the agent's report has a job to land on.
- z/OS Tracked Job (
zos-tracked) is for a job submitted on the mainframe outside OpCon. When the agent sees it start and reports it, the job is added to its workflow instance already started: it does not wait for its dependencies, and it finishes with the condition code the agent reports. - z/OS Queued Job (
zos-queued) is for a job the agent intercepts and holds. When the agent reports the held job, the job is added to its workflow instance and released to the agent once its dependencies are met, then finishes with the condition code the agent reports.
Either way, the successors of the job can depend on it, and its outcome is on the schedule like any other job's. What the job reports when it ends, and its output and kill, are as for a z/OS Batch Job — see What the job reports and Its output.
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | A z/OS agent. When the agent reports the job, the job is assigned to the agent that reported it, whatever its own assignment says. |
| Duplicate Dataset Action — Normal, Duplicate Dataset Action — Restart, GDG Regression | No | as for a z/OS Batch Job | The job's only other fields — see Restart options. |
A Tracked or Queued job runs only from the agent's report. Started any other way — built with its workflow and released before the agent has reported it, say — it fails without anything being sent to the agent, with A zos-tracked job runs only after its agent reports it (a tracked or queued job event). This run carries no tracking data from the agent, so nothing was sent to the agent (RC_LSAM_PRIVATE_DATA_REQUIRED) (or zos-queued).
What happens when the agent reports one
The agent reports with one of three events that only an agent can raise — Track Job
($JOB:TRACK), Queue Job ($JOB:QUEUED) and Update Queued Job ($JOB:QUESTATE). Each names
the schedule date, the workflow and the job; the line, and why one is refused, is in
Tracked and queued jobs. The job is matched by its
name, ignoring case, among the jobs of the workflow as deployed, and the workflow instance must
already be built for that date.
| The agent reports | What Continuum does |
|---|---|
| A job it saw start (Track Job) | Adds the job to the workflow instance as Wait to Start (Forced), assigned to the reporting agent, with the agent's description of the job as its termination description. The job is sent to the agent without waiting for its dependencies, conflicts or resources, even when the agent is Marked Draining or has no free slot. While the agent is Offline or Marked Offline, the job waits in Wait to Start (Forced) and goes as soon as the agent can take work. |
| A job it is holding (Queue Job) | Adds the job to the workflow instance to wait for its dependencies as any job does. It is never added On Hold, even when the job is set to build on hold. The agent is told at once that the job was accepted. When the dependencies are met, the job is released to the agent, with the agent's ordinary checks. |
| An update to a job it queued (Update Queued Job) | With a status of I, updates the job's termination description only. With any other status, starts the job now, as a tracked job starts — unless it has already been sent to the agent, when the update is refused and nothing changes. |
When the job is already in the workflow instance, what a report does depends on the job that is there:
| The job already there | Track Job | Queue Job, multi-instance job | Queue Job, otherwise |
|---|---|---|---|
| Finished | Replaced | Replaced | Replaced |
| Not started, and not from an agent's report | Added as a copy | Replaced | Replaced |
| Not started, from an earlier report | Added as a copy | Added as a copy | Refused: the agent is told Duplicate Job |
| Running | Added as a copy | Added as a copy | Refused: the agent is told Duplicate Job |
A copy is named for the job with a number, as TRACKJOB$0001; a finished copy's number is reused
before a new one is taken.
These need a relay that carries the agent's reference. The job is sent to the agent with the reference the agent gave it, so the agent can match the two. A job for an agent pool whose relay has not reported that it can carry it is refused with RELAY_LSAM_FEATURE_UNSUPPORTED, and a relay that predates the two types refuses them with Unsupported z/OS task type for TX1 translation. Upgrade the relay — see Keeping a relay current.
z/OS LSAM — JCL access
Viewing and editing JCL
The JCL member a z/OS Batch Job submits, and the exec a z/OS REXX Exec runs, can be opened, edited and saved back to the mainframe from Continuum — to check what a job will submit, or to fix a dataset name or a symbolic before a rerun, without leaving for ISPF. View / edit JCL opens the JCL dialog from three places:
| Where | It is offered when |
|---|---|
| The job editor's Job Definition tab, with the job's parameters | The job's Agent Assignment is a Specific Agent with a Primary Agent of type z/OS |
| JCL, in the header of a daily job's detail card | The job's machine is one z/OS agent, whatever the job's status |
| View / edit JCL, in the ⋮ menu of the job's node on the workflow diagram | As for the detail card. While the job and its agent are being checked it shows disabled, with Checking JCL access… |
It is not offered for a job assigned to an agent pool or a legacy agent group, or for a z/OS Started Task, z/OS Console Command, z/OS Tracked Job or z/OS Queued Job.
What the dialog opens. Member, View DD and Save DD are filled in for you, and you can change any of them. Each is upper-cased as you type and must be a z/OS name.
| Field | z/OS Batch Job | z/OS REXX Exec |
|---|---|---|
| Member | Temporary Member, else Member, else the start of the job's name | Exec Name, else the start of the job's name |
| View DD — where Load reads from | DD Name, else XPSJCL | DD Name, else SYSEXEC |
| Save DD — where Save writes to | Override DD Name, else TEMPJCL | DD Name, else SYSEXEC |
In the job editor the defaults come from the job's parameters as they stand in the editor, saved or
not. A value that still holds a property token is left empty for you to fill in, and a DD Name
of $DUMMY fills in no View DD. On a daily job whose agent has reported where its output is,
Member and View DD come from what the agent reported for the run instead; Save DD keeps
the default above, so a z/OS Batch Job's save goes to TEMPJCL unless you change it.
Load reads the member into a fixed-width editor, with a column ruler, a guide at column 80 and a Line N, Col M indicator, and says Loaded N lines from XPSJCL(PAYROLL). Leading blanks are kept. Loading over edits you have not saved asks first: Loading replaces the text in the editor. Your unsaved edits will be lost. A member that does not exist gives Member PAYROLL was not found in XPSJCL., and one the agent will not read gives The agent refused to read the member: …. While the dialog waits for the mainframe it shows Contacting the mainframe… and its buttons are unavailable.
Save writes the editor's text to the Save DD under Member, as your z/OS user ID: the dialog says Saving as ZUSER01. Before anything is sent the text is checked, and the first problem is shown with its line number:
- a line longer than 80 characters — it is not cut to fit;
- a tab, or any other control character;
- a character the mainframe code page (Windows-1252) cannot store, or one that could not be read
from the member and shows as
�; - a line that starts with
%%, or that holds%%ETin any case, including one that runs across the end of a line into the next — the agent reserves it; - more than 2,000 lines, or no text at all.
Trailing blanks are dropped from each line and every line is sent as an 80-column card, so a member you load after a save matches what you saved, apart from trailing blanks. When Save DD is the same as View DD, the dialog warns first: You are saving over the library jobs read from.
| The dialog says | What happened |
|---|---|
| The agent's reply, such as Saved: TEMPJCL(PAYROLL), or Saved. | The member was saved. The reply is shown only when it has a known shape. |
| The save was refused: … | The agent refused the save, such as Edit Error - User Not Authorized. A reason the agent is not known to give is replaced by its message ID alone. |
| The save was not applied. Nothing was written to the member. | The save never reached the agent: it was not picked up within about 90 seconds, or the agent could not be reached. |
| The save may or may not have been applied. Load the member again to check. | The text was sent and no answer Continuum can trust came back, or none came within four minutes. Load the member to see what it holds. |
A save is never retried for you, and it cannot be cancelled: closing the dialog while it runs warns that it still runs on the mainframe and that the dialog will not show its result. Closing the dialog otherwise discards your edits. Continuum keeps a record of every save it sends — who saved, as which z/OS user ID, the agent, DD, member, line count and outcome — and never the JCL text. Once a save has its outcome, it is recorded in Audit History as an Update in the z/OS JCL category, again without the JCL text.
Who can do what.
- Load needs Agents Execute, granted under All workspaces — see The SAP lookups and JCL access are checked, and the grant must be tenant-wide. The agent reads the member under its own authority.
- Save needs z/OS JCL Edit — see z/OS JCL Edit — and a z/OS user ID assigned to you on the z/OS User IDs tab of the Permissions page — see z/OS User IDs. Without the permission the dialog offers no Save; with it but with no ID, Save is unavailable with You have no z/OS user id assigned. Ask an administrator. A save never falls back to the agent's own authority or to the job's User ID.
What JCL access needs. The agent must have a JORS Port set, or the dialog says This agent has no JORS port configured; JCL access needs one. — see Agents and agent pools. An agent with Use TLS for Scheduling on gives JCL cannot be viewed or saved for this agent because it uses TLS for scheduling.: JCL access is not available for it. The relay serving the agent must be current and online: an older one gives The relay serving this agent is too old for JCL access. Upgrade the relay to view or save JCL., and an offline one The relay serving this agent is offline. Try again when it is back online. When the mainframe takes too long to answer a Load, the dialog says The mainframe did not answer in time. Try again.
MCP
The MCP LSAM connector (opcon.lsam.mcp) has six job types.
| Job type | Agent type | Purpose |
|---|---|---|
MCP Run (mcp-run) | MCP | Run a program on a Unisys MCP LSAM agent — see MCP Run and MCP Start |
MCP Start (mcp-start) | MCP | Start a WFL job on a Unisys MCP LSAM agent, optionally keeping a job summary — see MCP Run and MCP Start |
MCP EAE/AB Suite (mcp-eae) | MCP | Run an EAE/AB Suite report on a Unisys MCP LSAM agent — see MCP EAE/AB Suite |
MCP Copy (mcp-copy) | MCP | Copy a file or a whole directory between disk packs and tapes on a Unisys MCP LSAM agent, optionally to another MCP host — see MCP Copy, MCP Change and MCP Remove |
MCP Change (mcp-change) | MCP | Rename a file or a whole directory on a Unisys MCP LSAM agent — see MCP Copy, MCP Change and MCP Remove |
MCP Remove (mcp-remove) | MCP | Remove a file or a whole directory on a Unisys MCP LSAM agent — see MCP Copy, MCP Change and MCP Remove |
MCP LSAM — the six MCP job types
The MCP job types are new, and their run path has not yet been verified against a real MCP LSAM. Confirm availability with your Continuous contact before planning work around them.
The MCP LSAM connector (opcon.lsam.mcp) provides six job types for a Unisys ClearPath MCP
system. MCP Run (mcp-run) runs a program, such as (PRODUSER)OBJECT/PAYCALC ON DISK.
MCP Start (mcp-start) starts a WFL job, such as WFL/PAYROLL/DAILY ON PACK, and can keep a job
summary for it. MCP EAE/AB Suite (mcp-eae) runs an EAE/AB Suite report, such as PAYROLLRPT,
with the parameters you give it. MCP Copy (mcp-copy), MCP Change (mcp-change) and MCP
Remove (mcp-remove) copy, rename or remove a file or a whole directory, such as
(PRODUSER)PAYROLL/= — see MCP Copy, MCP Change and MCP Remove. All six
run on an MCP agent, under an MCP usercode. The program, WFL job, report or file stays on the MCP
system; the job names it. Any of the six can have a prerun — a second MCP job the agent runs
first — see Prerun.
The MCP agent decides whether the job succeeded. There is no exit-criteria table: the Fail Code words you enter are matched by the agent against the MCP console display. Each job runs under a Connection — an MCP Batch User connection, which holds the usercode — and it is required. The job needs a current relay — one that reports it can run MCP jobs. An older relay refuses the job at dispatch, with RELAY_LSAM_PLATFORM_UNSUPPORTED naming MCP jobs.
An MCP agent's name must end with a period and be at most 24 characters, such as MCPPROD. — see
Agents and agent pools.
MCP Run and MCP Start — parameters
The fields appear in this order under Job Parameters. The two types share all of them except File Attributes, which only MCP Run has, and the three Job Summary fields, which only MCP Start has. MCP EAE/AB Suite has its own fields in place of File Title and Arguments — see MCP EAE/AB Suite.
| Parameter | Type | Required | Limits | Notes |
|---|---|---|---|---|
| Machine (from Agent Assignment) | Both | Yes | 1–24 chars | Target MCP agent. Comes from the job's agent assignment, and must match the agent name registered with the relay. |
| Connection | Both | Yes | — | The MCP usercode the job runs under: pick an MCP Batch User connection. |
| Access Code | Both | No | ≤17 chars, when the job runs | The usercode's access code, if it has one: upper-case letters, digits and spaces. It is part of the job, not of the batch user. |
| File Title | Both | Yes | ≤96 chars, when the job runs | The program (MCP Run) or the WFL job (MCP Start) as an MCP file title. No lower-case letters, and none of , ! @ # $ % ^ & | \ { } : ; ? < > ~ + `. |
| Arguments | Both | No | ≤200 chars, when the job runs | Parameters passed to the program or WFL job, such as ("STRING PARAM",5,TRUE). Quotes are sent as typed. |
| Task Attributes | Both | No | ≤10 rows, ≤300 chars each | Task attributes, one per row, such as PRIORITY=50. |
| File Attributes | MCP Run | No | ≤10 rows, ≤300 chars each | File attribute equations, one per row, such as FILE INFILE(TITLE=PAY/IN ON DISK). None of ! @ # $ % ^ | \ { } : ? < > ~ + `. |
| End-of-Task Notice | Both | No | on / off | Asks the agent for an end-of-task notice message when the job ends. Default off. |
| Template Display File | Both | No | ≤100 chars, when the job runs | The display template file the agent uses for this job. None of , ! @ # $ % ^ & | \ { } : ; ? < > ~ = + ' `. |
| Job Summary User | MCP Start | Depends | — | The usercode the job summary is kept under: a second MCP Batch User connection. Required when Job Summary Family or Job Summary Directory or File is set. |
| Job Summary Family | MCP Start | No | ≤20 chars, no | | The family (pack) the job summary is written to. |
| Job Summary Directory or File | MCP Start | Depends | ≤200 chars, no | | The directory or file name for the job summary, such as JOBSUMMARY/PAYROLL. Required when Job Summary Family is set. |
| Fail Code 1–3 | Both | No | ≤40 chars each, when the job runs | The first word of an MCP console display that means the job failed, such as DS*. End it with * to match any ending. No lower-case letters. |
| Fail Reset 1–2 | Both | No | ≤40 chars each, when the job runs | The first word of an MCP console display that cancels a fail-code match. End it with * to match any ending. No lower-case letters. |
Values are checked as they will be sent. A line break, tab or form feed in any value becomes a
space, the value is trimmed, and property tokens such as [[JI.NAME]] are resolved — and the value
that results is what is checked. A value over its limit is refused, never cut short. The form takes
longer values than these limits so that a token fits.
Values reach the agent exactly as you typed them, as on UNIX: quotes in Arguments arrive as
quotes, which is what WFL string parameters need. For the same reason </F> and <F I= cannot be
carried in any value and are refused at dispatch — see
UNIX values reach the agent exactly as you typed them (on the
UNIX/Linux tab, under Command).
A blank Fail Code or Fail Reset is simply not sent; the others keep their own number, so a Fail Code 3 with Fail Code 2 blank still reaches the agent as the third fail code.
MCP EAE/AB Suite
An MCP EAE/AB Suite job runs an EAE/AB Suite report. It takes the same Connection, Access Code, Task Attributes, File Attributes, End-of-Task Notice, Template Display File, Fail Code and Fail Reset fields as MCP Run, with the same rules, and three fields of its own in place of File Title and Arguments. The fields appear in this order:
| Parameter | Required | Limits | Notes |
|---|---|---|---|
| Machine (from Agent Assignment) | Yes | 1–24 chars | Target MCP agent, as for MCP Run. |
| Connection | Yes | — | An MCP Batch User connection: the usercode the report runs under. |
| Access Code | No | ≤17 chars, when the job runs | As for MCP Run. |
| Report Name | Yes | ≤256 chars, when the job runs | The EAE/AB Suite report to run, such as PAYROLLRPT. |
| Accept File | Yes | ≤256 chars, when the job runs | The file the agent creates to hold the report's parameters, such as PAYROLLACC. |
| Parameters | Yes | ≤256 chars, when the job runs | Every parameter the report needs, such as PAYROLL 20261024 ALL. The agent writes them into the Accept File. Sent as typed. |
| Task Attributes | No | ≤10 rows, ≤300 chars each | As for MCP Run. |
| File Attributes | No | ≤10 rows, ≤300 chars each | As for MCP Run. |
| End-of-Task Notice | No | on / off | As for MCP Run. |
| Template Display File | No | ≤100 chars, when the job runs | As for MCP Run. |
| Fail Code 1–3, Fail Reset 1–2 | No | ≤40 chars each, when the job runs | As for MCP Run. |
Report Name, Accept File and Parameters have no case or character rule of their own: lower-case letters and punctuation are sent as typed. Like every MCP value, they are checked once their property tokens are resolved, and a value over 256 characters is refused, never cut short.
The agent keeps the Accept File. It stays on the MCP system afterwards as a permanent file in
*SMA/LINC17/FILES/=. To clear that directory, schedule SMA/WFL/CLEANUP/LINC17/FILES.
MCP Copy, MCP Change and MCP Remove
MCP Copy, MCP Change and MCP Remove work on a file, such as (PRODUSER)PAYROLL/IN, or
on a whole directory, such as (PRODUSER)PAYROLL/=. Each takes the same Connection, Access Code,
Fail Code and Fail Reset fields as MCP Run, with the same rules, and fields of its own.
They have no Task Attributes, File Attributes, End-of-Task Notice or Template Display
File. The fields appear in this order:
| Parameter | Type | Required | Limits | Notes |
|---|---|---|---|---|
| Machine (from Agent Assignment) | All three | Yes | 1–24 chars | Target MCP agent, as for MCP Run. |
| Connection | All three | Yes | — | An MCP Batch User connection: the usercode the job runs under. |
| Access Code | All three | No | ≤17 chars, when the job runs | As for MCP Run. |
| Source File or Directory | All three | Yes | ≤256 chars, when the job runs | The file or directory to copy, rename or remove, as an MCP file title, such as (PRODUSER)PAYROLL/IN. |
| From | MCP Copy | Yes | ≤40 chars, when the job runs | The disk family or tape the source is on, such as DISK. |
| Source Kind | MCP Copy | No | Tape / Pack (disk) | Whether the source is on a tape or a disk pack. The form shows Tape, and a job that leaves it unset is sent as Tape. |
| Destination File or Directory | MCP Copy | Yes | ≤256 chars, when the job runs | The file or directory to copy to, as an MCP file title, such as (PRODUSER)BACKUP/PAYROLL/=. |
| To | MCP Copy | Yes | ≤40 chars, when the job runs | The disk family or tape to copy to, such as PAYTAPE. |
| Destination Kind | MCP Copy | No | Tape / Pack (disk) | Whether the destination is a tape or a disk pack. Sent as Tape when unset, as for Source Kind. |
| BNA Hostname | MCP Copy | No | ≤256 chars, when the job runs | The MCP host to copy to over a BNA network, such as MCPHOST2. Leave it blank for a copy on this host. |
| Copy & Compare | MCP Copy | No | on / off | Compares the copy with the source after copying. Default off. |
| New File or Directory | MCP Change | Yes | ≤256 chars, when the job runs | The new name, as an MCP file title, such as (PRODUSER)PAYROLL/OLD. |
| From | MCP Change, MCP Remove | Yes | ≤40 chars, when the job runs | The disk family or tape the file or directory is on, such as DISK. |
| Fail Code 1–3, Fail Reset 1–2 | All three | No | ≤40 chars each, when the job runs | As for MCP Run. |
The file names, From, To and BNA Hostname have no case or character rule of their own: they are sent as typed. Like every MCP value, they are checked once their property tokens are resolved, and a value over its limit is refused, never cut short. BNA Hostname can be set with either Kind.
MCP Remove fails when there is nothing to remove. By default the agent fails the job when no file or directory matches Source File or Directory.
These three need a relay that knows them. A relay that runs MCP jobs but predates these three types refuses them at dispatch with Unsupported MCP task type … (RC_MCP_UNSUPPORTED_TASK_TYPE) — see Keeping a relay current.
Prerun
Every MCP job type has a Prerun section at the end of Job Parameters. A prerun is a second MCP job, of any of the six types, that the agent runs before the job itself — a program that checks the day's input is complete, say, or a copy that brings a file in. Prerun Type chooses it: Run, Start, EAE/AB Suite, Copy, Change or Remove, or blank for no prerun. The prerun's type does not have to match the job's: an MCP Run job can have a Copy prerun.
A prerun that fails does not fail the job. The job moves to Pre-run Failed and is started again, prerun first, about three minutes later, until the prerun succeeds — see A prerun that fails is re-attempted (on the How It Works tab, under Shared Parameters). An MCP Remove prerun fails when there is nothing to remove, so a prerun that removes a trigger file holds the job until the file is there.
The section shows every prerun field, each titled with the prerun types that use it, such as Prerun File Title (Run, Start). Only the fields of the chosen Prerun Type are sent; a field that type does not use is ignored, not refused. Each field takes the same values, limits and rules as the job's own field of the same name — see the tables above:
| Prerun Type | Its fields |
|---|---|
| Run | Prerun File Title, Prerun Arguments, Prerun Task Attributes, Prerun File Attributes, Prerun End-of-Task Notice, Prerun Template Display File |
| Start | Prerun File Title, Prerun Arguments, Prerun Task Attributes, Prerun End-of-Task Notice, Prerun Template Display File, and the three Prerun Job Summary fields — User, Family and Directory or File |
| EAE/AB Suite | Prerun Report Name, Prerun Accept File, Prerun Parameters, Prerun Task Attributes, Prerun File Attributes, Prerun End-of-Task Notice, Prerun Template Display File |
| Copy | Prerun Source File or Directory, Prerun From, Prerun Source Kind, Prerun Destination File or Directory, Prerun To, Prerun Destination Kind, Prerun BNA Hostname, Prerun Copy & Compare |
| Change | Prerun Source File or Directory, Prerun New File or Directory, Prerun From |
| Remove | Prerun Source File or Directory, Prerun From |
The prerun runs under the job's own Batch User, with the job's Access Code, and the job's Fail Code and Fail Reset words apply to it too; it has no fields for them. A Start prerun's Prerun Job Summary User (Start) is a second MCP Batch User connection, as the job's own Job Summary User is.
No prerun field is marked required in the form, because which ones are required depends on the Prerun Type. A required one that is blank refuses the job at dispatch with a message that says why, such as … is required when Prerun Type is COPY — see What is refused at dispatch.
A job with a Prerun Type set is refused at dispatch by a relay that has not reported it can send the prerun — any relay older than this release — so it never runs without it: it goes to Initialization Error with RELAY_LSAM_FEATURE_UNSUPPORTED: relay '…' (agent pool '…') has not reported support for … Upgrade the relay. Upgrade every relay that serves an MCP agent before you give a job a prerun — see Keeping a relay current.
What the job reports
When an MCP job of any of the six types ends, the text the agent reports is the job's
termination description — the Termination row on the job's Summary tab, and the value an
Exit Description event compares — such as
12345COMPLETED. Its leading digits are the task's mix number, not an exit code: Continuum reads no
exit code from it, and the outcome is the agent's. An event matches it as text: Equal To with
the whole text, or Contains with part of it — see
Match a message ID exactly (on the IBM i tab, under Batch Jobs).
While the job runs, the agent's status text appears as its Job Status Description in the job's Agent Feedback (on the How It Works tab), which an LSAM Feedback event can watch, and as the Status Description row on the Summary tab. The same Agent Feedback holds the task's Mix Number, Process Name and its processor, I/O, print-line, elapsed, ready-queue and elapsed-wait statistics, which also reset with each run.
A kill the MCP refuses
If you Kill an MCP job and the MCP answers that it did not stop the job (its reply ends in
NOT DSED, which usually means the kill arrived after the job had finished), the job is not
failed. It stays at Job Running To Be Terminated until the MCP reports how the job really ended,
and then takes that outcome. Continuum no longer reports a Kill timeout failure for a job that is
still running on the MCP. If the MCP never answers the kill at all, the kill still times out and the
job fails as before.
Its output
The job's Output tab lists the output the MCP agent filed for the job, and opens each file — while the job is still running too, and for a job whose kill timed out. To ask the agent for it, Continuum uses where the agent reported it filed the output; when the agent did not report that, it uses the job's MCP usercode — the one the job was sent with, from its Connection — and the mix number at the start of the job's status description, as OpCon has always done.
When neither is available, the tab says JORS LIST failed: This MCP job's output can't be listed: OpCon has no record of where the agent filed it…. That happens when the job has not reported its status yet, ran before your relay was upgraded, or reported an empty status description. If it is still running, open the tab again once it has reported its status.
What is refused at dispatch
The form saves these; the job fails when it runs, before anything reaches the agent, with a message that names the field and never the value, such as MCP field fileTitle must be at most 96 characters (RC_INVALID_MCP_FIELD):
- no Connection — MCP job has no batch user attached; select a Batch User
(RC_MCP_NO_BATCH_USER) — or a Connection or Job Summary User that is not an MCP Batch
User connection (
RC_WRONG_CONNECTION_TYPE); - a blank File Title (MCP Run, MCP Start); a blank Report Name, Accept File or Parameters (MCP EAE/AB Suite); a blank Source File or Directory or From (MCP Copy, MCP Change, MCP Remove), Destination File or Directory or To (MCP Copy), or New File or Directory (MCP Change);
- on MCP Copy, a Source Kind or Destination Kind that is not exactly
TAPEorPACK— which only a job edited outside the form can hold; - a value over its limit once any property token in it is resolved, or holding a character the field does not allow, as in the table above;
- a lower-case letter in File Title, a Fail Code or a Fail Reset;
- an Access Code with anything but upper-case letters, digits and spaces;
- more than 10 Task Attributes or File Attributes rows, or a blank row;
- on MCP Start, a Job Summary Family or Job Summary Directory or File with no Job Summary User, or a Job Summary Family with no Job Summary Directory or File;
- a Prerun Type that is not one of the six, or a prerun that breaks any of these rules for its type — a required prerun field left blank is refused with … is required when Prerun Type is … naming the type. The job's own fields are checked first, so a mistake in the job is reported before one in its prerun;
- a character outside the Windows-1252 code page the agent connection uses, in any value;
</F>or<F I=in any value.
A job aimed at an agent that is not an MCP agent is refused at dispatch too, with LSAM_AGENT_TYPE_MISMATCH — see A legacy job aimed at an agent of another type.
Restarting it
An MCP job, of any type, restarts from the beginning only. Its Restart dialog offers no
step, and a restart that names one — from a $JOB:RESTART event, say — is refused with Cannot
restart on a step: restart on step is not available for MCP jobs. See
Restart on a step.
A kill reaches the agent by default. An MCP agent's Allow Kill setting is on unless you turn it off — see Allow Kill.
OS 2200
The OS 2200 LSAM connector (opcon.lsam.os2200) has one job type.
| Job type | Agent type | Purpose |
|---|---|---|
OS 2200 Job (os2200-ecl) | OS 2200 | Start an ECL runstream on a Unisys OS 2200 LSAM agent — see OS 2200 Job |
OS 2200 LSAM — OS 2200 Job
The OS 2200 Job type is new, and its run path has not yet been verified against a real OS 2200 LSAM. Confirm availability with your Continuous contact before planning work around it.
An OS 2200 Job (os2200-ecl) starts an ECL runstream on a Unisys OS 2200 system. The runstream
stays on the system: the job names the file that holds it, by Qualifier and File Name, and
optionally the Element in that file to start. The agent starts it with the run card values you
give, substitutes the runstream's tokens, and decides from your Completion Status rules whether
the run succeeded. The agent can also hold the job until a prerun finishes with Good Fin or
until files it depends on are ready — see LSAM Resolved Dependencies.
The job runs on an OS 2200 agent. It has no Connection field: the run card's User ID
names the user, and -SECURITY- uses the LSAM's own security user. The job needs a
current relay — one that reports it can run OS 2200 jobs. An
older relay refuses the job at dispatch, with RELAY_LSAM_PLATFORM_UNSUPPORTED naming OS 2200
jobs.
The agent's Job Start Format decides how the job is started: in XML, the default, or in the
older non-XML record for an older LSAM. XML needs a relay that has reported it can send it; on any
other relay a job for an agent set to XML is refused with RELAY_LSAM_FEATURE_UNSUPPORTED, and the
message says to upgrade the relay or set the agent to Non-XML. In XML, a value that contains </F>
or <F I= is refused at dispatch. See
OS 2200 Job Start Format.
One relay connection per OS 2200 machine
An OS 2200 LSAM accepts one scheduling connection from a relay at a time, and it can close a connection the moment a second one opens. A BIS agent is handled the same way. Three things follow:
- Reconnects slow down after a dropped connection. When a connection to an OS 2200 or BIS machine is accepted and then ends before the machine has answered, the relay waits before it tries again: about 30 seconds, then 60, then 120, and 120 each time after that, each varied by up to 20%. The wait applies even with a job queued for the machine, and it ends the first time the machine answers. A machine that refuses the connection outright, because it is down, keeps the usual retry pace. Other agent types are not affected.
- A clean relay stop releases the connection first. When the relay is stopped or restarted cleanly, it closes every connection to its legacy machines before its connector exits, so the restarted relay is not refused as a second connection.
- Point only one relay at an OS 2200 agent. Two relays that connect to the same OS 2200 machine can disconnect each other. Do not rely on the machine to settle which of two relays keeps the connection.
OS 2200 Job — parameters
The fields appear in this order under Job Parameters, in five sections with their own headings: Start Command, Run Card Control, Completion Status, LSAM Resolved Dependencies and Tokens.
| Section | Parameter | Required | Limits | Notes |
|---|---|---|---|---|
| — | Machine (from Agent Assignment) | Yes | 1–24 chars | Target OS 2200 agent. Comes from the job's agent assignment, and must match the agent name registered with the relay. |
| Start Command | Qualifier | Yes | ≤12 chars | The qualifier of the file that holds the runstream, such as QUAL. |
| Start Command | File Name | Yes | ≤20 chars | The file that holds the runstream, with its read and write keys if it has any, such as PAYFILE. |
| Start Command | Element | No | ≤25 chars | The element (and version) to start, such as ELT/VERS. Leave it blank to start the file itself. |
| Start Command | Priority | No | One letter, A–Z | Sent upper-case. Leave it blank for the system default. |
| Start Command | Options | No | ≤9 letters | Start options, from B, D, N, O, P, R, S, T, W, X and Y. Sent as typed. |
| Start Command | Condition Word (Octal) | No | 0–9999 | The octal value to place in the run's condition word, entered as written: up to four digits. Default 0. |
| Run Card Control | Run ID | No | ≤6 chars | Leave it blank to let the system assign one. |
| Run Card Control | Account | No | ≤12 chars | |
| Run Card Control | User ID | No | ≤12 chars | Enter -SECURITY- to use the LSAM's security user. |
| Run Card Control | Project | No | ≤12 chars | |
| Run Card Control | Maximum Pages | No | 0–9999 | Default 0. |
| Run Card Control | Maximum Cards | No | 0–9999 | Default 0. |
| Completion Status | Completion Status | No | ≤4 rows | Rules checked against the run's status words — see Completion Status. A new job starts with one row. |
| Completion Status | Anything Else | No | Good Fin / Bad Fin | How the run finishes when no Completion Status row matches. Default Good Fin. |
| LSAM Resolved Dependencies | LSAM Resolved Dependency | No | None / Prerun / File Dependencies | What the agent checks before it starts the job. Default None. See LSAM Resolved Dependencies. |
| Tokens | Tokens | No | ≤251 chars in total | Runstream tokens, one per row, as ?NAME?=value. The rows are sent joined with commas, and together, commas included, they may be at most 251 characters. |
Values are checked twice: when you save, and as they will be sent. Saving refuses a value that
breaks these limits — see What is refused. When the job runs, a line
break, tab or form feed in any value becomes a space, the value is trimmed, and property tokens such
as [[JI.NAME]] are resolved — and the value that results is checked again. A value over its limit
is refused, never cut short. The form takes longer values than these limits so that a token fits.
The workflow name the agent is sent is cut to 20 characters, the width the OS 2200 record holds for it — see The names the agent is sent (on the How It Works tab). Your workflow name is not changed.
Completion Status
Each Completion Status row says how the run finishes — Finish is Good Fin or Bad Fin — when one of the run's status words meets a condition:
| Field | Values |
|---|---|
| Join | And or Or: how the row joins the row before it. The first row has no Join: the form shows it from the second row on. |
| Finish | Good Fin or Bad Fin. |
| Word Part | S1–S6 or T1–T3. |
| Condition | Logical AND, Equal, Not equal, Greater than, Less than, Greater or equal, Less or equal or Range. |
| Value | 0–9999, entered in decimal: the decimal equivalent of the octal status value. The form shows the octal form of what you type beneath the box, such as Octal: 17 for 15. |
| End Value | 0–9999, in decimal, with its octal form beneath it. Shown only when Condition is Range, where it is required and must be greater than Value. Changing the condition away from Range clears it. |
A new job starts with one row: Bad Fin when S2 is Not equal to 0. Add fills a
new row with the same values. Remove the row, or change it, if your runstream needs other rules.
The agent applies the rules. Continuum sends them with the job and reads no exit code of its own; when no row matches, the run finishes as Anything Else says.
LSAM Resolved Dependencies
LSAM Resolved Dependency chooses one thing the agent checks before it starts the job — Prerun or File Dependencies — or None for neither. A job has at most one of the two. Choosing one shows its fields; choosing another, or None, clears them.
Prerun shows a second set of the job's own fields, each titled Prerun … — Prerun
Qualifier, Prerun File Name, Prerun Element, Prerun Priority, Prerun Options,
Prerun Condition Word (Octal), Prerun Run ID, Prerun Account, Prerun User ID,
Prerun Project, Prerun Maximum Pages, Prerun Maximum Cards, Prerun Completion Status
and Prerun Anything Else — with the same limits and rules as the job's, for a runstream the agent
starts first. The job starts only when the prerun finishes with Good Fin, as its Prerun
Completion Status rules and Prerun Anything Else decide. The prerun needs Prerun
Qualifier, Prerun File Name and from one to four Prerun Completion Status rows; a new
prerun starts with one row, Bad Fin when S2 is Not equal to 0, as the job does.
File Dependencies is a list of up to eight files. The job starts only when every file meets its condition:
| Field | Required | Limits | Notes |
|---|---|---|---|
| Qualifier | Yes | ≤12 chars | The qualifier of the file. Put # before or after it to look in the shared directory, such as #QUAL. |
| File Name | Yes | ≤12 chars | The file to check, such as DATAFILE. |
| Type | Yes | see the next table | What must be true of the file. |
| Size | For Size only | 0–999 | The size, in thousands of tracks, that the file must exceed. Shown only when Type is Size, where it is required. |
| Type | The file must be |
|---|---|
| Exists | present |
| Created | present, and catalogued today |
| Deleted | absent |
| Size | larger than Size |
| Referenced | referenced today |
| Assigned | assigned now |
| Backed Up | backed up, with a current backup |
| Unloaded | unloaded now |
A prerun that does not finish with Good Fin, or a file that does not meet its condition, does not
fail the job. When the agent reports the check as failed, the job moves to Pre-run Failed and
is started again, the check first, about three minutes later — as many times as it takes — see
A prerun that fails is re-attempted (on the How It Works tab, under
Shared Parameters). While the agent reports that it is running the prerun, the job reads
Pre-run Active. An agent that instead reports the job as running while it waits leaves it
Running, with the agent's text, such as FILE DEPEND NOT MET, as its Job Status Description
in Agent Feedback.
A job with a prerun or file dependencies is refused at dispatch by a relay that has not reported it can send them — any relay older than this release — so it never starts without them: it goes to Initialization Error with RELAY_LSAM_FEATURE_UNSUPPORTED: relay '…' (agent pool '…') has not reported support for … Upgrade the relay. Upgrade every relay that serves an OS 2200 agent before you give a job a prerun or file dependencies — see Keeping a relay current.
What the job reports
When the job ends, the text the agent reports is the job's termination description — the Termination row on the job's Summary tab, and the value an Exit Description event compares. Continuum reads no exit code from it, and the outcome is the agent's. An event matches it as text: Equal To with the whole text, or Contains with part of it — see Match a message ID exactly (on the IBM i tab, under Batch Jobs).
While the job runs, the agent's status text appears as its Job Status Description in the job's Agent Feedback (on the How It Works tab), which an LSAM Feedback event can watch.
Its output
The job's Output tab lists the
job's output, such as its PRINT$ file, and opens each file, only when the OS 2200 agent itself
reports where it filed it: Continuum has no other way to find an OS 2200 job's output, and
asks the agent nothing until it has that. So:
- for a run whose agent did not report it, the tab says JORS LIST failed: This OS 2200 job's output can't be listed: its agent did not report where the job's output is… — which is also what a job that ran before your relay was upgraded shows;
- while the job is still running and the agent has not reported it yet, the tab says This OS 2200 job is still running and its agent has not yet reported where its output is…. Try again later.
What is refused, at save and at dispatch
The editor refuses to save an OS 2200 Job that the agent could not be sent. Save in the job editor, and saving the workflow, stop with a message that names the field and never the value, such as Qualifier is required. or Completion Status row 2: End Value must be greater than Value for Range. — the workflow editor puts the job's name in front, as Job 'PAYROLL': …. Continuum refuses the same workflow however it is saved, and refuses it on import. Every OS 2200 Job in the workflow is checked on each save, including ones you did not change.
These are refused:
- a blank Qualifier or File Name, or a Qualifier that starts with
<D>; - a value over its limit;
- a Priority that is not one letter, or Options with a letter not in the list above;
- a Condition Word (Octal), Maximum Pages, Maximum Cards, Value or End Value
that is not a whole number from
0to9999; - more than 4 Completion Status rows; a row with no Finish, Word Part, Condition or Value; a Range row with no End Value, or one not greater than its Value; a Join on the first row, or none on a later row;
- a Prerun with a value that breaks the same rules as the job's — named Prerun …, such as Prerun Qualifier is required. — or with no Prerun Completion Status row, or more than four;
- File Dependencies with no row or more than eight, a row with no Qualifier, File Name
or Type, a Qualifier or File Name over 12 characters, or a Size that is missing on
a Size row, set on any other row, or not a whole number from
0to999— a row is named by its number, as File Dependency 3 Size; - Prerun or File Dependencies chosen with nothing filled in, such as Prerun is required when LSAM Resolved Dependency is Prerun., or values left in the one not chosen — which only a job edited outside the form can hold;
- a blank Tokens row, or Tokens over 251 characters in total, commas included;
- a control character, or any character outside the Windows-1252 code page the agent connection uses, in any value.
Save checks the value as you typed it, so a leading or trailing space counts toward its limit and a line break or tab in a value is refused. When the job runs, a line break, tab or form feed becomes a space and the value is trimmed before it is checked.
Some values are checked only when the job runs. A value that holds a property token, such as
[[JI.NAME]], is checked once the token is resolved; a change you save to one run of a job, rather
than to the workflow, is not checked at save; and neither is a value a transformation rule rewrites
when the workflow is deployed. Such a job fails when it runs, before anything reaches the agent, with
a message that names the field and never the value, such as OS 2200 field qualifier is required
(RC_INVALID_OS2200_FIELD). A Completion Status or Tokens row is named by its number, from
1.
A job aimed at an agent that is not an OS 2200 agent is refused at dispatch too, with LSAM_AGENT_TYPE_MISMATCH — see A legacy job aimed at an agent of another type.
Restarting it
An OS 2200 Job restarts from the beginning only. Its Restart dialog offers no step, and a
restart that names one — from a $JOB:RESTART event, say — is refused with Cannot restart on a
step: restart on step is not available for OS 2200 jobs. See
Restart on a step.
A kill never reaches the agent. Jobs on an OS 2200 agent can never be killed — see Allow Kill.
Troubleshooting
The outcomes every legacy job can end in come first. The troubleshooting rows that follow are grouped by platform: All Platforms holds what any legacy job can run into, including the File Arrival rows that apply on both Windows and UNIX.
- Outcomes
- All Platforms
- Windows
- UNIX/Linux
- IBM i
- SQL
- SAP R/3
- SAP BW
- z/OS
- MCP
- OS 2200
Outcomes
Outcomes
| Result | When |
|---|---|
| Success | The exit code satisfied the job's exit criteria (on the How It Works tab, under Exit Criteria) — or, with no table, was 0 — and no UNIX Signal Criteria or Core Dump Flag condition marked it failed. |
| Failed — by criteria | The exit code resolved to failed against the job's exit criteria (on the How It Works tab, under Exit Criteria), or a UNIX signal/core dump matched a failure condition. |
| Success or failure — SAP R/3 Job | The SAP agent's own verdict. There are no exit criteria; the leading code of the agent's text is the exit code. See What the job reports (on the SAP R/3 tab). |
| Success or failure — SAP BW Process Chain | The SAP BW agent's own verdict, in the same way. See What the job reports (on the SAP BW tab). |
| Success or failure — z/OS jobs | The z/OS agent's own verdict: from the job's condition-code range and its Step Control rows on a z/OS Batch Job or z/OS Started Task, from the return-code range on a z/OS REXX Exec, and finished once the command is issued on a z/OS Console Command. There are no exit criteria; the agent's text is the termination description. See What the job reports (on the z/OS tab). |
| Success or failure — every MCP job type | The MCP agent's own verdict, from the job's fail codes matched against the MCP console display. There are no exit criteria; the agent's text is the termination description. See What the job reports (on the MCP tab). |
| Success or failure — OS 2200 Job | The OS 2200 agent's own verdict, from the job's Completion Status rules, or its Anything Else setting when none matches. There are no exit criteria; the agent's text is the termination description. See What the job reports (on the OS 2200 tab). |
| Failed — dispatch rejected | The job couldn't be dispatched (e.g. the attached batch user is of the wrong platform's type or is incomplete, the job's machine doesn't match a registered LSAM agent, or it names an agent of another type). |
All Platforms
Troubleshooting
| Symptom | Likely cause | Resolution |
|---|---|---|
| An LSAM command finds no agent to run on | The relay is down, or the LSAM machine is unreachable | Check the relay and LSAM connectivity (Administrator) — this is not a Universal Agent issue. |
| Legacy agent shows Unknown (yellow) | Its relay is stale/down — reachability can't be determined (not the same as the agent being down) | Check the relay; agents leave Unknown within a heartbeat once it recovers (Administrator). |
| Newly registered legacy agent isn't picked up | Relay hot-reconcile hasn't run yet, or registration failed (e.g. duplicate name per relay) | Wait one heartbeat; confirm the agent was registered on the Agents tab of the Agents page with a name unique to that relay (Administrator). |
| Job can't be placed / unknown machine | The job's machine name doesn't match the LSAM agent name registered with the relay | Pick the registered agent under Agent Assignment, which sets the machine name (Builder/Administrator). |
| Job fails at dispatch with LSAM_AGENT_TYPE_MISMATCH: this job needs an agent of type …, but agent '…' on relay '…' has type … | The job's machine is an agent of another type than the job type's platform | Point the job at an agent of the type the message names (Builder). See A legacy job aimed at an agent of another type. |
Any legacy job rejected at dispatch: RC_WRONG_CONNECTION_TYPE | The job's connection is of the wrong platform's type — usually from a hand-edited config, an import, or a transformation rule | Re-pick the account in the Connection field (Builder). |
| Job ran but is marked failed | An Exit Criteria (or UNIX Signal Criteria or Core Dump Flag) condition matched | Confirm the criteria match the program's real success/failure codes (Builder). |
| Job rejected at dispatch on its exit criteria | More than 20 conditions, a missing or unrecognised operator, or a value that isn't a whole number in range | Use up to 20 conditions, each with a valid operator and a whole-number value in range (Builder). |
Job with Finish OK, a range row, or more than five conditions never starts | The relay serving that machine predates those forms and refuses the dispatch rather than sending a table it cannot represent | Upgrade the relay. Until then, express the table as up to five plain failure conditions with Exit Criteria Result Fail (Administrator). |
| Job finished OK but the agent's log says it failed | Its table was evaluated centrally, so the agent was given the platform default and reached the opposite conclusion | Read the Exit Criteria Evaluated entry on the job's Status History tab in Processes, which records the exit code, the row that matched and the verdict (Operator). |
| File Arrival job rejected at dispatch | The file name is blank, too long (4000 characters on Windows, 255 on UNIX), or contains a refused character (" on Windows; see What the path may contain (on the UNIX/Linux tab, under File Arrival) on UNIX); an offset or stable time isn't a whole number in range; or End Offset is earlier than Start Offset | Give a valid file name; use whole-number offsets and stable time in range, and keep End Offset at or after Start Offset. The form won't catch these — they surface only at dispatch (Builder). |
| File Arrival job fails with exit code 2 | The directory in File Name doesn't exist on the machine — only the file name part may use wildcards | Point File Name at a directory that already exists on the LSAM machine, creating it first if needed (Builder / Administrator). |
| File Arrival job never finds the file | The window (Start Offset / End Offset) closed before the file arrived, or the path or wildcard pattern doesn't match | Widen the window to when the file really appears, and check the path and pattern. Check the Day Offset boxes too — a minute count typed there reads as days. For a bare "does it exist now?" check, set both offsets to zero (Builder). |
Windows
| Symptom | Likely cause | Resolution |
|---|---|---|
Windows job rejected at dispatch: no login name / no password (RC_NO_RUNAS_LOGIN / RC_NO_RUNAS_PASSWORD) | The attached Windows Batch User is a named account missing its login name or password | Complete the connection (Administrator), or leave the job's Connection empty to run as the agent's service account (Builder) — see Connections. |
| File operation (Copy/Move/Rename/Delete) rejected at dispatch | A required field is blank, an attribute filter value is invalid, or the assembled command line or working directory is too long | Fill every required field and shorten the command. The form won't catch these — they surface only at dispatch (Builder). |
| File Delete removed the wrong files, or nothing | The comma-separated file list or an attribute filter doesn't match what you intended | Re-check the list and the filters: Include deletes only files carrying that attribute, Exclude skips them (Builder). |
| Windows Command job has no way to embed a script | The Windows Command job type has no embedded-script field | Use the Windows - Embedded Script job type to run a library script, or Start Image and Arguments with a script already on the machine (Builder). |
| A Windows Command job with a Process Priority other than Normal goes to Initialization Error with RELAY_LSAM_FEATURE_UNSUPPORTED | The relay serving the agent predates process priorities | Upgrade the relay (Administrator), or set Process Priority back to Normal. |
| WS_FTP Pro job rejected at dispatch | A required field is blank, a required text field contains a ", a profile field contains a :, or the assembled command line/working directory is too long | Fill every required field; remove " from all five text fields and : from the profile names; shorten the location/options (Builder). The form won't catch these — they surface only at dispatch. |
| WS_FTP Pro transfer fails to authenticate | The referenced source/destination site profile doesn't exist on the agent, or its stored credentials are wrong | Create/repair the named profiles in the WS_FTP Pro installation on that machine (Administrator); the job carries no credentials of its own. |
| Windows job ignores environment variables | The relay or agent is running an older build that doesn't emit them on Windows | Confirm the relay and agent are on a current build (Builder/Administrator). |
| Windows job now fails (or succeeds) where it didn't before | Its Exit Criteria are being evaluated. Earlier builds discarded them and failed on any non-zero exit code, so criteria that were configured but ineffective now take effect | Re-check the criteria against the program's real codes — this is the criteria doing what they say, not a new fault (Builder). |
UNIX/Linux
| Symptom | Likely cause | Resolution |
|---|---|---|
UNIX job rejected at dispatch: RC_NO_UID / RC_NO_GID | No batch user is attached, or its user or group is blank | Attach a complete UNIX Batch User connection (Builder/Administrator). |
| UNIX job fails with a permissions error | The attached UNIX Batch User's user or group lacks rights to run the Start Image | Attach a batch user whose user/group has the right permissions on the LSAM machine (Builder/Administrator). |
IBM i
| Symptom | Likely cause | Resolution |
|---|---|---|
| IBM i job refused at dispatch naming a list, a row and a column | A message-monitor, spool-file or dynamic-variable row breaks its column rules, a list holds more than 99 rows, or a Restricted Mode job carries a message or spool list | Fix the row the message names. The form does not check list rows; the relay does, at dispatch, and nothing reaches the machine. See IBM i lists (on the IBM i tab, under Lists) (Builder). |
| IBM i - FTP job refused at dispatch: IBM i FTP … requires a … | The chosen Action needs a field the job lacks — a Local File Name for PUT, MPUT or GET, a Local Library or Directory for every action but DEL and MDEL — or the Action or Transfer Type is not one of its exact values | Fill the field the message names, or pick the Action from the list. See IBM i - FTP (on the IBM i tab, under FTP) (Builder). |
| An IBM i job with message monitors, spool files or dynamic variables goes to Initialization Error with RELAY_LSAM_FEATURE_UNSUPPORTED | The relay serving the machine has not reported that it can send the lists | Upgrade the relay (Administrator). See The lists need a current relay (on the IBM i tab, under Lists). |
| IBM i - File Arrival job refused at dispatch | The File Name is missing or holds a tab or line break; the window end is earlier than its start; or an authority check is on with no Connection | Fix the field the message names, or attach an IBM i Batch User for the authority check (Builder). See IBM i - File Arrival (on the IBM i tab, under File Arrival). |
| IBM i - File Arrival job ends without waiting for a file that arrives later | With no re-check interval, the agent checks for the file once | Add a $@RECHKFREQ dynamic variable with the seconds between checks, and set the window end (Builder). See IBM i - File Arrival (on the IBM i tab, under File Arrival). |
| An Exit Description event never fires on an IBM i job | The job's description is a message ID such as SMA0037, and the event compares it with a number — or the relay predates the change and passes no message ID | Set the event's Value to the message ID itself, such as Equal To SMA0037, and upgrade the relay (Builder / Administrator). See The message ID is the job's Exit Description (on the IBM i tab, under Batch Jobs). |
SQL
| Symptom | Likely cause | Resolution |
|---|---|---|
| MS SQL DTExec job refused at dispatch naming the batch user | The package is on the file system, which always runs as the agent's service account, and the job still names one — written outside the job editor, which hides and clears Connection for the file system | Switch Package Location away from File system and back to clear it, and save (Builder). See MS SQL DTExec (on the SQL tab, under MS SQL DTExec). |
| MS SQL DTExec job goes to Initialization Error with RELAY_LSAM_FEATURE_UNSUPPORTED naming configuration files | The job has a Configuration File, and the relay serving the machine has not reported support for it | Upgrade the relay (Administrator), or clear Configuration File and put the path in Other Options as a /ConfigFile switch (Builder). See The configuration file. |
MS SQL DTExec job refused at dispatch naming configFile | The Configuration File is set for a package that is not on the file system, holds a double quote, ends with a backslash, is too long with Other Options, or is set while Other Options already has a configuration-file switch | Fix the path, or keep only one of the two (Builder). |
MS SQL Script or Other DB job refused at dispatch naming scriptSource | Script Source disagrees with the script field that is filled in — a job written outside the job editor | Open the job, set Script Source to match, and save (Builder). |
| MS SQL DTExec job refused at dispatch naming the package path or server | The Package Path contains a double quote, ends with a backslash or holds a character outside Windows-1252, or the Server Name is missing or contains a space or a double quote | Correct the field the message names (Builder). |
| MS SQL DTExec job fails with return code 4, 5 or 6 | DTExec could not find the package (4), could not load it (5), or hit a syntax or internal error (6) — often a wrong Package Location or Package Path, or a bad switch in Other Options | Read the job's output, which is DTExec's own, and correct the location, path or switch (Builder). |
| Oracle job refused at dispatch naming a field or the batch user | No batch user, a batch user with no password, a missing Script File Path, Server Name or Service Name, inline SQL statements left on the job, or a character the SQL*Plus command line cannot carry | Fix the field or the credential the message names (Builder / Administrator). See What is refused at dispatch (on the SQL tab, under Oracle). |
Oracle job fails with return code 1017 | Oracle refused the batch user's user name or password | Correct the credential on the SQL Batch User connection (Administrator). |
Oracle job fails with return code 1 | The script file is not at the path given, or another operating-system error stopped SQL*Plus | Check Script File Path on the agent machine, and that SQL*Plus is installed there (Builder / Administrator). |
| Other DB job refused at dispatch: must not contain a password or other secret or must not name a login | The connection string holds a password, secret, token or login key — even inside a quoted or braced value | Remove the key from the connection string and put the login in a SQL Batch User (Builder). See The connection string must carry no login or secret (on the SQL tab, under Other DB). |
| Other DB OLE DB job refused at dispatch naming the batch user | A batch user with Windows Authentication off — the agent cannot pass a database login to an OLE DB provider | Turn Windows Authentication on, or remove the batch user (Builder). |
| Other DB job fails with no return code | The connection did not open — a driver or DSN not installed on the agent, or a wrong server or login | Read the job output for the driver's message, and check the driver or DSN on the agent machine (Administrator). |
| MySQL job refused at dispatch naming a field or the batch user | No batch user, a batch user with no password, a missing Script File Path, Server Name or Database Name, inline SQL statements left on the job, or a character the Windows command line cannot carry | Fix the field or the credential the message names (Builder / Administrator). See What is refused at dispatch (on the SQL tab, under MySQL). |
| MySQL job refused at dispatch: cannot be used for the Script File Path when MySQL user variables are set | With user variables set, the path holds & | ^ ( ) ;, or a folder, file or server name that starts with a lowercase g | Rename the folder or file, or move the script (Builder / Administrator). |
MySQL job fails with return code 1 | A MySQL error, a failed login or connection, a missing script file — or, with no user variables, a mysql client that is not installed or not on the PATH | Read the mysql client's message in the job output, and check the credential, Server Name, Script File Path and the client on the agent machine (Builder / Administrator). |
MySQL job fails with return code 255 | The job has user variables and the mysql client is not installed or not on the agent machine's PATH | Install the mysql client on the agent machine and add it to the PATH (Administrator). |
SAP R/3
| Symptom | Likely cause | Resolution |
|---|---|---|
| SAP R/3 Job refused at dispatch naming a field | A missing or over-long SAP Job Name, a name containing * or ?, a SAP Job Number that is not 1 to 8 digits, an Exec. Target over 64 characters, or a character outside Windows-1252 | Fix the field the message names (Builder). See What is refused at dispatch (on the SAP R/3 tab). |
| SAP R/3 Job fails at dispatch with RELAY_LSAM_PLATFORM_UNSUPPORTED … SAP R/3 and CRM jobs | The relay serving the agent predates the SAP R/3 Job type | Upgrade the relay (Administrator). See Serving legacy LSAM agents. |
SAP R/3 Job fails with exit code 70006 | The SAP job with that name and number does not exist, or is not in Scheduled or Ready status | Check the job name and number in SM37, and that the job is released or scheduled in SAP, or pick it again with Search… on SAP Job Name. To find every job with a stale number, use Check SAP Job Numbers (Builder). |
| Search… is missing beside an SAP field | The job's Agent Assignment is a pool, no Primary Agent is chosen, or the editor is read-only | Choose Specific Agent and a Primary Agent (Builder). See Looking up the job and the server in SAP. |
| Search… is unavailable: You don't have permission to run agent lookups. Ask an administrator. (or Start Check on Check SAP Job Numbers is unavailable) | Your roles do not grant Agents Execute under All workspaces | Grant it to the role under All workspaces (Administrator). See SAP lookup permission. |
| A search or check reads The relay serving this agent can't run SAP lookups. It may need to be upgraded, or its job output (JORS) channel is turned off. | The relay does not report that it can run SAP lookups | Upgrade the relay (Administrator). See Keeping a relay current. |
| A search or check reads The relay serving this agent is offline. or The relay has not loaded this agent yet. | The agent's relay is down, or has not yet picked up the agent | Bring the relay online, or wait and search again (Administrator). |
| A search or check reads The agent's query port could not be reached. Check the agent's Query Port setting and that the port is open from the relay. | The agent's Query Listener Port does not match the port the agent listens on, or the port is blocked from the relay | Correct Query Listener Port on the agent, or open the port (Administrator). See Agents and agent pools. |
| A search or check reads The agent closed the connection without answering. Check the agent's SAP system settings. | The agent's SAP system details may not match its own SAP configuration | Correct them on the agent (Administrator). See SAP system details. |
| A search or check reads The agent rejected the search: (or the query:) and the agent's own text | The agent refused it, for the reason in its text | Act on the agent's text (Administrator). |
| A search reads The agent did not answer in time. or The agent has not answered yet. Search again later. | The agent did not answer the relay in time, or no answer came within 3 minutes | Search again later (Builder). |
| Check SAP Job Numbers shows Cannot Check with The job's agent and its machine name differ | The machine name stored in the job is not its primary agent's name | Open the job and choose its Primary Agent again, then check again (Builder). |
SAP R/3 Job fails with exit code 70004 or 70005 | No SAP user is configured on the agent, or its SAP logon failed | Correct the SAP user in the agent's own configuration on the machine (Administrator). |
| Kill on a SAP R/3 Job is refused with does not allow its jobs to be killed | The agent's Allow Kill is off, which is the default for SAP R/3 and CRM | Turn on Allow Kill for the agent if its jobs should be killable (Administrator). See Allow Kill. |
SAP BW
Troubleshooting
| Symptom | Likely cause | Resolution |
|---|---|---|
| SAP BW Process Chain job refused at dispatch naming processChainName | The name is missing, longer than 25 characters once its tokens are resolved, or holds a character outside the Windows-1252 code page | Correct the Process Chain Name, and check what any property token in it resolves to (Builder). See What is refused at dispatch (on the SAP BW tab). |
| SAP BW Process Chain job fails at dispatch with RELAY_LSAM_PLATFORM_UNSUPPORTED … SAP BW jobs | The relay serving the agent predates the SAP BW Process Chain type | Upgrade the relay (Administrator). See Keeping a relay current. |
SAP BW Process Chain job fails with 1 - NOT FOUND | SAP does not know a chain by that name | Check the technical name in RSPC, or pick it with Search… on Process Chain Name (Builder). See Looking up the process chain in SAP. |
| Search… on Process Chain Name is missing, unavailable, or the search fails | The same causes as on an SAP R/3 Job | See the SAP lookup rows on the SAP R/3 tab. |
| An SAP BW agent never connects | Its ports do not match the agent's configuration. An SAP BW agent registered earlier on the defaults kept 13100 and 13110, not the SAP BW agent's own 14100 and 14110 | Set LSAM Port and JORS Port to the ports the agent listens on (Administrator). See Agents and agent pools. |
| Kill on an SAP BW Process Chain job is refused with Cannot kill: jobs on SAP BW agents cannot be killed | Jobs on an SAP BW agent can never be killed | Let the chain end, or stop it in SAP (Administrator). |
z/OS
| Symptom | Likely cause | Resolution |
|---|---|---|
| z/OS Batch Job refused at dispatch: z/OS field … must be 1-8 characters (A-Z, 0-9, @, # or $) and must not start with a digit | A Member, DD Name, User ID, Temporary Member, Override DD Name or Pre-run System is not a z/OS name once its property tokens are resolved | Correct the field the message names, and check what any property token in it resolves to (Builder). See What is refused at dispatch (on the z/OS tab). |
| z/OS Batch Job refused at dispatch: z/OS field memberName is blank and the job name does not give a valid z/OS member name; set Member | Member is blank, and the start of the job's name is not a valid member | Set Member on the job (Builder). See The member a blank Member submits (on the z/OS tab). |
| z/OS Batch Job refused at dispatch naming stepControl | A Step Control row breaks its rules, two rows repeat the same step name and condition-code range, or the list holds more than 80 rows | Fix or remove the row the message names (Builder). See Step Control (on the z/OS tab). |
| z/OS Batch Job refused at dispatch naming datasetPrerun | A Dataset Pre-run row breaks its rules — an Exists row with a wildcard or a .G0000V00 ending, for example — two rows repeat each other, or the list holds more than 80 rows | Fix or remove the row the message names (Builder). See Dataset Pre-run (on the z/OS tab). |
| A z/OS job with Step Control rows, pre-runs or Restart options goes to Initialization Error with RELAY_LSAM_FEATURE_UNSUPPORTED | The relay serving the agent has not reported that it can send them | Upgrade the relay (Administrator). See Keeping a relay current. |
| z/OS Batch Job fails at dispatch with RELAY_LSAM_PLATFORM_UNSUPPORTED … z/OS jobs | The relay serving the agent predates the z/OS Batch Job type | Upgrade the relay (Administrator). See Keeping a relay current. |
| A z/OS REXX Exec or z/OS Console Command restarted on a step is refused with restart on step is not available for this kind of z/OS job | Only a z/OS Batch Job or z/OS Started Task can restart on a step | Restart it from the beginning, or clear the event's Restart Step (Builder). See Restarting it (on the z/OS tab). |
A z/OS job restarted from a $JOB:RESTART event is refused with the start or end step is not one of the steps this job reported or the end step comes before the start step | A step in the event's start|end range is not one the agent reported, or the range runs backwards | Use steps from the job's Restart dialog, in order (Builder). See Restarting it (on the z/OS tab). |
| A z/OS Started Task, REXX Exec or Console Command fails at dispatch with Unsupported z/OS task type for TX1 translation | The relay serving the agent predates the job type | Upgrade the relay (Administrator). See Keeping a relay current. |
| A z/OS job refused at dispatch with z/OS field … is not used by a … job; remove it | The job carries a field of another z/OS job type, left by an edit on the Code tab or through the API | Remove the field the message names (Builder). See What all three share (on the z/OS tab). |
| A z/OS Tracked Job or z/OS Queued Job fails without reaching the agent, with … runs only after its agent reports it … (RC_LSAM_PRIVATE_DATA_REQUIRED) | The job started from its workflow — built and released, or force-started — before the agent reported it, so there was nothing to send | Let the agent's report add the job. See z/OS Tracked Job and z/OS Queued Job (on the z/OS tab). |
| A z/OS Tracked Job or z/OS Queued Job is refused with RELAY_LSAM_FEATURE_UNSUPPORTED | The relay serving the agent's pool has not reported that it can carry the agent's reference for the job | Upgrade the relay (Administrator). See Keeping a relay current. |
| A job the z/OS agent tracks or queues never appears in its workflow | The agent's report was refused — the workflow is not built for that date, the job is not on it or is not a Tracked or Queued Job, or the job is already queued — and the agent was told why | Check the job's type and that the workflow instance is built for the date, and look for the event in the event log, filtered to source AGENT (Builder). See Tracked and queued jobs. |
| A z/OS Console Command refused at dispatch with z/OS field operatorCommand is required | Operator Command is blank once its property tokens are resolved | Enter the command, and check what any property token in it resolves to (Builder). |
| View / edit JCL is not offered on a z/OS job | The job is a Started Task, Console Command, Tracked Job or Queued Job, or it is assigned to a pool or a legacy agent group rather than one z/OS agent | Assign the job a Specific Agent (Builder). See Viewing and editing JCL (on the z/OS tab). |
| The JCL dialog offers no Save, or says You have no z/OS user id assigned. Ask an administrator. | You do not hold z/OS JCL Edit, or no z/OS user ID is assigned to you | Ask an administrator to grant the permission or assign the ID on the z/OS User IDs tab (Administrator). See Viewing and editing JCL (on the z/OS tab). |
| The JCL dialog says The save may or may not have been applied. Load the member again to check. | The text was sent, and no answer Continuum can trust came back | Load the member to see what it holds before you save again (Operator). |
| A z/OS Batch Job's Output tab says This z/OS job's output can't be listed: its agent did not report where the job's output is for this run… | The z/OS agent did not report where it filed this run's output — or the run predates your relay's upgrade — and Continuum has no other way to find it | Read the job's output on the mainframe (Operator). See Its output (on the z/OS tab). |
| A z/OS Batch Job's Output tab says This z/OS job is still running and its agent has not yet reported where its output is | The agent reports where the output is at the latest when the job ends | Open the tab again later (Operator). |
| A z/OS Batch Job's Output tab says JORS port not configured for agent: … | The agent's JORS Port is empty, which is where a z/OS agent starts | Set JORS Port to the port the agent's JORS listener uses (Administrator). See Agents and agent pools. |
MCP
| Symptom | Likely cause | Resolution |
|---|---|---|
| MCP job refused at dispatch: MCP job has no batch user attached; select a Batch User | The job has no Connection | Pick an MCP Batch User connection in Connection (Builder). See MCP Batch User. |
| MCP job refused at dispatch: MCP field … (RC_INVALID_MCP_FIELD) | A value breaks its field's rules once its property tokens are resolved — too long, blank where it is required, a lower-case letter in File Title or a fail code, a character the field does not allow, or, on MCP Copy, a Kind other than TAPE or PACK | Correct the field the message names, and check what any property token in it resolves to (Builder). See What is refused at dispatch (on the MCP tab). |
| MCP job fails at dispatch with RELAY_LSAM_PLATFORM_UNSUPPORTED … MCP jobs | The relay serving the agent predates the MCP job types | Upgrade the relay (Administrator). See Keeping a relay current. |
| MCP Copy, MCP Change or MCP Remove job fails at dispatch with Unsupported MCP task type … (RC_MCP_UNSUPPORTED_TASK_TYPE) | The relay serving the agent runs MCP jobs but predates these three types | Upgrade the relay (Administrator). See Keeping a relay current. |
| Creating or renaming an MCP agent is refused with MCP agent names must end with a period (.) and be at most 24 characters | The name has no trailing ., or is longer than 24 characters | Give the agent a name such as MCPPROD. (Administrator). See MCP agent names. |
| An MCP job refused at dispatch with MCP field prerun… is required when Prerun Type is … | The chosen Prerun Type needs a prerun field that is blank. No prerun field is marked required in the form | Fill in the field the message names, or clear Prerun Type (Builder). See Prerun (on the MCP tab). |
| An MCP job with a prerun goes to Initialization Error with RELAY_LSAM_FEATURE_UNSUPPORTED | The relay serving the agent has not reported that it can send the prerun | Upgrade the relay (Administrator). See Keeping a relay current. |
| An MCP job sits in Pre-run Failed and starts again every few minutes | Its prerun failed — an MCP Remove prerun with nothing to remove, for example — so the job is re-attempted rather than failed | Fix what the prerun checks, or Cancel or Skip the job (Operator). See A prerun that fails is re-attempted. |
| An MCP job's Output tab says This MCP job's output can't be listed: OpCon has no record of where the agent filed it… | The agent did not report where it filed the output, and the job has no status description to take the mix number from — it has not reported its status yet, ran before your relay was upgraded, or reported an empty one | If the job is still running, open the tab again later; otherwise read its output on the MCP system (Operator). See Its output (on the MCP tab). |
An MCP job restarted from a $JOB:RESTART event is refused with restart on step is not available for MCP jobs | The event names a Restart Step, and an MCP job restarts from the beginning only | Clear the event's Restart Step (Builder). |
OS 2200
| Symptom | Likely cause | Resolution |
|---|---|---|
| An OS 2200 Job will not save: Qualifier is required., Completion Status row 1: Value is required. or another message naming a field | A value breaks its field's rules — blank where it is required, too long, a number outside 0–9999, a Completion Status row that is incomplete, or Tokens over 251 characters in total. Every OS 2200 Job in the workflow is checked, including ones you did not change | Correct the field the message names, in the job the message names (Builder). See What is refused, at save and at dispatch (on the OS 2200 tab). |
| OS 2200 Job refused at dispatch: OS 2200 field … (RC_INVALID_OS2200_FIELD) | A value breaks its field's rules once its property tokens are resolved, or a change saved to one run or a transformation rule set a value that save did not check | Correct the field the message names, and check what any property token in it resolves to (Builder). See What is refused, at save and at dispatch (on the OS 2200 tab). |
| OS 2200 Job fails at dispatch with RELAY_LSAM_PLATFORM_UNSUPPORTED … OS 2200 jobs | The relay serving the agent predates the OS 2200 Job type | Upgrade the relay (Administrator). See Keeping a relay current. |
| An OS 2200 Job finishes Good Fin or Bad Fin other than you expected | No Completion Status row matched, so Anything Else decided, or a row's Value was entered as the octal status value rather than its decimal equivalent | Check the rows and enter each Value in decimal (Builder). See Completion Status. |
| An OS 2200 Job will not save: Prerun is required when LSAM Resolved Dependency is Prerun., File Dependency 2 Size … or another Prerun … or File Dependency message | The chosen LSAM Resolved Dependency is empty, or one of its values breaks its rules | Fill in or correct the field the message names, or set LSAM Resolved Dependency to None (Builder). See LSAM Resolved Dependencies (on the OS 2200 tab). |
| An OS 2200 Job sits in Pre-run Failed and starts again every few minutes | Its prerun did not finish with Good Fin, or a file dependency is not met | Fix what the prerun or the file checks, or Cancel or Skip the job (Operator). See LSAM Resolved Dependencies. |
| An OS 2200 Job with a prerun or file dependencies goes to Initialization Error with RELAY_LSAM_FEATURE_UNSUPPORTED | The relay serving the agent has not reported that it can send them | Upgrade the relay (Administrator). See Keeping a relay current. |
| An OS 2200 Job's Output tab says This OS 2200 job's output can't be listed: its agent did not report where the job's output is… | The OS 2200 agent did not report where it filed this run's output — or the run predates your relay's upgrade — and Continuum has no other way to find it | Read the job's output on the OS 2200 system (Operator). See Its output (on the OS 2200 tab). |
| An OS 2200 Job's Output tab says This OS 2200 job is still running and its agent has not yet reported where its output is | The agent has not reported it yet | Open the tab again later (Operator). |
| Kill on an OS 2200 Job is refused with Cannot kill: jobs on OS 2200 agents cannot be killed | Jobs on an OS 2200 agent can never be killed | Let the run end, or stop it on the OS 2200 system (Administrator). |
An OS 2200 Job restarted from a $JOB:RESTART event is refused with restart on step is not available for OS 2200 jobs | The event names a Restart Step, and an OS 2200 Job restarts from the beginning only | Clear the event's Restart Step (Builder). |
Contact support when
- The relay and LSAM machine are confirmed healthy and the job's machine name is correct, but jobs still can't be dispatched or never report completion.
- Exit/signal handling doesn't match what the LSAM machine actually returned.
Include the connector (UNIX, Windows, IBM i, SQL, SAP R/3 and CRM, SAP BW, z/OS, MCP or OS 2200), the job type, the machine name, the Start Image and Parameters, the configured Exit Criteria, and the relay/LSAM status.