Re: Check if a particular job is always active

On 27-Sep-2016 12:31 -0500, GioArt wrote:

I need to be sure that a particular batch job is always active; is
there any system service that can I use to check that ?

This is my scenario : at IPL start my job IMP_ORD (that import
customer orders) and another job that should check every five
minutes if the IMP_ORD is active.

There is no [direct\supported\ ability to access the _system_ portion of the IPL for that purpose. The _user_ portion of the IPL is the job started for activating the System Startup Program [as defined by the] (QSTRUPPGM) system value.

Similarly to that automatic startup, those jobs can be activated by one or more Autostart Job Entry (AJE) just as is the StrUpPgm; see Add Auto Start Job Entry (ADDAJE). Be aware that the Controlling Subsystem [as determined by the system value] (QCTLSBSD) could change; the QSTRUPPGM is started for at least both QCTL and QBASE with the QSTRUPJD Job\Job Description (JOBD).

But for that scenario, as described, consider [and I think the idiom from which these name-references are derived, is sufficiently international]: the JOB(CHICKEN) and the JOB(EGG), *or is that* the JOB(EGG) and the JOB(CHICKEN)?

This morning for some IPL problem (that I have to understand) this
jobs are not started and so I have had problem with orders import.

My first approach would be to investigate and understand the origin of that specific failure [not yet communicated to readers of this topic; e.g. the error message identifiers\context], and code\configure to prevent that and any similar potential problems; i.e. put off pursuing /monitoring/, esp. if the processing had long been starting and running without difficulties until this particular incident.

So I would find a system job (that I'm sure that run always ) that I
could use as monitor for my IMP_ORD job.. I hope I shall be clear

IIRC the system job QALERT [while unable to start or actively monitor the two jobs] does supply a means to inform if the job fails while starting or ends [should also be able to establish a means to inform of a condition logged by program that requests but fails to start the jobs. See GO CMDALR for the Alert (ALR) related commands; and IIRC there was a separate manual for alerts.

For monitoring, review the Watch feature; see GO CMDWCH for Watch (WCH) related commands. There are surely more creative means to utilize those than just a specific Message Identifier (MSGID) which might be too limited; e.g. a Watched message queue (WCHMSGQ) might utilize another Notify (NTFY) feature that sends a message to a queue. This feature should be able to /watch/ for the same messages that alerts could be defined to handle; while alerts may have the ability to notify externally via SNMP traps, AIUI the watches will require a job on the server to react.