Package Deployment when Subsystems launched from Scheduler

Has anyone seen this? Client is at Xe SP20 XU6:

There are a number of subsystem jobs launched after the backup on a daily basis from the OneWorld scheduler.

Subsequent attempts to deploy packages to the server result in the scheduler kernal blocking the package deployment kernal, i.e. it doesn't respond to the request to release locks. The package deployment goes into SEMW status and the whole system basically freezes from then on.

Our current feeling is that because the scheduler tracks the jobs it launches, and the subsystems never end, the scheduler kernal can never release its locks and therefore block the package.

Has anyone deployed server packages whilst subsystem jobs are active, either called from the command line, the application or the scheduler? I'm sure it works if run from the command line or application but can't put my finger on exactly when I have seen it happening.

Thanks

Paul,

I believe that to deploy a package to the enterprise server, NO batch jobs can be running. I think this is because a package will replace programs/files that are used to run batch jobs.

Also, I'm not familiar with AS400 and subsystems, so I don't know if there are any platform-specific issues here.

Don, you are bang on the money with your answer. Regardless of platform, all
batch jobs have to held while a deployment to the ES is in progress for the
reason you give.


regards
Sid Perkins
Independent IT Consultant
[email protected]

----- Original Message -----
From: "DSauve" <[email protected]>
To: <[email protected]>
Sent: Thursday, February 27, 2003 3:58 PM
Subject: Re: Package Deployment when Subsystems launched from Scheduler


jobs can be running. I think this is because a package will replace
programs/files that are used to run batch jobs.Also, I'm not familiar with
AS400 and subsystems, so I don't know if there are any platform-specific
issues here.

Hi PAul,

This is a known issue. I think it is fix in SP 20_N1.

I have two clients:

One just upgrades to SP20_N1 and the problem went away.

The other is on SP19_D1 and the CNC guy stops and restarts the services after deploying every package to the enterprise server. It is a pain but this guarantees the scheduler works properly.

Paul,

I don't believe this to be a bug. It's standard behaviour and makes perfert sense. You don't want to update specs of a UBE that is currently running. The results would probably be pretty nasty... I believe that this would lead to spec corruption. If you are installing a package to the server, you definitely don't want any jobs running be it by users or the scheduler.

Jim.

Subsystem jobs are a bit different, as they mostly just sit there in a TIMW status. It would therefore not be unfair for the package deployment to request those to hold while the deployment takes place, after all the deployment itself waits for any jobs to complete and holds all others submitted after it.

I'm going to have a look at SP20_N1 to see if I can identify SAR in question, and I think the problem only started after two seperate events:

1) A Package deployment was aborted

2) The subsystems were started from the scheduler.

Cheers

Subsystem jobs can be running whilst deploying packages, user submitted jobs cannot. whilst you deploy a package to the server it allows batch jobs to be submitted, but they do not run until the deployment is completed. I have succssfully deployed full and partial packages on AS/400 platforms without problems even with subsystem running.

The platform in question was AS/400, SP17.1, Xe