Batch jobs suddenly taking forever to run

Has anyone come across a problem on their application where batch jobs all of a sudden are taking forever to complete.

The database group over the weekend was doing some tweaking with database switches. Since we have noticed a huge slowdown in batch job processing.

They supposedly switched everything back but it does not seem to have resolved the problem.

We cannot explain what seems to be happening. It is happening with our custom batch reports and some larger JDE canned reports.

Does anyone have any idea what we should be looking for. We are at a loss on how to resolve the problem. We have been cancelling jobs, holding jobs and switching job priorities.

Any help would be much appreciated.

Cathy,

the possibilities are endless. However, the fact that they changed database parameters on a weekend implies that some of the changes were those that required a database shutdown and startup.

After they "supposedly switched everything back ", was the database shutdown and restarted again?

Perhaps they switched from cost-based optimization to rules based optimization. That would explain a massive slow-down on some queries. Even if they switched back the statistics probably now need to be regenerated/computed.

I trust that the database group is assuming full responsibility for this debacle?

Good Luck!

It is not good to start laying blame on anyone. There are all sorts of changes that can be made to a system and in Information Technology you never know when a change in the DB switches may affect an application (I don't work in DB area I am an programmer who writes custom code).

We are assuming that this is the cause. We have since noticed that any batch jobs selecting specific data runs very fast but jobs that select ranges of data seems to be taking 10x or more longer to run than before. Some large jobs it never seems to come back from the SQL call.

Please ask your database guys to generate the indexes for all the tables. This will fix your issue. If you would like take one UBE and generate the indexes for the tables it is using.

Hope it helps.
DD
ERP Application Manager

Turned out that we had some jobs running that generated huge job logs. We had to remove all the job logs to clear up some space because after the system was brought down it would not come back up because there was not enough available space.

We then brought the system down then back up then everything was okay.

Before we did this it got to the point where all the jobs in the system were processing then ending in error.

We are now trying to figure out what is causing all these job logs to be generated. Some are as large as 2 gig per run.

Interesting! Did someone turn on logging in a scheduled job (via the JDE scheduler), or did someone turn on debugging in the enterprise server's JDE.INI file? Just a couple suggestions to check. Good luck sleuthing.

It wasn't the DB Group that caused this. The slowdown was caused because tracing was turned on for the batch jobs (and some of the jobs produced A LOT of trace messages). Tracing was turned on when trying to debug problems. We turned tracing on using the JDE monitoring tool, thinking it would be reset when we bounced the system. I guess this stayed on, affecting our batch jobs. Some of the batch jobs running produced so many trace messages that it filled up the log file for that job (2 GB max file size). Then, the whole temp directory filled up with all these log files, causing everything to start bombing. Once we deleted all the log files (to free up the temp directory) and the trace flag was turned off the jobs started running normally.

Hope this helps someone else resolve problems.