DSJE_TIMEOUT

ArndW · Post by **ArndW** » Thu Feb 28, 2008 8:42 am

Did you notice if the machine was slower than usual when this happened? The timeout is often seen on heavily overloaded boxes, so the message isn't surprising. But if you have never seen this before and haven't changed anything significant on the machine it needs to be investigated.

chulett · Post by **chulett** » Thu Feb 28, 2008 9:07 am

30 minutes?

dspxlearn · Post by **dspxlearn** » Thu Feb 28, 2008 9:54 am

Thank you ArndW,

We have reported this to out UNIX server admin.

throbinson · Post by **throbinson** » Thu Feb 28, 2008 11:56 am

How many jobs were attempting to start/run concurrently?
Check you uvconfig settings.

With the DataStage default configuration parameters, you
normally can run 50 to 60 jobs concurrently. After setting GLTABSZ = 120,
RLTABSZ = 120, MAXRLOCK = 119, you can run about 100 jobs concurrently. So the number of DataStage jobs that can be run concurrently is determined by the DataStage lock table.

Setting those values to their default or too low can result in the DSJE_TIMEOUT
you are experiencing.

dspxlearn · Post by **dspxlearn** » Thu Feb 28, 2008 2:59 pm

Hi throbinson,

Our default configuration settings(uvconfig) are as below:
GLTABSZ =120
RLTABSZ =220
MAXRLOCK =219

We had 80 to 90 jobs started at the same time (04:00 AM) out of which 32 failed. Out of these jobs few were running in multiple instances.

dspxlearn · Post by **dspxlearn** » Thu Feb 28, 2008 3:06 pm

By searching the forum, i found that many people suggest that increasing these parameters will not help much.

Anyways, we raised this to our admin. I will post their solution once i hear from them.

fridge · Post by **fridge** » Sat Mar 01, 2008 4:00 pm

this is caused by a hardwired timeout limit of 60 seconds in which the job has to kick in - and this can be missed if lots of jobs kick in at the same time, the main problem is each job has to open universe files for logs, config , status etc - this can be tweaked via uvconfig - but only two cetain limits.

if u have no luck with UNIX guys -there is a patch that incorparates a way of changing the default time out from 60 secs - speak to your datastage support provider