Datastage Jobs Not Running

priyadarshikunal · Post by **priyadarshikunal** » Thu Aug 12, 2010 3:56 am

do some load balancing and don't overload the server. You can also ask IBM to get a patch for increasing the timeout defined for server. Keep the &PH& folder clean and don't let logfiles grow too much.

paranoid · Post by **paranoid** » Thu Aug 12, 2010 4:56 am

Hi,

Thanks for the reply. I would contact IBM on this as you suggested.
Does clearing DS logs in the director clears the files in &PH& automatically? Or the files in that directory needs to be removed manually? Sorry for this stupid question.

Sue

Sainath.Srinivasan · Post by **Sainath.Srinivasan** » Thu Aug 12, 2010 5:03 am

Removing or purging DS logs is different from &PH&.

Logs are stored in RT_LOG hashed files and are related to the specific job's logs.

&PH& stores the phantom process history - more of kernel level. This is one central area where your process communication occurs.

To clear, log into DS Admin and select the project you wish to clear and run the command

Code: Select all

CLEAR.FILE &PH&

Clearing &PH& will assist DataStage to 'turnaround' better.

chulett · Post by **chulett** » Thu Aug 12, 2010 6:31 am

Most people set up a cron script to prune that directory, i.e. delete files older than x days to keep a rolling x days in the phantom directory. It's important to note that CLEAR.FILE is like a truncate so if you take that route you should only do that when no jobs are running.

And the solution to your -14 error is to stop overloading your server. Sometimes that's not about how many are running but rather it means don't attempt to start so many at the same time - stagger out their start a little bit. While there (allegedly) is a "patch" for this to increase the timeout value, it's not something IBM just gives anyone that asks, you have to make a pretty compelling business case for this from what I understand.

vivekgadwal · Post by **vivekgadwal** » Thu Aug 12, 2010 7:22 am

paranoid wrote:The jobs failed today as well and finally i could find the error code when running manually on the server. It says "Status code = -14 DSJE_TIMEOUT".

Sue,

This was exactly our problem too. Are the jobs getting fired at all later? I mean, if you try to run them manually, do they run? Or they just keep quiet, not logging anything and stay as they were before?

priyadarshikunal · Post by **priyadarshikunal** » Thu Aug 12, 2010 9:00 am

paranoid wrote: I would contact IBM on this as you suggested.

I would first suggest to clean &PH& folder and old job logs then check it. Also monitor the resource usage after you trigger the jobs, itf its taking more than 90-95% CPU , try to reduce the number of jobs running concurrently else including other error messages, performance will be hit. IMO

paranoid · Post by **paranoid** » Fri Aug 13, 2010 4:04 am

Thanks all. Can i clear all of the files in &PH& folder? Will there be any issue with the existing jobs if i delete all the files in it?

If we need to clear that folder, should we need to make sure that no job is running at that time?

@ Vivek -- Yes after some time when we ran them manually, they are running fine.

Sue

Sainath.Srinivasan · Post by **Sainath.Srinivasan** » Fri Aug 13, 2010 4:09 am

Yes - you can clear all files from &PH&. The prescribed method is to clear from TCL.

Yes - you must ensure that none of your jobs in that project are running when you do this.

paranoid · Post by **paranoid** » Fri Aug 13, 2010 4:11 am

Thank you sai. Currently the size of the &PH& folder is 263 MB.
Can i use the command given by you in DS admin to clear the files?

Sainath.Srinivasan · Post by **Sainath.Srinivasan** » Fri Aug 13, 2010 4:13 am

paranoid · Post by **paranoid** » Mon Aug 16, 2010 12:43 am

Hi,

Today the jobs are failing again, however got the message 'Unable to lock project' in the PH folder logs. This error occurs exactly at 2 am EST and continues till 4 am EST. Any ideas on why this is happening?

Thanks,
Sue

chulett · Post by **chulett** » Mon Aug 16, 2010 6:30 am

You sure it's not "Project is currently locked"? Couldn't find your message anywhere in the forums but this one I've seen myself.

ray.wurlod · Post by **ray.wurlod** » Mon Aug 16, 2010 4:50 pm

Do "they" quiesce the database (or similar) between 02:00 and 04:00 for system maintenance such as backups?

paranoid · Post by **paranoid** » Tue Aug 17, 2010 2:07 am

Hi everyone,

We have finally reseolved this issue after a struggle of 10 days or so..

After a DS restart and clearing the files in PH folder we could resolve this issue. No more failures today

As per IBM Support, the number of files in the PH folder should not exceed more than 1000. Since we have more than 3000 files, he advised us to clear the files in PH folder. The same suggestion is given by DS experts in this post as well.

Before clearing the folder we have restarted the DS server as well.

I am very much thankful to each and everyone who replied to this post.

Have a nice day!!

Sue :D