Scratch Cleanup

[dougcl]

Hi folks, I noticed that we are getting a growing number of old files in the scratch area. Here's an example:

Code: Select all

-rw-------   1 dsadm    dsdev    2227442 Jul 12 16:49 etleng2node1.0000000000000014.00000000

Shouldn't the scratch area be clear after all the jobs are finished? Perhaps we are accidentally storing performance data there or something.

Thanks,
Doug

[anbu]

The files created in Scratch disk will be there even after the job finishes. You have to delete manually or create some script to handle those files

[ray.wurlod]

I disagree. If jobs finish OK then files on scratchdisk should be cleaned up automatically. If jobs abort then the files may be left there, and would need to be cleaned up manually. This is one of many reasons why I try to create jobs that don't abort.

[anbu]

I got confused resource disk with scratch disk.

[dougcl]

Awesome. Thanks Ray. I suppose creating jobs that don't abort is an uncontroversial "best practice"

[ray.wurlod]

I don't know. Many people here ask questions about how to force jobs to abort. To me that's anathema. So it's my personal best practice to create jobs that don't abort, even though that may require more thought than just cranking out the code.

Deadlines? Pah! You will spend less time doing it right the first time than having to do it a second or third time.

And doing it right includes documentation!

[chulett]

There's nothing about Ray that is "uncontroversial", including his best practices. And some DataStage features require jobs to abort to work.