Information Server Dataset Directory

Raftsman · Post by **Raftsman** » Tue Jul 26, 2011 9:40 am

Our jobs started to abort this morning due to space issues. After review, we found many files existed under IBM\InformationServer\Server\Datasets. I know that under the Project directory, there is also a directory called Datasets. It doesn't have anywhere near the same amount of datasets.

My question is, can we just purge the InformationServer\Server\Dataset directory. It looks like fragment of Lookups and aborted jobs. Will it corrupt future running jobs.

I opened a PMR just the same but I thought maybe someone here has come across the same problem.

Thanks in advance

jwiles · Post by **jwiles** » Tue Jul 26, 2011 12:10 pm

No, don't just purge the Datasets directory. This is where the segment files that contain the data for your parallel datasets are located. If those files disappear, you will likely see jobs begin to fail because datasets are corrupted (missing data segment files). The path you gave is the default (at installation) location for this directory and is still actively used if you have not changed the "resource disk" entries in your configuration (.apt) files.

If you are still using the default location, and the drive/mount point which ran out of space is where this is located, you might consider creating a new storage location (separate, larger mount point) for the Datasets directory and migrating your configuration files to begin using it.

Regards,

Raftsman · Post by **Raftsman** » Tue Jul 26, 2011 12:36 pm

It's funny because there are ton of Lookuptable files that were left around from previous jobs. We couldn't get ORCHADMIN RM command to work and tried other way. None successful. So since it's a development box, we deleted all files. I am current testing existing jobs that use datasets. So far so good. IBM provided documentation but we couldn't get it to work.

I'll update everyone shortly on the testing progress.

Raftsman · Post by **Raftsman** » Thu Jul 28, 2011 8:25 am

We deleted the files in the Server\Dataset directory and tested a few sequence run. Everything seems to work as it did before.

ray.wurlod · Post by **ray.wurlod** » Thu Jul 28, 2011 4:44 pm

You probably have a heap of orphaned *.ds and/or *.fs files hanging around now.