Certain tables return "The Report Server is not responding" [Solved]

[DrCandycane]

The issue I am describing just started. The only relevant changes are a TEST instance of 10.2.2 was installed on a separate server and we have slowly begun using the Lifecycle manager tool in anticipation of an upgrade from 10.1.1 (where the problem is present) to 10.2.2. The affected reports are run ad-hoc and I am not sure if there is a correlation or if the timing of the issue matches with the timing of the new TEST instance / LifeCycle manager, but I suspect it is involved somehow.

We have a framework/package published with about 30 or so tables in it. One of the tables (Table A, ~10,000 rows, 8 columns, so not a large table) returns a "DPR-ERR-2056", "The Report Server is not responding." error when queried in Report Studio or run from Cognos Connection.,  All of the other tables in the package are fine. Just today, I found a second table (B), in a different framework / package that is having the same problem.

The tables query just fine in SQL, they also validate and test in Framework manager just fine. Additionally, if I select the option to run the reports they are used in to my e-mail (so not interactively, but running in the background) they will also work fine, it seems the problem is limited to running them interactively from Cognos connection or Report studio. Other tables from the same data source in SQL work just fine whether interactively or in the background, no issues.


Steps I've tried:
- Restarted Cognos Configuration (PROD 10.1.1)
- Restarted SQL Server
- Created a new framework / package with only the first table, table A present. Same error.
- Removed table A from it's original framework completely, re-added it. Same error.
- Queried table A from report studio on our DEV 10.2.2, this works fine!

I'd greatly appreciate any input or suggestions you may have on this issue.

Thank you!

[bdbits]

This may be obvious to you, but I have to start somewhere, so...

Are you using the 10.1.1 version of FM with the 10.1.1 instance, and 10.2.2 FM with 10.2.2 test? It is very important that the FM version matches the Cognos version. I would also advise that the packages on 10.2.2 be re-published with that version of FM. And if you have published to 10.1.1 with FM 10.2.2, re-publish out of 10.1.1.

Have you checked the Cognos server error logs? There might be something in there that is not making it back to client apps, so I would check there also.

Also, googling on the error it appears it is often related to disk errors, e.g. a lack of space. I know you said the table was only 10k rows, but I would check anyway and if it is fluctuating a lot then produce the error while monitoring. Seems unlikely to be the problem I know, but stranger things have happened to me with various software (not just Cognos).

Edit: I thought of one other thing. When switching between versions, always be sure to clear your browser cookies and cache. That can cause some really strange problems if you do not.

[DrCandycane]

Thank you for the response, the correct versions of framework were used in each case and disk space wasn't an issue. This one has been 'solved'. It looks like Lifecycle manager got stuck at some point in reading from the Source, 10.1.1 and crashed the instance. I looked further into this and figured I'd strip out any of the Lifecycle remnants in 10.1.1. Lifecycle had not been in use since I noticed the problem with our PROD instance as that took priority over the upgrade.

http://www-01.ibm.com/support/docview.wss?uid=swg21408219

See above link for instructions on how to remove logging changes to your Source in Lifecycle. This fixed the issues we were having with the two faulty tables in 10.1.1.