Error ANR0162

[filthyPierre]

Hi, recently upgraded an AIX-based TSM server from 5.5 to 6.2. Have started getting these errors:

ANR0162W Supplemental database diagnostic information: -1:HY000:-1042 ([IBM][CLI Driver][DB2/AIX64] SQL1042C An unexpected system error occurred. SQLSTATE=58004).

They have started after we created a disk storage pool with a FILE type devclass for de-duplicating. IBM diagnostic info is vague for this error, but says to look at the actlog entry immediately before this error, which is probably the real cause of the problem. However, immediately before these errors is not an error, simply an "I" message:

ANR1176I Moving data for collocation set 1 of 1 on volume /tsmdpool/filepool/00000004.BFS. (PROCESS: 7872)

No other errors around those times.

Anybody seen this?

Thanks in advance,

FP

 

[moon-buddy]

I suggest to go back to where you were and see if the errors had disappeared. Version 6.2 is bleeding edge that I won't probably touch for a sometime.

 

[filthyPierre]

Thanks, but that's easier said than done... a new LPAR was created and network migration done.. since then the old LPAR has been nuked and storage re-provisioned. It has been running for nearly two months, without problems, seems coincidental that it's started happening after we have started de-duplication...

FP

 

[cheffern]

from what I've been hearing, de-dup and caching both seem to be an issue on the 6.x versions.
I would try setting de-dup to NO, move data (if you can) -
and going forward see if this helps.

the "I" message you noted looks perfectly normal to me -
I see that when I move data from a disk pool to tape -
it's purely information.
but the "W" is no good...

-Chef.

ps- I'd revisit this later after IBM works the kinks out of 6.2

 

[filthyPierre]

Thanks, yes I have turned off dedup for that pool. No more errors since but I am still not sure if that was the cause.

Other weirdness - after deduping the pool and then creating copypool copies ... tried to run reclamation on the pool to remove the duplicates. It refused to reclaim, saying there was no data to process. Found a recent web-presentation from IBM and in there was a command "show deduppending", which I ran on the pool. There was around 173 GB of data in the pool and "show deduppending filepool" said there was about 39 GB of duplicates. But could not get reclaim to remove it - all volumes showed 100% utilised, i.e. no percentage of reclaimable space, which I guess accounts for the "no data to process".... maybe we are supposed to run expiry after dedup, then reclaim? The manuals do not mention expiry though... all up a frustrating waste of time...

So, as you suggest Chef, may revisit this after we've had a few fix releases.

Thanks,

FP