Dirty Page in Buffer Pool

[mikeatkc]

After TSM ran the full db backup my recovery log does not clear to zero. Additionally, I'm getting the below messages when doing "show logpinned." Do I need/want do a dumpdb to correct the problem or just auditdb?

The problem occurs every so often, after an incremental db backup or another full db backup the recovery log clears to zero.

//error messages

Dirty page Lsn=8924142.85.2090, Last DB backup Lsn=8930021.131.2768, Transaction table Lsn=8930204.192.2278, Running DB backup Lsn=0.0.0, Log truncation Lsn=8924142.85.2090

Lsn=8924142.85.2090, Owner=DB, Length=56 Type=Update, Flags=82, Action=SetIcb, Page=14926848, Tsn=0:858751236, PrevLsn=8924142.85.1885, UndoNextLsn=0.0.0, UpdtLsn=0.0.0 ===> Bit Offset = 526

The recovery log is pinned by a dirty page in the data base buffer pool. Check the buffer pool statistics. If the associated transaction is still active then more information will be displayed about that transaction.

Database buffer pool global variables:

CkptId=33454, NumClean=50030, MinClean=127904, NumTempClean=128000, MinTempClean=63952, BufPoolSize=128000, BufDescCount=139059, BufDescMaxFree=140800, DpTableSize=192000, DpCount=77968, DpDirty=77968, DpCkptId=30473, DpCursor=14374, NumEmergency=0 CumEmergency=0, MaxEmergency=0. BuffersXlatched=0, xLatchesStopped=False, FullFlushWaiting=False.

//end

 

[heada]

Do you have any client activity when this occurs? A client performing a long operation might cause this. Some TDP processes can pin the log for hours it seems.

-Aaron

 

[mikeatkc]

Yes, there's one client (mail server) that recently took all day, almost continuous backup of its file system (and presently continues to do so). Do I need to adjust this node's backup and not worry about the dumbdb or auditdb?

 

[heada]

There is an undocumented show logpinned command that should show if this is the client that is keeping a log buffer locked (pinned) while the dbbackup is running. I would lean to that being the cause and not a problem with the TSM DB. Try the DBBackup without the client running and see if you get the same error. If so, call IBM and open a PMR.

-Aaron