db2 version 8 fix 7 on Solaris 5.8

Our online backup failed again and again without any clue.

The error message is:
SQL0911N The current transaction has been rolled back because of a deadlock
or timeout. Reason code "68". SQLSTATE=40001
SQL0843N The server name does not specify an existing connection.
SQLSTATE=08003

I checked, but there was no deadlock, no lock-wait to force application.
Is there any clues cause online backup corrupted. Disk space is ok for backup.

Please help me in urgent.

Thanks much

How and when did you check this one ?
"no lock-wait to force application."

Can you post more details ?

FYI, an online backup fails when the catalog is in an inconsistent state .. For eg, a LOAD or RUNSTATS or REORG is in progress, a create/drop object is in progress or has not been committed, etc.

Sathyaram

Lock-wait and deadlock was monitored a few minutes after failing because I was watching for this.

We did not run reorg, runstat and loading. How can I check if the query is not committed yet. or the loading is going on?

Thanks much

Could you guide me how to find the timestamp for incremental backup?
If my full ONLINE Backup is failed. How much database I can restore.
There is no base line for incremental backup. How can I still take the backup.

Please help

Try
db2 list utilities show detail

What does that show?

db2 list utilities show detail


show that
ID = 1109
Type = RUNSTATS
Database Name = WBVDB
Partition Number = 0
Description = WBV.FACTOR
Start Time = 07/24/2006 15:49:23.828430
Throttling:
Priority = Unthrottled

After making sure no utilities are running(you can kill them) run the backup; you have to look in the diaglog as to why db2 backup is failing. If it fails, tell us what command you used and what the diaglog says at that time.

Could you please explain to me during the time utilities like reorg, runstats, load running, the online backup failed. ONLINE backup need locks like runstats load ... This would help me much.

The very first step in the online backup is to take a backup of the catalog tablespace
and this copy of the catalog is required to be a consistent one, For this DB2 takes table level share locks (not sure if it is S locks or some internal lock simialr to S)
When a runstats, load or reorg is running, an update lock is taken the rows of catalog tables that reference the user table on which the utilities are running.
As you can see, the above two are conflicting locks.
Therefore, if a backup starts when utilities are running, the backup waits for locktimeout seconds to obtain locks on the catalog. If it is unable to, then just like any other application, the backup fails.
Remember, that if a backup has finished copying catalog tablespace, then utilites can start on tables and both of them can contine in paralllel. In this case, just like for any other table, chnages to the catalog tables is 'tracked' using the transaction logs.

Hope this answers your question

Oh BTW, IIRC, there are lockking issues with backup, when lobs are stored in SMS tablespaces as well

Sathyaram