A project I am on uses EMC Legato to backup DB2 database using the appropriate agents. It frequently complains when it kicks off two parallel backups from the same instance when one of the backups needs a lock on a shared tablespace which more often than not is "syscatspace". Escalatation to IBM didn't get anywhere and we are finding it hard to schedule the work off-peak.

Has anyone out there experienced this or anything similar?

What is the point of two parallel backups?

Just to make myself clear, we are talking about the concurrent backup of two or more databases in the same instance. One point is to save time. Another point is to not have to bother about backup scheduling. Legato and other backup managers can do things in parallel.

Sounds more like an locking issue from the Application.

It is a locking error - in DB2. Had it been the Legato software, we would have seen the error in Legato's logs but the error is reported in db2diag.log. The tablespace has been marked as BACKUP_IN_PROGRESS and that makes it exclusive from a backup perspective.

Sorry, but your explanation does not make much sense. No tablespace is ever shared between databases, so backups of two separate databases can run simultaneously without affecting each other. Each tablespace is placed into "backup in progress" state as it is being backed up, which is not an error.

I think you need to make sure that multiple Legato agents do not attempt to back up the same database at the same time. If you see any particular error in db2diag.log you should probably post it here.

[removed duplicate post]

If absolutely no tablespace is shared, then where does syscatspace sit? Does every database have a syscatspace? You can assume that no more than one backup of a database is set off at any given time.

As requested, a snippet from db2diag.log

2010-01-28-03.00.08.553634+060 E9180A740 LEVEL: Error
PID : 1998902 TID : 39529 PROC : db2sysc 0
INSTANCE: ptlinst1 NODE : 000 DB : RELDB61
APPHDL : 0-12405 APPID: 10.58.45.97.38010.100128020007
AUTHID : DB2INST1
EDUID : 39529 EDUNAME: db2agent (RELDB61) 0
FUNCTION: DB2 UDB, database utilities, sqlubBuildAppTbsp, probe:1653
MESSAGE : SQL2048N An error occurred while accessing object "". Reason code:
"".
DATA #1 : String, 166 bytes
The BACKUP_IN_PROGRESS state is already on for this tablespace.
This tablespace is already in the midst of being backed up and
cannot have a concurrent backup on it.
DATA #2 : String, 11 bytes
SYSCATSPACE

Each Database has a SYSCATSPACE. It's the System Catalog of the database, where the System-tables reside. The ID of the tablespace is always 0.

db2 list tablespaces show detail

db2 list tablespace containers for <ID> (0 for syscatspace)

Does DB2 perform Auto-Backups?
Is AUTO_DB_BACKUP set to on?
(db2 get db cfg for RELDB61 | grep AUTO_DB_BACKUP)

I think this tells you all you need to know.

The error in db2diag.log that you posted obviously contradicts this statement.

Nick, thanks for setting me right on the syscatspace issue. Wrong assumption on my part and, as you said, all indications are that we've got two or more online backups of the same database occurring at the same time.

I have been thrown off by the fact that IBM say I cannot do a parallel backup of six databases in the one instances. Quote

But the real problem here is that you are trying to execute online backups for all 6 databases at same time, and this is not compatible.

Unquote.

I had a PMR out on this so it should have been an official response. We're now saying that response is false. My question went out to the forum more as a feeler to see if anybody else out there might find it convenient to backup multiple databases at once (no doubt it is) with a view to push for a feature request as opposed to fixing up some mistake which I may have done.

I'll assume the latter and go over my steps again.

Not sure what they meant by "not compatible", it's difficult to tell outside the context. Technically speaking, there is nothing that prevents you from backing up multiple databases simultaneously. However, there's something to keep in mind:
- all utilities, including backup, share the same memory pool, which should have sufficient size to allow simultaneous backups;
- if the vendor library is used to access backup media (which I suspect may be the case with Legato), make sure it supports simultaneous operations, especially if the backup images go to different media and/or storage classes;

Could it be that your db is cataloged as some alias name and your backup job is getting the db name from "list db directory" and trying to backup the same database in parallel.

Here is an example of me reproducing this error:

bkogan@panipuri /home/bkogan > db2 catalog db sample as sample2
DB20000I The CATALOG DATABASE command completed successfully.
DB21056W Directory changes may not be effective until the directory cache is
refreshed.

bkogan@panipuri /home/bkogan > db2 terminate
DB20000I The TERMINATE command completed successfully.


From one session:

bkogan@panipuri /home/bkogan > db2 backup db sample online
Backup successful. The timestamp for this backup image is : 20100129163547


From another session:
bkogan@panipuri /home/bkogan > db2 backup db sample2 online
SQL2048N An error occurred while accessing object "0". Reason code: "6".



2010-01-29-16.35.46.025030-300 E9772A731 LEVEL: Error
PID : 884820 TID : 3667 PROC : db2sysc
INSTANCE: bkogan NODE : 000 DB : SAMPLE
APPHDL : 0-660 APPID: *LOCAL.bkogan.100129213539
AUTHID : BKOGAN
EDUID : 3667 EDUNAME: db2agent (SAMPLE2)
FUNCTION: DB2 UDB, database utilities, sqlubBuildAppTbsp, probe:1653
MESSAGE : SQL2048N An error occurred while accessing object "". Reason code:
"".
DATA #1 : String, 166 bytes
The BACKUP_IN_PROGRESS state is already on for this tablespace.
This tablespace is already in the midst of being backed up and
cannot have a concurrent backup on it.
DATA #2 : String, 11 bytes
SYSCATSPACE

Hi db2girl. You hit the nail on the head. We've got the same databases on two instances and aliases for the db2inst1 user in both environments were pointing to one. Nothing to do with shared tablespaces in the one instance. Just a dud setup. Thanks you Nick, Cougar for your input.