Reasons for Tsm DB corruption

[prabhav]

Hi

I am using TSM 5.4 on Win 2k3 server. Recently we had a problem and the TSM services were unable to start. We logged a case with IBM and they informed that the TSM DB has corrupted and neds to be restored. We hav restored the DB successfully.

I wanted to know what had caused the TSM DB to corrupt. However, Ibm informed that it is not possible to dtermine this as the DB has been restored.

Can anyone please tell me what might be the reasons for the DB corruption. Not necessary only in this case but in general.

thanks in advance.

following is the error that I was recieving:

ANR0306I Recovery log volume mount in progress.
ANR9999D_0506045865 (dballoc.c:1378) Thread<0>: Zero bit count mismatch
for
SMP
page addr 197632; Zero Bits = 0, HeaderZeroBits = 43.
ANR9999D Thread<0> issued message 9999 from:
ANR0361E Database initialization failed: error initializing database
page
allocator.

 

[LED888]

Any issues in the Disk Subsystem, the Disk Controllers or the Communication Channels or even the Disk Drivers can cause databases to become corrupt.

Kernel Drivers and softwares like Antivirus, that have the ability to take Kernel Level Handle on files, are also known to cause database corruption issues.
If the TSM Server is hosted on a Unix system, more than likely this would not be the cause.

- Hard disk error \ failure caused due to hard drive crash, heat, water, natural disasters, and more.
- Memory error
- Network error
- Power glitch (power loss, dip, or surge)
- File system corruption

User actions like these can corrupt a database:
- Rebooting or switching off the system when the TSM Server is writing data.
- System crashing (freezing) when the TSM Server is writing data.


Co-worker :
Too many queries, reports via Operational Reporter /Server Graph.
The Oeperational Reporter and Server Graph, just run select statments I have never encounter this causing the database to be corrupt. Performane issues, yes but corrupt database.

Good Luck,
Sias

 

[LED888]

I seem to have forgotten on.

Doing a kill -9 on the dsmerv process while data are being written.

Not sure if this would be consider a possible cause to corrupt the database.
Having the database on a raw file system, then increasing the size of the raw file system.

Good Luck,
Sias

 

[prabhav]

Thanx for the info LED888

I was wondering if the DB corruption can occur if expire inventory runs while backups are running

Also, my db volumes are of 20GB each. (3 volumes - Total DB size is 60GB). I have read that the recommended value is 2GB for each volume. Can this be a problem?

 

[LED888]

I know that your really don't want to run expiration at the same time as migration, or reclamation.

I would not expect the database to be corrupted when we are performing an exp inv while clients are performing there schedule backups.

As in term of size of the database.
Again I would not expect the database to be corrupted just becuase we have 20GB, or even 40GB, volumes. If we have various sizes of the dbvolume may expect some performance issues. I would much rather have 3-20GB dbvolumes than 30-2GB dbvolumes.

Good Luck,
Sias

 

[wim]

The reason why you don't want to run expiration at the same time as migration, or reclamation is for performance reason not because of possible corruption.
For performance reason it is recommended to have about 8 volumes for the database.

Regards. Wim