Incremental backup scan takes forever

[kvwar]

Hi

I am trying to perform incremental backup of a 700 GB volume. The volume is mapped as CIFS share on a Windows volume. The fullback took me 48 hours.
After a 1% change of data I fired an incremental backup. This is not a scheduled backup. It is a manual backup. There are close to 4 million files in this volume. even after 24 hours the scan is still going through. It hasnt event started the backup. I am relatively new to TSM. Following are my questions:

1. Is it normal for a CIFS incremental of the above dataset to take such a long time?
2. Which log will give me information of what folder or file is being scanned by the client. I am not sure what it is doing.

Please help.

 

[Sunhillow]

Hi

we have several CIFS shares holding more than 4 million files. They are backed up every night via a Win2008 Server. Scan + Backup runs for 6 to 8 hours.

Do you have a WAN connection between Server and Filesystem? This would make it slow. Another reason might be your backup server has not enough RAM. The backup process allocates at least 1.5 GB with such a lot of files. There is an option to consume less memory (memoryefficientbackup yes), but this won't make things faster.

AFAIK the tsm client does not log the directories it is scanning, only the amount of scanned files in an increment of 500. The verbose option is on by default

 

[kvwar]

The RAM is 2 GB. I do not see any activity whatsoever on my storage. Server and the CIFS storage are in a LAN.When you say that the file count increments by 500 where can I find that increasing. I have fired the backup using the client GUI. Is there something that would tell me what is going on right now with regards the backup.

 

[Sunhillow]

By default the schedlog file is in the directory %programfiles%\tivoli/tsm\baclient. But I don't think a manual backup writes to a logfile.

If you use the commandline client, the log output is printed to the screen or you can redirect it to a logfile. The logfile itself then can be shown using the tail command.

 

[JohanW]

.. There are close to 4 million files in this volume ..

1. Is it normal for a CIFS incremental of the above dataset to take such a long time?

Yes.

Have a look at journalling.

 

[kvwar]

Is Journaling supported on a Windows 2003 Server Enterprise 32 bit

 

[JohanW]

That's where I'm using it .. but you might have a look at the IBM website here.

 

[Sunhillow]

Journaling only works on local filesystems, not on a CIFS share

 

[kvwar]

Thanks a lot Johan. I have run a full backup using the option Incremental. It backed up 700 GB of Data. Can I now fire an incremental straight away with Journal option or am I supposed to run a full again with Journal option. Sorry to bug you with so many queries.

 

[JohanW]

Journalling needs one backup without a journal to initialise itself.

Sorry ..