fbackup file warnings

[Randy Hagedorn]

Hi,

I occasionally get warnings during our full backup with fbackup that a file (not always the same file) was not backed-up.

fbackup(3009): WARNING: File number 142 (/u03/oradata/ORD/user03.dbf)
was not successfully backed up

Can any one give clues as to how I can find out why the file wasn't backed-up?

Thanks in advance.
Dewy

[Sanjay_6]

Hi,

The file was probably in use by the database / process.

Hope this helps.

Regds

[A. Clay Stephenson]

That usually means that fbackup has tried the maximum number of retries on an active file and is giving up. It means that fbackup has detected modifications to this file during the backup process. By the way, if you are trying to backup Oracle without shutting down the database (cold backup) or putting the database files into backup mode for a hotbackup then you have a worthless backup regardless of whether you get these fbackup warnings.

[Paula J Frazer-Campbell]

hi

As Sanjay has said tje file was in use and fbackup had exceeded the maxretries and then ignored the file .

You can increase maxretries in:-

/etc/sam/br/fbackup_config

blocksperrecord 32
records 32
checkpointfreq 32
readerprocesses 2
maxretries 10
retrylimit 5000000
maxvoluses 100

If it is at default (10) then try 15 and then 20.

HTH

Paula

[Randy Hagedorn]

Clay: To add to your comments about the Oracle database's backup mode. We have it in HOT backup mode during the fbackup.

It seems that these warnings are a fairly new occurrence.

[James R. Ferguson · Accepted solution]

Hi:

If this warning was preceeded by one that indicates that the (same) file was inuse (fbackup(3007)) then you have attempted to backup a file that changed between the time it was first targeted for transfer to tape and the time it finished transfering to tape.

'fbackup' note the timestamp of the file on disk as it begins its transfer to tape and again when it has finished its replication to tape. If the timestamp's differ, 'fbackup' will attempt to recopy the file 'maxretries' times as governed by this parameter's value in the 'fbackup' 'config' file (see the 'fbackup' man pages for more details).

The dire problem, here is that it appears that your database was inuse during the backup process. While you might get lucky and have a situation where you could copy various portions of your database, they would be inconsistent if you tried to restore them.

Too, if a file is in use when 'fbackup' reties its copy, tape is wasted. 'fbackup' marks the first (second, third, etc.) tape copy as "bad" and proceeds to copy the file (again) from disk until a good copy (same timestamp) exists on tape or until 'maxretries' is exhausted.

Regards!

...JRF...

[A. Clay Stephenson]

If you are it hotbackup mode then as suggested you need to increase the number of retries in the config file. I suspect that why this is occurring now is that the database has become more active during the backup window.

I might suggest a Plan B. You might consider doing a shutdown, a snapshot mount of all your Oracle filesystems, and then a startup. You would have a total downtime of about two minutes. You then backup the snapshot mount points. You then have all the warm-fuzzy safety of a true cold backup with almost all the uptime of a hotbackup. You must have OnlineJFS to do this.