Restore Issue

[iman_1]

Hi All,

I have taken an image backup of the system using compression with below command

backup/image/log/verify/ignore=(nobackup,interlock) $1$dua1: mkb0:dua1_050806.bck/rewind/label=dua1bk

The imag backup completed successfully and after this we have increased the disk capacity on the HSD30 RAID5 disk array from 8 disks to 11 disks.

I had to restore the same image backup and while doing so, it did restore the first tape but as soon as it took the second tape from tape auto loader it started giving the below error.

-SYSTEM-F-VOLINV, volume is not software enabled
%BACKUP-E-WRITEBLOCK, error writing block 1331456 of $1$dua3:[directory_structure]file_name.RDA;1

Can anyone let me know why we are facing this error ?

Also the backup took about 4 hours to write on one tape and 6 hours to verify the first tape but its taking as long as 23 hours to restore !!!!

Any Suggestions ?

Cheers,
Iman

[Robert Gezelter]

Iman,

It is no surprise that the restore takes longer than the backup. File allocation alone is a fairly expensive operation. Also, remember that RAID 5 performance is poorest for writes.

The error writing a block is more troubling. Are you sure that your RAID set is configured correctly?

- Bob Gezelter, http://www.rlgsc.com

[Volker Halle]

Iman,

did you show the backup error message in the right order ? A system service status message should normally follow the %BACKUP-E-WRITEBLOCK error message.

Did you check for any errors on the output device $1$DUA3: ? Was the disk still mounted foreign ?

Any errors on the console of the HSD30 ?

Volker.

[Burkart Beat · Accepted solution]

Hi

After building a new RAID-5 set, the controller calulates all the checksums back to disk. This consumes a lot of resources on the disks and controllers. Writing data to the diskset during this phase is possible but very slow. That's why it takes so long.

Enabling WRITEBACK Cashing during the restore will also speedup the process.

The error means, that during restore the drive vent offline, maybe because the RAID-5 building process still was active and took all resources... Happend on my site too using a HSG80 controller.

Best regards
Beat

[iman_1]

Hi Beat,

Yes I saw this suggestion and also think enabling the writeback cache can solve the problems.

Just one quesry, can this be done online when the restore is running on the system, I mean this parameter bde updated dynamically without affecting the restore on the system ?

Your suggestions are really helpful

Cheers,
Iman

[iman_1]

Hi Volker,

Yes the sequence of the error has been as quoted earlier. I have now made the writeback cache parameter enabled and hope so this solves the issue. Also restore has been started from different set of non compressed tapes. Could it be an issue of compression coupled with the slow speed in able to transfer the data which could cause these errors ?

Cheers,
Iman

[John Abbott_2]

Hi Iman,

You can see how far your RAID set has got in reconstructing from the controller by issuing a

SHOW lun (look under the users column for the raid set name)

SHOW raid-set-name
.
here you will see the members of the raid set and a percentage on reconstruction
.

Time permitting, I generally don't start a restore until the raid set has reconstructed.

Kind Regards
John.