07-31-2015 02:14 PM
Hi All,
I have been facing an issue with catalog recovery. Symantec was not able to fix that.
I had taken a catalog backup of Master<auspxxxx)to a single tape CA0080(LTO6)and decommissioned the auspxxxx. That catalog backup had two images. Unfortunately we got a request to restore for a server from auspxxxx. So we build a vm with same name, configuration(NBU 7.6.0.3) and OS(SUSE Linux). It does not have any tape drives attached to it. what i did was, I imported CA0080 into my other Master(cmsxxxx), duplicated the catalog images to disk and mounted the disk path to auspxxxx server. I was able to import and create DRfile. When i recover the catalog(bprecover -wizard -copy 2), it was able recover to configuration and images from one image and could not recover NBDB from other image. it says Failed to recover NBDB as it was looking for CA0080 to mount. Not sure why it is looking for that tape even though we have same images from disk set as primary copy. I contacted symantec support for this but no luck. Could you anyone help getting this fixed.
Solved! Go to Solution.
08-04-2015 03:43 AM
I think the problem is when the original catalog backup image gets restored along with all other images.
The original image contains the tape media id.
When it is time to restore nbdb, it finds the media-id in the image header and therefore requests the media id.
Can you identify the image-id for the catalog backup?
Check backup destination(s) for this image with:
bpimagelist -backupid <image-id> -L
Based on what we see in this output, we can figure out a way to expire the image or the copy number, and then use bprecover again with dr-file to only recover nbdb.
08-04-2015 03:43 AM
I think the problem is when the original catalog backup image gets restored along with all other images.
The original image contains the tape media id.
When it is time to restore nbdb, it finds the media-id in the image header and therefore requests the media id.
Can you identify the image-id for the catalog backup?
Check backup destination(s) for this image with:
bpimagelist -backupid <image-id> -L
Based on what we see in this output, we can figure out a way to expire the image or the copy number, and then use bprecover again with dr-file to only recover nbdb.
08-13-2015 02:31 PM
If you browse the catalog backup image you imported (or bplist it) can you see and recover just NBDB.db file?
08-13-2015 11:53 PM
Marianne has the answer I do believe.
After the NBDB fails to recover, you could change the FRAG lines in the catalog to point at the disk image. I think cat_export > edit the 'header file' and cat_import should do it.
However, this is probably 'unsupported' as you're not really meant to be editing this manually.
A better way, would be to use nbcatsync, which should correct the media ids in the catalog.
areznik:
You can't just recover the NBDB.db file. You can however recover 'ALL' the files in /usr/openv/netbackup/db/staging to and bring up the NBDB from that, so you were on the right lines, but it's not a single file.
09-04-2015 02:40 AM
Thinking about this, the nbcatsync should work for this issue, and resync the correct media ids for the images in the catalog.