Greetings!!
We had a couple concurrent disk failures on our Solaris server running
DiskSuite and lost the RAID volume/filesystem that the netbackup/db
directory was on. I was able to tar off the db directory, but the one
db/backup/IMAGE1 file had some I/O errors on the read, so I have a
feeling it's not viable. Unfortunately, I was not aware that we were
backing up the DB to disk instead of to tape, and we were backing it up
to the same damaged volume (ugh!).
I rebuilt the volume and filesystem and re-tarred all the files back. I
was able to restart NB and randomly browse backups without any problems.
In fact, I was even able to restore a file. This leads me to believe
that the catalog may not be damaged, or at least some of it might be
salvageable. I've temporarily deactivated all backups including the DB
backup (which I've since config'd to backup to tape now!) since I don't
want anything to be overwritten just yet until I hear back from Veritas.
In the meantime, I have a couple or three questions:
1) The tech at Veritas said there is a CLI for checking the catalog
file, does anyone know what that is and what the command format is?
2) When the catalog is backed up to disk, where does the backup go?
Since my file was in the "backup" directory, does that mean the original
catalog file is someplace else, and if so, where? This could mean that
the catalog is intact and only the backup of the catalog got corrupted.
3) If the catalog is corrupted, is there a way to rebuild w/out having
to read every single tape? And how do I restore a catalog from backups
that were written to disk instead of tape? At least we will have only
lost a couple days worth of backups in the catalog, and that's a lot
less tapes to re-read.
As always, thanks for any info!!!!!
....k
-==-==-==-==-==-==-==-==-==-==-==-==-==-==-==-==-==-==-==-
Kevin Freels kfreels AT askjeeves DOT com
Senior UNIX SysAdmin 510.985.7444
Ask Jeeves ask.com
"We keep you alive to serve this ship. Row well and live."
|
