On Wed, Apr 25, 2007 at 12:46:36AM +0200, Steen wrote: > Tirsdag 24 april 2007 03:35 skrev Ross Boylan: > > On Mon, Apr 23, 2007 at 06:23:27PM -0700, Ross Boylan wrote: > > > I created a bsr file with the names of all my volumes in order, > > > separated by |'s. There seem to be too many: > > > # bscan -s -m -b vols.bsr -v /var/spool/bacula -c > > > /etc/bacula/bacula-sd.conf -P xxx -h localhost > If I remember correctly, you need all the volumes that belong to one job in > one go, not necessarily all existing volumes at once. Maybe you can look at > the timestamps on the volume files. If you take the latest you would > hopefully have one version of the catalog, then you can get the rest one job > at a time.
I suspect that would work, at least unless individual jobs had too many volumes (or volumes with too many letters). Since my volumes have jobs run together, it's not so likely to find a neat breaking point. I think I've solved the "too many volumes" problem another way. Apparently the bscan -b file is allowed to be a regular bsr file as well as the formats discussed in the bscan section. This allows each volume to go on a separate line. So I used Volume=Vol01 Volume=Vol02 etc. for my .bsr file. My crash occurred the day after a differential backup, so I only put in the Full volumes and the volumes from the last Differential. That approach allowed a single pass through the volumes. In an earlier solution I used the actual .bsr files from the individual jobs. That required more work, since it included lots more volumes (all the ones between the full and the final differential) and required processing some volumes repeatedly (the ones holding several jobs). Even the trimmed down approach took over 3 hours to reconstruct the catalog (about 486,000 records in the files table). 3GHz PIV with a reasonable, but single, SATA drive, communicating over unix domain sockets with Postgresql 8.1 on Debian GNU/Linux. The restores themselves took only minutes. I'm not completely sure if that intermediate solution (using the actual .bsr files from jobs) is OK, since I ended up with a corrupted catalog after using it. I suspect it was corrupted by my earlier, failed, efforst with the long lines of volume names. I wiped it all out and started over. Ross P.S. I can't say definitively that the preceding approach worked. I've done restores and see no problems with them so far. Esoterica: some of my original backups failed because the disk filled. The result were some 0 byte volumes, as well as some truncated and damaged ones. Originally I had the 0 byte ones on the list; bscan said it couldn't find the volume header and was waiting for me to mount a new volume. I removed those entries from the .bsr, and bscan seems to have tossed the remaining bad bits pretty well. ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
