>
> I am struggling to find a method of keeping consistent off-site without
> breaking easy restores.
>
> My planned schedule was as follows:
>
> First Sunday of Month: Full Backup to Disk
> Monday-Saturday: Incremental Backup to Disk
> Friday (after the incremental): Virtual Full Backup to Tape
> Subsequent Sundays in Month: Differential Backup to Disk
>
> Keeping 3 months (or more depending on rate of change and how well
> compression works, which are currently unknowns) on disk for restores,
> and the Friday virtual full going off-Site in case of disaster.
>
> The problem I ran into (well not actually still in testing, and not
> production), is on the Second Friday of the month, the virtual full
> fails, because it can't find the previous virtual full to build the new
> one from. How do I make it go back to the actual full instead of the
> previous Virtual Full?
>
> Will I be stuck creating a virtual full to another disk pool, then
> running a copy job for off-site backups, which unfortunately greatly
> reduces the history of on disk data I can keep for restores.
>
> Or is there some method I have yet to discover that will allow me to
> mark the tape volume unavailable so that it ignores that job on the
> subsequent virtual fulls. I have tried playing with setting the enabled
> status to disabled, and updating the volstatus parameter, without
> getting anywhere.
>
My backup regime is:
. Full backups Friday and Saturday night (spread over two nights because too
much data to do in one night)
. Incrementals 3 times a day every other day
. Every Sunday-Thursday night a virtual full + catalog backup to USB disk for
offsite
. After the virtualfull, purge the offsite volume
The offsite disk will only ever be used in the case of a total loss of the
backup server, so a catalog restore will be required then anyway, so purging
the volume isn't a problem.
I originally modified Bacula so that it could exclude the virtual full medium,
but that was a bit of a hack.
My post-catalog backup script does the purging. I have one director doing
backups for two sites so there are actually two offsite usb disks (one for each
site).
I use autofs for mounting the usb disk automatically. It gets mounted on
/backup/offsite. The sd's are completely separate machines to the director too,
hence the need for the scp.
The script I use follows this email (some stuff redacted).
James
#!/bin/sh
/etc/bacula/scripts/delete_catalog_backup
/usr/bin/mysql --skip-column-names bacula -ubacula -p<password> <<EOF |
SELECT DISTINCT VolumeName
FROM Job
JOIN Pool
ON Job.PoolId = Pool.PoolId
JOIN JobMedia
ON Job.JobId = JobMedia.JobId
JOIN Media
ON JobMedia.MediaId = Media.MediaId
WHERE Pool.Name IN ('site1-offsite', 'site2-offsite');
EOF
while read media
do
echo Purging $media
echo "purge volume=$media" | /usr/bin/bconsole >/dev/null 2>/dev/null
done
# copy catalog bsr to usb too
scp /var/lib/bacula/BackupCatalog.bsr
site1-sd-server:/backup/offsite/BackupCatalog.bsr
------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60133471&iu=/4140/ostg.clktrk
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users
|
