Re: filestorage :: some misconfiguration

Bill Arlofski via Bacula-users <[email protected]> Tue, 19 Aug 2025 16:54:05 +0000

Newsgroups	gmane.comp.bacula.user
Message-ID	<[email protected]>

On 8/18/25 3:48 PM, Rob Gerber wrote:
>
> At a glance, your most recent config looks reasonable. I will note that with one 'drive' device in your SD, and no
> autochanger, you will only be able to write one job at a time. This is why one might want to have multiple 'drives' and an
> autochanger. MAYBE your spooling will change this. Not sure, I haven't spooled a disk volume job.

Hello Rob,

Just a few comments here:

If a Drive device has `MaximumConcurrentJobs = [a number greater than 1]` then more than one job can write its data
concurrently (interleaved) to the same volume using the save drive device. This is also true for tape drives.

Having said this it simply makes no sense at all (in my experience) to not use file Autochangers with several drive devices
in them. It gives so much more flexibility.

Typically, I will create my Autochangers with a minimum of 10 devices (usually more like 15 or 20), usually each usually
having `MaximumConcurrentJobs = 1` (can be set to anything, I just prefer more drives, and 1 ConcurrentJob per drive).

I will often also create a few extra devices with `ReadOnly = yes` so that there are a few devices available for critical
restores if backups are using the rest of the other devices in the Autochanger.

Enabling DataSpooling when using disk or cloud devices makes no sense and will not speed up anything. It will just slow
things down since the data gets written to a spool file, then gets written to a drive device.

Also, AttributeSpooling should always be enabled - unless debugging/troubleshooting some issue. With Attribute spooling
enabled, the SD spools the metadata to disk when the job is running, then sends the 1 attribute spool file to the Director to
be quickly and efficiently batch inserted into the database.

With attribute spooling off, the SD will send metadata for each file backed up in real time, which is very inefficient and
will very noticeably slow things down.

Here are some AttributeSpooling notes from the 15.0.2 manual:
----8<----
◾ When data spooling is enabled, Bacula automatically turns on attribute spooling. In other
words, it also spools the catalog entries to disk. This is done so that in case the job fails,
there will be no catalog entries pointing to non-existent tape backups.

◾ Attribute despooling occurs near the end of a job. The Storage daemon accumulates file
attributes during the backup and sends them to the Director at the end of the job. The
Director then inserts the file attributes into the catalog. During this insertion, the tape
drive may be inactive. When the file attribute insertion is completed, the job terminates.

◾ Attribute spool files are always placed in the working directory of the Storage daemon.

◾ When Bacula begins despooling data spooled to disk, it takes exclusive use of the tape.
This has the major advantage that in running multiple simultaneous jobs at the same time,
the blocks of several jobs will not be intermingled.

◾ It probably does not make a lot of sense to enable data spooling if you are writing to disk
files.
----8<----

Hope this helps,
Bill

--
Bill Arlofski
[email protected]

_______________________________________________
Bacula-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/bacula-users

signature.asc (application/pgp-signature, 583 B)