On Thursday 07 July 2005 13:47, Arno Lehmann wrote:
> Hello,
>
> today, I think I found a problem with baculas handling of disk space.
>
> The situation:
> I use spooling, and had one large job running overnight. The spool space
> is not big enough to hold the complete job, so the job is run in slices:
> spool until disk is full, despool, repeat until done.
>
> The spool directory and baculas working directory are on the same
> partition.
>
> When the regular backup jobs for today were about to start that
> partition was full.
> The result: The scheduled jobs didn't run, are marked with status
> "Error" and I received an error email without any contents. Nothing in
> the console log as well.
>
> I assume that the log messages could not be created because the disk
> space wasn't available, the same with the information mails.
>
> I also assume that the jobs were not run because they couldn't create
>
> their spool files. In 'status dir' I have entries like this:
> > Terminated Jobs:
> > JobId Level Files Bytes Status Finished Name
> > ========================================================================
>
> ...
>
> > 2610 Full 5 113,578 Error 07-Jul-05 08:37 BackupMail
>
> While I can understand that behaviour, I'd like to know three things:
> First, has anybody else encountered this behaviour?
> Second, are my assumptions correct?
> Third, shouldn't this behaviour be changed? One solution would be to
> hold a job when there's not enough spool space when it's started, I'd say.
>
> And
> #version
> goblin-dir Version: 1.36.3 (22 April 2005)
>
> #status storage=DLT
> Connecting to Storage daemon DLT at goblin:9103
>
> goblin-sd Version: 1.36.3 (22 April 2005) i586-pc-linux-gnu suse 8.1
>
> Spool space is on a LVM partition across several SCSI disks, no SCSI or
> filesystem errors according to the system log. And I know that more
> spool space _is_ desirable...
If reasonable messages were not printed in the job log when jobs failed, then
it is something that interests me. I don't plan to spend any time, with
perhaps some time for more documentation, building a fail safe operating
system if you over fill your working directory. If I am not mistaken, the
document warns you that you must have a reasonable space available in the
working directory. Concerning spooling, I give you the tools to put the
spool directory in a place other than your working directory, and I give you
the tools to limit the amount of spool space used. Beyond that, as long as
Bacula produces reasonable error messages, it is up to you to manage things.
If you think this is not a correct attitude on my part, then simply fill up
your tmp directory and or your root and see how many jobs your OS holds for
rescheduling ... :-) Another good one to try is delete /dev/null, then
re-create it with "touch /dev/null" and see how well your OS runs ...
--
Best regards,
Kern
(">
/\
V_V
-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
