Greetings..

 

I believe I may be hitting a concurrency bug in the storage director code, but I’m pretty new to Bacula and may just not understand what I’m looking at in the trace output.  The quick outline of my scenario is:

 

Bacula version 5.0

SD and DIR running on same box (CentOS 5.4, x86_64)

Completely file based backup, using a custom autochanger script (which does not appear to be the problem, but I mention it because it may be relevant)

 

In my setup:

-          I have a single autochanger defined (changer0)

-          I have 10 drives of DeviceType=File defined for changer0

-          Maximum Concurrent Jobs = 1 is set on each Device entry for each drive

-          Each drive shares the same media type, changer0_media

-          I have three pools – OnSiteFull, OnSiteIncremental, and OnSiteScratch

 

I have 5 backup jobs that all start at the same time (1:30am), and which all will want either OnSiteFull or OnSiteIncremental depending on the day.

 

The problem that I am seeing is that volumes are being incorrectly marked as being in Error status each night.  My storage director trace output can be found at:  http://pastebin.com/raw.php?i=Ne0XJbBh

 

One of the jobs in the trace output that marks a volume as in Error is job 36.  If someone more familiar with how volume selection works could take a peek and give me some guidance I’d really appreciate it.. in looking through the trace and trying to follow the code, it looks like there is a concurrency issue when there is a volume which is wanted on a drive other than the one it is currently loaded into.

 

Thanks,

 

Joe