I am seeing fairly regular occurances of one or other pool grabbing 
every available scratch pool tape.

I'm not the only one who's reported it and I've filed it as a bug a 
couple of times but the issue has never been fixed.

The problem is intermittent, apparently impossible to predict and leaves 
no indication of errors in the logfile. It's happened using on several 
versions of Bacula (3.2 up to current (but I only started using scratch 
pools at 3.2)) MySQL (several versions) and Postgres (Several versions), 
using LTO2 and LTO5 tapes.


The only thing which I can think of which _might_ cause this is that if 
several jobs get started at once when there is no tape in "append" state 
in their pool they may all independently decide to grab one from the 
scratch pool (is there locking to prevent this happening?)


Can someone take a look at the code and see if this scenario is 
possible? Does anyone have any other ideas why multiple tapes would be 
transferred from a scratch pool?

For those who might think this is minor (which I did): I have been on 
leave for 10 days. This happened while I was away and caused major 
disruption to backups because the backup operators had no idea how to 
cope with the situation - and our director decided that as it wasn't 
supposed to happen they should wait for me to return before attempting 
to fix it.

(35 new, unused tapes which had been added to the Scratch pool were 
pulled into one of the backup pools, leaving the other 4 pools starved 
of resources! They were afraid if they added more tapes it would happen 
again)





------------------------------------------------------------------------------
Learn Windows Azure Live!  Tuesday, Dec 13, 2011
Microsoft is holding a special Learn Windows Azure training event for 
developers. It will provide a great way to learn Windows Azure and what it 
provides. You can attend the event by watching it streamed LIVE online.  
Learn more at http://p.sf.net/sfu/ms-windowsazure
_______________________________________________
Bacula-devel mailing list
Bacula-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-devel

Reply via email to