Hello Ralf, I've copied the user's list, because they will probably be interested in this question ...
On Thursday 13 December 2007 10:51, Ralf Gross wrote: > Kern Sibbald schrieb: > > On Sunday 02 December 2007 16:40, Ralf Gross wrote: > > > Kern Sibbald schrieb: > > > > On Sunday 02 December 2007 11:07, Ralf Gross wrote: > > > > > Ralf Gross schrieb: > > > > > > [...] > > > > > > This changes makes the scratch pool useless, because bacula does > > > > > > not prune volumes during the 'status dir' command, but still > > > > > > decides if a volume of the scratch pool will be used. This way a > > > > > > volume of the scrach pool will be moved to the Differential (or > > > > > > Full) pool all the time the command is issued... > > > > > > > > > > ...and no other volume with state Append, Recycle etc. is available > > > > > in that specific pool. E.g. for each new weekly Differential backup > > > > > a new scratch pool volume will be used if the status dir command is > > > > > issued before. > > > > > > > > > > Just to make it a bit clearer. > > > > > > > > I *think* the patch I attached to your bug report #1019 will fix > > > > these problems. > > > > > > Thanks, I'll apply the patch tomorrow and report what happens next - > > > this might take one or two weeks. > > To give some feedback... the status dir command doesn't move any scratch > pool volume to differenet pools anymore. Thanks for the additional feedback. At least that is one bug that is now fixed ... For those of you who don't know what this is: when the algorithm was modified not to prune when doing the "status" command it "uncovered" a previous bug where if a volume was not found during the status command (much more likely without pruning) it would move one out of the scratch pool -- not really something that one wants done during a "status" command. There is a patch that fixes this ... > > > [...] > > If you want to trigger pruning, you will need to use the prune command. > > Then you can check with the status command. > > I found some references regarding the 'list nextvol' command on the user > list. This command still purges volumes as the 'status dir' command did > before. This is fine for me, I'm just wondering if this is still the > desired behaviour now that 'status dir' doesn't purge volumes in 2.2.x? Yes, thanks for pointing this out. I had forgotten about that command when I was answering you. Yes, it does prune, and I think it is appropriate for it to prune (though this could be a point of discussion) because you are explicitly asking what the next volume will be. One thing that I have noticed and have been tracking is that since I modified the pruning algorithm when Bacula is looking for a Volume (i.e. the SD asks for the next volume) so that it does a minimum pruning, and since I modified the status command not to prune (correct behavior IMO), my databases have been growing. This is explained by the fact that there is no longer any Bacula mechanism that "automatically" forces pruning. Within Bacula's philosophy of keeping everything as long as possible (which was by the way, not previously the case), this is correct behavior. However, it does mean that there are a lot of records in databases that are expired and are not pruned. Some users have complained that this is not logical, but within the Bacula design it *is* logical and the desired behavior. The problem I have is that there are probably some of you out there (like myself) who might want Bacula to more strictly adhere to the retention periods that you have setup. In principle, there is the "prune" command that does this, but at the moment, that needs to be manually done -- i.e. there is no fully automatic way to get Bacula to keep the database fully pruned to the retention periods ... I'm thinking about how to resolve this but have not yet found a solution that I consider ideal (some way of doing it via a directive or script within an Admin Job). In the mean time, I suggest that you all monitor your databases if you are running 2.2.x and if it is growing larger than you want, either you schedule a job that explicitly does a prune command or does a "list nextvol" command. I'll probably have more on this subject once I have had a bit more time to investigate all the angles ... Best regards, Kern > > *list nextvol Job=VU0EM003 days=3 > The next Volume to be used by Job "VU0EM003" (Pool=Incremental, > Level=Incremental) will be 06D122L3 The next Volume to be used by Job > "VU0EM003" (Pool=Differential, Level=Differential) will be 06D127L3 > +-------+----------+---------------------+------+-------+-----------+------ >-----------+-----------+ > > | jobid | name | starttime | type | level | jobfiles | > | jobbytes | jobstatus | > > +-------+----------+---------------------+------+-------+-----------+------ >-----------+-----------+ > > | 283 | VU0EM003 | 2007-06-28 00:08:01 | B | I | 1,118 | > | 2,823,838,461 | T | 285 | VU0EM003 | 2007-06-29 00:08:30 | B | > | I | 1,081 | 12,410,251,160 | T | 287 | VU0EM003 | > | 2007-06-30 00:08:29 | B | I | 1,575 | 1,724,824,930 | T > | | > > [...] > > You have messages. > *mess > 13-Dez 10:41 VU0EM005-dir JobId 0: There are no more Jobs associated with > Volume "06D127L3". Marking it purged. 13-Dez 10:41 VU0EM005-dir JobId 0: > Recycled volume "06D127L3" > > > Ralf > > ------------------------------------------------------------------------- > SF.Net email is sponsored by: > Check out the new SourceForge.net Marketplace. > It's the best place to buy or sell services > for just about anything Open Source. > http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplac >e _______________________________________________ > Bacula-devel mailing list > [EMAIL PROTECTED] > https://lists.sourceforge.net/lists/listinfo/bacula-devel ------------------------------------------------------------------------- SF.Net email is sponsored by: Check out the new SourceForge.net Marketplace. It's the best place to buy or sell services for just about anything Open Source. http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
