[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Bacula-devel] Director bug when using two storage daemons?


Hello,

Sorry, I was too jet lagged yesterday afternoon, so I asked an incomplete 
question.  Would you please run it one more time and add just after the "list 
volumes" a:

  llist volume=TestVolume0001

Note, it really is "llist" with two l's.

This will tell us if your retention periods are setup correctly.  I suspect 
they might be wrong, which is causing the pruning.   If they are OK, then we 
have some strange pruning bug ...

Regards,

Kern

PS: you have done a good job creating a test. I wish more people would do so, 
but there are just two more steps.  1. (a bit painful) is to convert your new 
xxx-director.conf  into a xxx-director.conf.in by replacing all 
the "variable" parameters by @xxx@ keywords.  Then 2. (trivial) adding one 
new line in scripts/do_sed to cause the xxx-director.conf to be created using 
your config file parameters by substituting them into the 
xxx.bacula-dir.conf.in file.


On Friday 14 November 2008 17:51:23 Graham Keeling wrote:
> On Fri, Nov 14, 2008 at 05:13:52PM +0100, Kern Sibbald wrote:
> > On Tuesday 11 November 2008 12:51:56 Graham Keeling wrote:
> > > Hello,
> > >
> > > Since my last email, I found out that you only need one director, one
> > > file daemon, and one storage daemon in order to see the problem.
> > >
> > > Since there has been no response to my last email, I tried to come up
> > > with an easy way for others to demonstrate it for themselves.
> > >
> > > So, I made a regress test.
> > >
> > > I've called the test 'broken-media-bug-test', and it does this:
> > > a) creates a huge 300M file in the build directory
> > > b) starts job "First" (using fileset Set1) backing up the build
> > > directory c) waits 5 seconds
> > > d) starts job "Second" (using fileset Set2) backing up the build
> > > directory e) waits for both jobs to finish
> > > f) tries to restore the build directory using fileset Set1.
> > >
> > > The restore doesn't manage to restore all the files, and looking back
> > > at the debug output, you can see the volume that the first backup
> > > creates being stomped on by the second backup.
> > >
> > > The patch for the test is attached, along with a file containing the
> > > REGRESS_DEBUG=1 output that it gives. The problem can be seen in the
> > > output around line 453, where it decides to purge the Volume.
> >
> > Please add a "list volumes" after step c and send us the job output.
>
> OK, it says this (I've also turned on some debug for the director). The
> full output for this run is also attached, along with my regress config
> file (without my email address).
>
>
> setdebug level=150  Director
> localhost-dir: ua_cmds.c:951-0 setdebug:setdebug level=150  Director:
> list volumes
> localhost-dir: ua_cmds.c:1931-0 UA Open database
> localhost-dir: ua_cmds.c:1950-0 DB regress opened
> localhost-dir: ua_output.c:255-0 list: list volumes
> Automatically selected Catalog: MyCatalog
> Using Catalog "MyCatalog"
> Pool: Default
> +---------+----------------+-----------+---------+----------+----------+---
>----- ------+---------+------+-----------+-----------+-------------+
>
> | MediaId | VolumeName     | VolStatus | Enabled | VolBytes | VolFiles |
> | VolRete
>
> ntion | Recycle | Slot | InChanger | MediaType | LastWritten |
> +---------+----------------+-----------+---------+----------+----------+---
>----- ------+---------+------+-----------+-----------+-------------+
>
> | 1       | TestVolume0001 | Used      | 1       | 210      | 0        |
> | 3153600
>
> 0     | 1       | 0    | 0         | File      | 0           |
> +---------+----------------+-----------+---------+----------+----------+---
>----- ------+---------+------+-----------+-----------+-------------+



-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
Bacula-devel mailing list
Bacula-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.sourceforge.net/lists/listinfo/bacula-devel


This mailing list archive is a service of Copilot Consulting.