Bacula Bug???

[tqz]

Hello All

When trying to access the director status (either via webmin or using the command prompt) the system seems to hang. When doing this via the command prompt at the server it gives the following info: When the Daemon started and heap information etc then the columns for the table for scheduled jobs appear but not the data in the tables, the server then just hangs.

If i restart bacula, then go to the bacula director using webmin I see that the jobs that should have been backed up yesterday are still in the scheduled backup jobs table. There is no error in the completed backup jobs table.

Is there a bug??? I haven’t changed anything in bacula and everything was working okay the previous day....

Any help would be greatly appreciated on how to rectify this problem.

[Irithori]

The Director is waiting for the database.

Connect to the bacula DB.

If it is a mysqld, issue a
mysql> show processlist;

You will notice some heavy requests.
Possible deleting of pruned jobs.
Or adding a job´s file attributes of a client with lots of files.

I had to tune mysqld, reconfigure raid layout and finally move
to a more powerfull machine.
From time to time I still face this issue.
In my case, there are 650million files in that catalog, so well..
Perhaps I should consider postgres

[Irithori]

And about that:

Hello All

If i restart bacula, then go to the bacula director using webmin I see that the jobs that should have been backed up yesterday are still in the scheduled backup jobs table. There is no error in the completed backup jobs table.

What happened is:
The jobs were still running.

At least not all steps were finished, as after the actual backup this happens:
a) the Storage Daemon has to spool data to tape (if you use the spool feature (you should))
b) and then the file attributes are copied from SD to Director and finally to DB

If you just kill the director while jobs are running, these jobs are not reported.

BTW: the bweb webinterface will falsely report them as "running", but with ridiculous running times.
These JobIDs can be safely deleted, as the jobs are not complete anyway.

What you see in the scheduled section is just that.
The next schedules for these jobs.

Hope that helps a little

[tqz]

Hi Irithori and thanks for your responses!

I did check the process list and didnt see any heavy processes (for postgres or bacula or anything else for that matter) eating up memory, strangely everything pretty much sorted itself out as soon as I posted....everything seems to be okay since then???

Thanks again for your help!