On Friday 30 May 2008 03:34:13 pm Thomas Mueller wrote:
> Stephan Jennewein schrieb:
> > On Monday 26 May 2008 07:33:24 am Thomas Mueller wrote:
> >>>> post your "Maximum Concurrent Jobs" definitions in bacula-dir.conf (in
> >>>> resources director,client,storage and job), bacula-fd.conf  and
> >>>> bacula-sd.conf to see if it's correct allover the different places.
> >>>> And if you use "Maximum Volume Jobs" also post that.
> >>>
> >>> director:  Maximum Concurrent Jobs = 20
> >>> storage: Maximum Concurrent Jobs = 10
> >>
> >> this is in bacula-dir.conf, right?
> >
> > Yes, this is in my bacula-dir.conf.
> >
> >>> client1: no Maximum Concurrent Jobs Definition
> >>>   all jobs have no Maximum Concurrent Jobs Definition
> >>
> >> default for client resource in bacula-dir.conf  "Maximum Concurrent
> >> Jobs" is 1.
> >>
> >>> client2: same here
> >>> client3: same here
> >>> client4: Maximum Concurrent Jobs = 1
> >>>   jobs no definition
> >>> client5:  Maximum Concurrent Jobs = 4
> >>>   jobs no definition
> >>> client6: nothing here
> >>> client7: Maximum Concurrent Jobs = 20
> >>>
> >>>   jobs no definition
> >>>
> >>> i cannot post the bacula-fd.conf because the clients dont belong to me.
> >>> i only realize the backup server in our institute.
> >>
> >> in bacula-fd.con the default "Maximum Concurrent Jobs" is 2 (1 for data
> >> / 1 for "status" requests).
> >>
> >> the other entries seem to be ok. the My understandig is, that the
> >> limiting "Maximum Concurrent Jobs" is the one from bacula-dir.conf in
> >> resrouce storage with 10 concurrent connections if you changed the
> >> default in bacula-fd.conf.
> >
> > maybe but this not my point. i want that the jobs are scheduled
> > automatically in an alternating way, because otherwise the scheduled jobs
> > would wait for the first computer with more than one job. that's why i
> > want something like: computer 1 - job 1
> > computer 2 - job 1
> > computer 3 - job 1
> > computer 2 - job 2
> > computer 3 - job 2
> > computer 1 - job 2
> > and so on.
> > instead of
> > computer 1 - job 1
> > computer 1 - job 2
> > computer 1 - job 3
> > computer 2 - job 1
> > computer 2 - job 2
> > computer 2 - job 3
> > computer 3 - job 1
> > computer 3 - job 2
> > computer 3 - job 3
> > thats what happen at the moment.
> > increasing the 'maximum concurrent jobs' in the bacula-fd.conf would not
> > fix the problem because the harddrive isn't faster just because i want to
> > run more jobs parallel. so i only can parallelize if i can dynamically
> > run jobs of different computers at the same time.
>
> so, i don't know why it isn't working like you expect it.
How do you schedule the jobs? I only have one schedule for each server so all 
of the jobs of one server starts simultaneously. But I have much more than 3 
jobs per computer so i don't think the schedule is the right way to go.

> i've got 4 servers and they backup in parallel one job per server.
>
> bacula-dir.conf
>    Director: "Maximum Concurrent Jobs" = 10
>    Storage: "Maximum Concurrent Jobs" = 4
>
> Anyway you have set "Maximum Concurrent Jobs" in bacula-dir for
> client4,5,7 as you want only start 1 job per computer?
To what should I set the option for each client?

> And the backup jobs share write to a volume in the same pool?
I have two pools one for full backups and one for differential and incremental 
and an autochanger device with two drives. ok, i splitted the catalog up 
because of performance issues. but i dont think this is the problem. most of 
the computers are still in one catalog.

Stephan

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users