Re: [Bacula-users] Spooling can slow Bacula

[John K. <ko...@co...>]

On Mon, Feb 20, 2006 at 11:36:45PM +0100, Ludovic Strappazon wrote:
> Wolfgang Denk wrote:

> >And assuming you are backing up some  big  RAID  array  it  might  be
> >difficult to provide big enough spool files.

That's certainly the case in my situation.  I don't have 4 TB sitting
idle that I can dedicate to spool space, and even if I did, it
wouldn't help performance any because of the sequential nature of
spooled backup jobs.

> >So the only real advantage of data spooling as I see it  is  that  it
> >helps  to  reduce  tape  (drive) wear by avoiding frequent start-stop
> >sequences - which is only  a  problem  if  your  file  system  cannot
> >provide  the  data  faster  than the tape can write. In my experience
> >this is only the case for differential and incremental backups -  and
> >then the data size might be small enough to fit in a spool file.

Yes, I think that that's right, but it wasn't obvious from the
documentation.  That's what lead me to give this a try.

> >Ideally I'd like to be able to turn spooling  off  for  level  "full"
> >backups   while   leaving   it  turned  on  for  "differential"  and
> >"incremental" ones.
> 
> You can override your job's default in the schedule, something like :
> 
> Schedule {
>  Name = "MySchedule"
>  Run = pool="MyPool" Full fri at 20:30
>  Run = SpoolData=yes pool="MyPool" Storage="LTOdrive" Differential mon-thu at 20:30
>  Run = SpoolData=yes pool="MyPool" Storage="LTOdrive" Incremental fri at 11:35
> }
> 
> Ludovic Strappazon.

This looks like the way to go.  If I leave spooling enabled at all, it
will be with a setup like this where its use can be limited to the
remote clients that stand to benefit.

-- John Kodis.