Re: [Bacula-users] Testing out data spooling

[Kern S. <ke...@si...>]

Hello,

On Thu, 2004-04-08 at 18:16, Matt Howard wrote:
> First, the setup:
> 
> spool directory /var/bacula/spool  (excluded from backups)
> max spool size is 5GB   (is this reasonable?)
> backup of 17GB fileset (Full level)
> 
> As you can tell, the data spool filled/drained a few times.  When the job
> finished, "status storage" reported a negative number for active job bytes. 
> Max job bytes was correct, though.

You'll need to send me the exact output so I can track it down. Without
actually "seeing" it in context, I would just be guessing.  It is
possible that I stupidly edit the bytes as a 32 bit quantity rather than
64 bits, which would be harmless but cause it to display negative.

> 
> After the job had appeared to finish, the FD was done scanning through the
> files, the SD was writing no more to tape, but the job was still a
> "running" job, and sat that way for a few minutes.

This is quite normal, and I should finish the documentation on this over
the weekend, but to ease your worries, when doing data spooling, it also
forces attribute spooling so that the catalog is not updated until after
the data is written. This is to avoid that the catalog points to
something non-existing on tape. Sending the attributes to the Director
and getting them into the catalog take a bit of time -- especially for 5
GB of data (probably quite a few files).

> 
> I also fired off an incremental job after that.  The active job bytes
> reported positive numbers, but now max job bytes is negaative.  This job
> also appeared to block for a few minutes when it was done.

Same as above.

> 
> Anyway, just a little weird.  Everything claims to have completed, and I'll
> try to restore some files later...  

Please let me know the results.

Regards, Kern