Zitat von Antony Mayi <antonymayi AT yahoo DOT com>:

>> ________________________________
>> From: Jérôme Blion <jerome.blion AT free DOT fr>
>> To: bacula-users AT lists.sourceforge DOT net
>> Sent: Wednesday, 20 February 2013, 9:41
>> Subject: Re: [Bacula-users] backup concurrency
>>
>> Le 2013-02-20 10:19, Antony Mayi a écrit :
>>> Hi community,
>>>
>>> I am running backups of multiple database servers and each backup job
>>> is defined with "RunScript" command that dumps the databases on each
>>> particular server and these dumps are then taken for backups.
>>>
>>> The dumps take different times on each server - minutes to couple of
>>> hours. Since the dumping doesn't involve bacula storage daemon I want
>>> another jobs to be running in parallel that can be sending data to
>>> storage while the db servers are dumping the databases. Also the
>>> database dumps can be running in parallel instead of sequentially as
>>> that's purely local matter.
>>>
>>> I've increased the "Maximum Concurrent Jobs" in "Director" resource
>>> but I can still see only one job running at a time. I don't think I
>>> want to increase this as well in "Storage" resource and definitely
>>> not
>>> in the "Job" resource.
>>>
>>> What am I missing here?
>>>
>>> ...running bacula 5.2
>>>
>>> thx,
>>> Antony.
>>
>> Hello,
>>
>> What is the state of the other jobs while one is running?
>> How did you set up your Maximum Concurrent jobs?
>>
>>
>
> good point.
>
> Running Jobs:
> Console connected at 20-Feb-13 11:02
>  JobId Level   Name                       Status
> ======================================================================
>     45 Full    DB1Backup.2013-02-19_23.05.00_21 is running
>     46 Full    CatalogBackup.2013-02-19_23.10.00_22 is waiting for  
> higher priority jobs to finish
>     47 Increme  Srv1Backup.2013-02-20_11.02.40_26 is waiting on max  
> Storage jobs
>     48 Increme  Srv2Backup.2013-02-20_11.02.45_27 is waiting on max  
> Storage jobs
>
>
> so it seems the DB1Backup is blocking the storage although not using  
> it really since it is busy with running the "RunScript" db dump  
> which takes in this case about 4 hours. so that suggests I need to  
> increase the concurrency for accessing the Storage which I wanted to  
> avoid to not to interleave writes from multiple Jobs. I find it  
> quite inefficient blocking the storage resource for several hours  
> although not really using it. Is there a way around it or is the  
> only approach enabling concurrent access to the Storage (which is  
> discouraged from restoration performance point of view)?
>
> thx,
> Antony.

If you want to avoid concurrent (interleaved) write to tape by  
multiple jobs have a look at the data spooling feature. With this you  
split up your jobs in configurable chunks which will be written to  
tape non-interleaved. Your second possibility is to limit the actual  
tape device to 1 job, but not the storage daemon (not sure if this  
works though).

Regards

Andreas



------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_feb
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users