Bacula-users

Re: [Bacula-users] How to troubleshoot a slow client, when there is no obvious answer

2013-01-22 18:12:44
Subject: Re: [Bacula-users] How to troubleshoot a slow client, when there is no obvious answer
From: dweimer <dweimer AT dweimer DOT net>
To: <bacula-users AT lists.sourceforge DOT net>
Date: Tue, 22 Jan 2013 17:09:18 -0600
On 2013-01-22 15:34, Adrian Reyer wrote:
> On Tue, Jan 22, 2013 at 01:37:43PM -0600, dweimer wrote:
>> If you have checked disk I/O, CPU, memory, network, on both the 
>> client
>> and the server, all seem great, both from a statistics look, showing 
>> all
>
> What about many small files, possibly very many in single 
> directories? I
> have seen these with huge impact on backup performance, especially on
> busy directories.
>
> Regards,
>       Adrian

There really hasn't been a huge change in the data, below are from the 
client status, the full I canceled this morning has the majority of 
files done, but not the bulk of the data.  The biggest chuck of which is 
the around 30 iTunes movie files, which is where I should see the best 
performance in the backup, at the point I canceled this it was in the 
process of processing one of those 4+G movie files.

Full     67,353    284.8 G  OK       15-Jan-13 03:30
Incr          0         0   OK       16-Jan-13 00:05
Incr         18    5.145 M  OK       17-Jan-13 00:05
Incr         23    9.016 M  OK       18-Jan-13 00:05
Incr          8    11.50 M  OK       19-Jan-13 00:05
Incr         31    16.09 M  OK       21-Jan-13 00:08
Incr         31    16.09 M  OK       21-Jan-13 00:08
Full     53,174    125.0 G  Cancel   22-Jan-13 08:46

However you see there are multiple incremental backups ran on the 21st. 
I did have some failures Monday morning, I had rebooted the server on 
Sunday and forgot to remount my backup volume.  It was in the server 
during reboot, but is encrypted so it doesn't mount automatically.  
Those jobs had marked all of my disk volumes with an error state, I 
hadn't noticed, so when I remounted the volume for its Monday backups 
most jobs failed, due to lack of usable online volumes but a couple 
completed, and got kicked off a second time after I marked the volumes 
back online.  Its possible all of this had been caused by snapshots that 
had been stuck around since the failure of the backup job on the 
bacula-dir and bacula-sd server, the database volume was included in 
this, that few day old snap shot may have been slowing down writes to 
the database.

I have cleared that and its running a dbcheck on the database now, I 
will kick off new backups when that finishes and see how things go.  
Hopefully this was the cause, however I may have to do some more 
performance searching because the snapshot shouldn't have caused the 
performance to degrade that much.

-- 
Thanks,
    Dean E. Weimer
    http://www.dweimer.net/

------------------------------------------------------------------------------
Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS,
MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current
with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft
MVPs and experts. ON SALE this month only -- learn more at:
http://p.sf.net/sfu/learnnow-d2d
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users

<Prev in Thread] Current Thread [Next in Thread>