Bacula-users

Re: [Bacula-users] Bacula 5.0.3 hangs on MySQL 5.5 -- slow DELETE's

2013-11-19 17:05:43
Subject: Re: [Bacula-users] Bacula 5.0.3 hangs on MySQL 5.5 -- slow DELETE's
From: "Novosielski, Ryan" <novosirj AT ca.rutgers DOT edu>
To: "Novosielski, Ryan" <novosirj AT ca.rutgers DOT edu>, Uwe Schuerkamp <uwe.schuerkamp AT nionex DOT net>
Date: Tue, 19 Nov 2013 17:01:58 -0500
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 11/15/2013 01:58 AM, Novosielski, Ryan wrote:
> On 10/09/2013 05:10 AM, Uwe Schuerkamp wrote:
>> On Wed, Oct 09, 2013 at 01:39:52AM -0400, Novosielski, Ryan 
>> wrote:
>>> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
>>> 
>>> Hi all,
>>> 
>>> I upgraded my system from Bacula 2.2.8 to 5.0.3 awhile back and
>>> upgraded MySQL from 5.0 to 5.5 if I'm not mistaken. Since that
>>> time, I've been having reliability problems. My schedule is to
>>> run incrementals every day except for the 2nd and 4th Tuesday
>>> and fulls on those days instead. When I tend to have trouble is
>>> on the 2nd and 4th Tuesday. The backups on those days hang -- I
>>> need to restart the storage daemon and the director it seems
>>> like to clear the hang. That is another problem. However, I
>>> believe the cause of the hang is a slow delete when the
>>> director goes to purge the "full" tape. I have tried running
>>> 'status dir' right before one of these backups is to take place
>>> and there is a pause for probably more than a minute. MySQL at
>>> that time grows to 20% of the system in CPU use. I checked
>>> 'mysqladmin processlist' and I see what is taking so long is
>>> the following:
>>> 
>>> | 1063 | bacula | localhost | bacula | Query   | 71   | 
>>> updating | DELETE FROM File WHERE JobId IN 
>>> (14485,14486,14487,14488,14489) |
> 
>> Hi Ryan,
> 
>> how large is your File table? Have you tried a "repair table 
>> File" and maybe a dbcheck? Have you checked the indices for 
>> damage? Are you using innodb or MyISAM for your storage backend?
>>  How much RAM is installed in your box, and how large is your 
>> bacula database overall?
> 
>> Depending on the number of rows in your file table, volume 
>> recycling (which involves deleting all records for jobs on that 
>> volume from the File table) can take a very long time. For large 
>> tapes and File tables (a billion rows or more) I have seen 
>> upwards of 4 hours for a single volume recycle, but thankfully 
>> those days are past now for us since we got rid of copy jobs and
>>  migrated to disk-based backups with separate devices, storages 
>> and pools for each client.
> 
> I'm expecting that my catalog is not even close to "large" for
> what this software can handle, even on my older system. My file
> table has 1.4M rows. The whole database takes up 401M on disk. Not
> sure if that's the measure of "large" you were looking for. My
> tables are MyISAM. System is a 4x480MHz UltraSparc-II w/4GB of RAM.
> And like I said, this machine functioned fine on Bacula 2.2.8 with
>  MySQL probably 5.0. Whatever Blastwave was offering for packages 
> (probably could get the exact MySQL version if it mattered).
> 
> I seem to have improved the times some but I still have the hang, 
> and the only time it seems to occur is on days when a Full backup 
> tape needs to be recycled in order to be used. So I can't figure 
> out whether it's related to MySQL or not, but the hang happens and
>  I can't figure out what information would be helpful for 
> troubleshooting it. A truss shows all of the processes don't really
> seem to be doing anything. A mount, however, will hang. Seems like
> the storage daemon is locked up when this happens.

Today I got a hang recycling my weekly incremental tape, which is
generally a quick recycle. Any advise for attempting to diagnose this?
It is making Bacula rather unreliable for me.

PS: I'm considering upgrading to see if it helps, but I was using
packages from OpenCSW which is why I'm at the version I'm at. I guess
I could see about trying to learn the packaging system/take what
exists and update the packages to a current release.

The symptom I get is the storage daemon appears to be completely
locked up. To solve it, I have to restart it and then the director.

- -- 
 ____*Note: UMDNJ is now Rutgers-Biomedical and Health Sciences*
 || \\UTGERS      |---------------------*O*---------------------
 ||_// Biomedical | Ryan Novosielski - Sr. Systems Programmer
 || \\ and Health | novosirj AT rutgers DOT edu - 973/972.0922 (2x0922)
 ||  \\  Sciences | OIT/EI-Academic Svcs. - ADMC 450, Newark
      `'
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iEYEARECAAYFAlKL39UACgkQmb+gadEcsb4jiwCgiefY6+DeX4J2SO+Qr9xATpEg
clkAnRKoDhGfziT3A0T6jaeHYXBmee4H
=Hx6o
-----END PGP SIGNATURE-----


------------------------------------------------------------------------------
Shape the Mobile Experience: Free Subscription
Software experts and developers: Be at the forefront of tech innovation.
Intel(R) Software Adrenaline delivers strategic insight and game-changing 
conversations that shape the rapidly evolving mobile landscape. Sign up now. 
http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk
_______________________________________________
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>