Bacula-users

Re: [Bacula-users] Network timeouts while running enormous backup

2011-05-30 15:11:28
Subject: Re: [Bacula-users] Network timeouts while running enormous backup
From: S H <shdashbeta AT gmail DOT com>
To: bacula-users AT lists.sourceforge DOT net
Date: Mon, 30 May 2011 15:08:12 -0400
On Thu, May 26, 2011 at 4:58 PM, S H <shdashbeta AT gmail DOT com> wrote:
> Hello,
>
> I've got a very large backup set (55M files, 1TB of data) that's
> simply not backing up. This is always the failure message:
>
> 26-May 16:35 buny1-dir JobId 5223: Fatal error: Network error with FD
> during Backup: ERR=Connection timed out
> 26-May 16:35 buny1-dir JobId 5223: Fatal error: No Job status returned from 
> FD.
>
> I've set "Heartbeat Interval = 1 minute" on the FD, the SD, and the
> Director to no avail. This was happening at random times during the
> backup job until I enabled spooling -- then the job ran for 27 hours,
> spooled up all of its data to the SD and failed with the above message
> when it started to despool. It was heartbreaking to watch a terabyte
> of data and 20+GB of attribute spool just disappear.
>
> The Director and SD both run on the same server: OpenBSD 4.6 (32-bit)
> with Bacula 2.4.4. The client is FreeBSD 8.1 (64-bit) with Bacula
> 2..4.4. I know these are outdated versions; I just inherited the
> environment and they're working for everything else so I haven't
> wanted to go through the pain of upgrading until I have some spare
> cycles to dedicate to it.
>
> Compression and FD encryption are off but network traffic runs over
> TLS everywhere.
>
> Is there anything I can possibly check that might help?
>
> -SH
>

I added a network interface to the backup server on the same segment
as the client and the timeout disappeared, so it seems my firewall was
expiring a TCP session somewhere along the line. Unsure whether it's a
configuration problem on the Bacula or firewall side -- I expected the
heartbeat to prevent this.

Anyhow, I'm not sure Bacula is the right solution for this type of
backup: the attribute despool has been running for nearly three days
straight now. Does anybody else have experience backing up tons of
files? Any recommendations on making it not suck?

-SH

------------------------------------------------------------------------------
vRanger cuts backup time in half-while increasing security.
With the market-leading solution for virtual backup and recovery, 
you get blazing-fast, flexible, and affordable data protection.
Download your free trial now. 
http://p.sf.net/sfu/quest-d2dcopy1
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users