Bacula-users

Re: [Bacula-users] Win32 FD / Write error sending N bytes to Storage daemon

2011-06-06 14:00:28
Subject: Re: [Bacula-users] Win32 FD / Write error sending N bytes to Storage daemon
From: Josh Fisher <jfisher AT pvct DOT com>
To: bacula-users AT lists.sourceforge DOT net
Date: Mon, 06 Jun 2011 13:56:58 -0400
On 6/6/2011 12:46 PM, Yann Cézard wrote:
> Le 20/05/2011 10:22, Yann Cézard a écrit :
>> Hi everyone,
>>
>> Since a few weeks, I am facing a really strange problem with
>> my win32 bacula-fd.
>>
>> It seems that the problem started when I upgraded my SD + DIR
>> to the 5.0.X (I was still using the 2.4.4 until that time).
>> The problem is that almost (this is only observable on Full backups
>> of several GB) all my win32-fd Full backups now fails with the 
>> following errors :
>> - with a 2.4.4 client :
>> 12-May 22:22 msadpau-fd JobId 2538: Generate VSS snapshots. Driver="VSS Win 
>> 2003", Drive(s)="E"
>> 13-May 00:00 msadpau-fd JobId 2538: Fatal error: ../../filed/backup.c:892 
>> Network send error to SD. ERR=Input/output error
>> 13-May 00:02 msadpau-fd JobId 2538: VSS Writer (BackupComplete): "System 
>> Writer", State: 0x1 (VSS_WS_STABLE)
>> [...]
>> 13-May 00:02 msadpau-fd JobId 2538: VSS Writer (BackupComplete): "NTDS", 
>> State: 0x1 (VSS_WS_STABLE)
>> 13-May 00:02 backuppa-sd JobId 2538: JobId=2538 
>> Job="msad-stockage-pau.2011-05-12_22.00.00_37" marked to be canceled.
>> 13-May 00:02 backuppa-sd JobId 2538: Job write elapsed time = 01:39:44, 
>> Transfer rate = 6.296 M Bytes/second
>> 13-May 00:02 backuppa-sd JobId 2538: Error: bsock.c:518 Read error from 
>> client:10.1.2.17:36643: ERR=Connection reset by peer
>> 13-May 00:02 backuppa-dir JobId 2538: Error: Bacula backuppa-dir 5.0.2 
>> (28Apr10): 13-May-2011 00:02:15
>>    Build OS:               x86_64-pc-linux-gnu debian squeeze/sid
>>    Backup Level:           Full
>>    Client:                 "BLABLABLA" 2.4.4 (28Dec08) 
>> Linux,Cross-compile,Win32
>> - after upgrading the client to 5.0.3 (error message is more verbose, 
>> but
>> the problem is still there) :
>> 16-May 17:22 msadpau-fd JobId 2552: Generate VSS snapshots. Driver="VSS Win 
>> 2003", Drive(s)="E"
>> 16-May 20:14 msadpau-fd JobId 2552: Error: 
>> /home/kern/bacula/k/bacula/src/lib/bsock.c:393 Write error sending 65562 
>> bytes to Storage daemon:backuppa:9103: ERR=Input/output error
>> 16-May 20:14 msadpau-fd JobId 2552: Fatal error: 
>> /home/kern/bacula/k/bacula/src/filed/backup.c:1024 Network send error to SD. 
>> ERR=Input/output error
>> 16-May 20:15 msadpau-fd JobId 2552: Error: 
>> /home/kern/bacula/k/bacula/src/lib/bsock.c:339 Socket has errors=1 on call 
>> to Storage daemon:backuppa:9103
>> 16-May 20:16 msadpau-fd JobId 2552: VSS Writer (BackupComplete): "System 
>> Writer", State: 0x1 (VSS_WS_STABLE)
>> [...]
>> 16-May 20:16 msadpau-fd JobId 2552: VSS Writer (BackupComplete): "WMI 
>> Writer", State: 0x1 (VSS_WS_STABLE)
>> 16-May 20:16 backuppa-sd JobId 2552: JobId=2552 
>> Job="msad-stockage-pau.2011-05-16_17.22.21_58" marked to be canceled.
>> 16-May 20:16 backuppa-sd JobId 2552: Job write elapsed time = 02:53:20, 
>> Transfer rate = 6.234 M Bytes/second
>> 16-May 20:16 backuppa-sd JobId 2552: Error: bsock.c:518 Read error from 
>> client:10.1.2.17:36643: ERR=Connection reset by peer
>> 16-May 20:16 backuppa-dir JobId 2552: Error: Bacula backuppa-dir 5.0.2 
>> (28Apr10): 16-May-2011 20:16:02
>>    Build OS:               x86_64-pc-linux-gnu debian squeeze/sid
>>    Backup Level:           Full
>>    Client:                 "BLABLABLA" 5.0.3 (04Aug10) 
>> Linux,Cross-compile,Win32
>> - after upgrading the DIR/SD from 5.0.2 to 5.0.3 (Debian squeeze => 
>> wheezy) :
>> 19-mai 10:07 msadpau-fd JobId 2565: Generate VSS snapshots. Driver="VSS Win 
>> 2003", Drive(s)="E"
>> 19-mai 10:09 msadpau-fd JobId 2565: Error: 
>> /home/kern/bacula/k/bacula/src/lib/bsock.c:393 Write error sending 65536 
>> bytes to Storage daemon:backuppa:9103: ERR=Input/output error

Try setting "Maximum Network Buffer Size" to 32768 in both the storage 
daemon and client configs. It looks like something doesn't like the 
default 65536 buffer size.

Also, a backup job can run a long time, and Bacula keeps the TCP 
connections open the entire time. Make sure a switch isn't timing out a 
TCP connection when it shouldn't. You may also want to look up the 
"Heartbeat Interval" setting in the Bacula docs.

...

------------------------------------------------------------------------------
Simplify data backup and recovery for your virtual environment with vRanger.
Installation's a snap, and flexible recovery options mean your data is safe,
secure and there when you need it. Discover what all the cheering's about.
Get your free trial download today. 
http://p.sf.net/sfu/quest-dev2dev2 
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users