Bacula-users

Re: [Bacula-users] Failed backup: Network send error to SD. Broken Pipe

2008-06-10 06:00:12
Subject: Re: [Bacula-users] Failed backup: Network send error to SD. Broken Pipe
From: Arno Lehmann <al AT its-lehmann DOT de>
To: bacula-users AT lists.sourceforge DOT net
Date: Tue, 10 Jun 2008 11:59:44 +0200
Hello,

10.06.2008 09:12, Christian Nolte wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> 
> Christian Nolte schrieb:
>> Christian Nolte wrote:
>>> Christian Nolte wrote:
>>>> Josh Fisher wrote:
>>>>> Christian Nolte wrote:
>>>>>> John Drescher wrote:
>>>>>>  
>>>>>>>> Does no one have a clue what could be wrong with my configuration?
>>>>>>>> Anything that I might try to get more information about the problem?
>>>>>>>>
>>>>>>>>       
>>>>>>> My first guess is a faulty NIC or switch.
>>>>>>>     
>>>>>> Unfortunately I have no access to the hardware outside our company. The
>>>>>> system to be backed up is hosted by a provider (STRATO in Germany).
>>>>>>
>>>>>> The connection in our company is working fine. No problems with large
>>>>>> downloads or something like that.
>>>>>>
>>>>>> Is there a way to get further information from bacula so that I can
>>>>>> debug this issue?
>>>>>>   
>>>>> Yes. Run bacula-sd manually, setting the debug level to 150 and
>>>>> capturing debug output to a file. Something like:
>>>>> touch /var/lib/bacula/sdlog
>>>>> chown bacula /var/lib/bacula/sdlog
>>>>> chgrp disk /var/lib/bacula/sdlog
>>>>> bacula-sd -d150 -c /etc/bacula/bacula-sd.conf -u bacula -g disk
>>>>> &>/var/lib/bacula/sdlog
>>>> Thanks, Josh. With your information I was able to get the storage daemon
>>>> log below. What puzzles me is, that this seems not to be a network issue
>>>> because the log indicates that there is an issue with writing the data
>>>> the FD sent to the SD. I just have checked the partition and there are
>>>> no errors in the file system. I will try another partition to store the
>>>> backup.
>>>> Here is the excerpt from the logfile:
>>>> fc6_32_build2-sd: askdir.c:351 >dird: fc6_32_build2-sd: askdir.c:357
>>>> Didn't get vol info vol=Full-Ext-Vol-0001: ERR=Network error on
>>>> bnet_recv in req_vol_info.

If this is your longest-running job, you might have found a network 
issue where the connection between SD and DIR is dropped after some 
time of inactivity. Firewalls and routers sometimes do this...

>>>> fc6_32_build2-sd: block.c:748 dir_update_volume_info terminate writing
>>>> -- ERROR
>>>> fc6_32_build2-sd: block.c:771 Leave terminate_writing_volume -- ERROR
>>>> fc6_32_build2-sd: append.c:220 Got write_block_to_dev error on device
>>>> "FileStorageExternal" (/mnt/backup-store/backups).
>>>> fc6_32_build2-sd:
>>>> fd_cmds.c:160 <filed:
>>> I get the same error message when using an other physical device to back up:
>>> ---
>>> fc6_32_build2-sd: askdir.c:351 >dird: fc6_32_build2-sd: askdir.c:357
>>> Didn't get vol info vol=Full-Vol-0005: ERR=Network error on bnet_recv in
>>> req_vol_info.
>>> fc6_32_build2-sd: block.c:748 dir_update_volume_info terminate writing
>>> -- ERROR
>>> fc6_32_build2-sd: block.c:771 Leave terminate_writing_volume -- ERROR
>>> fc6_32_build2-sd: append.c:220 Got write_block_to_dev error on device
>>> "FileStorage" (/mnt/backup-store2/backups).
>>> ---
>>> As mentioned: Other backups work fine on the same devices. This specific
>>> backup *always* stops after 2,991,804,324 bytes have been written...
>>
>> If I run the full backup the SD does not write any bytes. I have tested
>> a small test set (/etc/bacula, about 34MB) and this backup does work fine.
>>
>> Are there any things I could try to narrow this issue down?
>>
> 
> Ok. I did try something else: I have disabled the compression for the
> file set. The backup finished correctly (about 47 GB of data). I have
> reason to believe that the compression is the problem here, but I do not
> know what exactly goes wrong here. Other hosts however work fine with
> the compression enabled.

Which versions of FD, SD and DIR are involved here?

Also, can you verify the network connection between the daemons stays 
open while the job runs?

Arno

-- 
Arno Lehmann
IT-Service Lehmann
www.its-lehmann.de

-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
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>