Regression: Bacula jobs fail with kernel 3.8-rc1

Regression: Bacula jobs fail with kernel 3.8-rc1

[Tilman Schmidt]

[-- Attachment #1: Type: text/plain, Size: 1281 bytes --]

With kernel 3.8-rc1, my nightly Bacula backup job reproducibly aborts
20 minutes into spooling with a "broken pipe" error on the network
socket connecting the file and storage daemons:

31-Dec 00:45 xenon-sd JobId 859: Spooling data ...
31-Dec 01:04 xenon-fd JobId 859: Error: bsock.c:429 Write error sending
65536 bytes to Storage daemon:xenon:9103: ERR=Broken pipe
31-Dec 01:04 xenon-fd JobId 859: Fatal error: backup.c:1200 Network send
error to SD. ERR=Broken pipe
31-Dec 01:04 xenon-dir JobId 859: Error: Director's comm line to SD dropped.

Returning to kernel 3.7.1 makes it work again. (Provided the crash
I reported in http://marc.info/?l=linux-kernel&m=135337052232021&w=2
doesn't hit, of course.)

Configuration:
Intel Pentium D, Intel DQ965GF mainboard, 6 GB RAM
onboard S-ATA controller driving two 500 GB S-ATA disks
and a Pioneer DVR-216D DVD-RW drive
Adaptec 29160B Ultra160 SCSI adapter driving a
Tandberg TS400 LTO-2 tape drive
openSUSE 11.4 userspace
Bacula 5.2.12, all daemons on the same machine, spooling to local disk

Regards,
Tilman

-- 
Tilman Schmidt                    E-Mail: tilman@imap.cc
Bonn, Germany
Diese Nachricht besteht zu 100% aus wiederverwerteten Bits.
Ungeöffnet mindestens haltbar bis: (siehe Rückseite)


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 261 bytes --]

Re: Regression: Bacula jobs fail with kernel 3.8-rc1

[Tilman Schmidt]

[-- Attachment #1: Type: text/plain, Size: 918 bytes --]

On 31.12.2012 16:07, /me wrote:
> With kernel 3.8-rc1, my nightly Bacula backup job reproducibly aborts
> 20 minutes into spooling with a "broken pipe" error on the network
> socket connecting the file and storage daemons:
[...]
> Returning to kernel 3.7.1 makes it work again.

Update/correction: in the meantime

- I had some jobs succeed on kernel 3.8-rc1,
  so the failure is not fully reproducible

- it happened with kernel 3.7.1 too, though much less frequently

- I went back to 3.6.11 where I the failure hasn't occurred so far

BTW the fact that the job spends 20 minutes spooling appears to be
part of the problem; a job which aborted after 20 mins on 3.7.1
completed in 3 mins on 3.6.11.

-- 
Tilman Schmidt                    E-Mail: tilman@imap.cc
Bonn, Germany
Diese Nachricht besteht zu 100% aus wiederverwerteten Bits.
Ungeöffnet mindestens haltbar bis: (siehe Rückseite)



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 260 bytes --]