Zitat von b_...@mail.ru: > On Nov 29, 2012, at 11:45 PM, lst_ho...@kwsoft.de wrote: > >> >> Zitat von b_...@mail.ru: >> >>> On Nov 29, 2012, at 11:05 PM, Dan Langille <d...@langille.org> wrote: >>> >>>> >>>> On Nov 29, 2012, at 2:26 PM, b_...@mail.ru wrote: >>>> >>>>> Hello, >>>>> >>>>> I have a couple of hosts which backup through bacula. I have >>>>> confronted with next situation, one host can't perform full >>>>> backup, process of backup always finished with this error: >>>>> Error: bsock.c:389 Write error sending 262144 bytes to Storage >>>>> daemon:IP ADDRESS:9103: ERR=Broken pipe >>>>> Fatal error: backup.c:1190 Network send error to SD. ERR=Broken pipe >>>>> Error: Director's comm line to SD dropped >>>>> Error: Bacula dir.bacula.HOSTNAME. 5.2.6 >>>>> >>>>> Time of occurrence of this error is not always the same, here >>>>> are the logs: >>>>> Elapsed time: 11 mins 56 secs >>>>> Elapsed time: 1 hour 10 mins 18 secs >>>>> Elapsed time: 47 mins 36 secs >>>>> Elapsed time: 1 hour 14 mins 40 secs >>>>> Elapsed time: 1 hour 40 mins 1 sec >>>>> >>>>> I don't think that this related to network issue. But client side >>>>> have a lot of I/O operation, disk system is busy almost all the >>>>> time. Maybe this is a cause. >>>>> How can I solve this? I don't see any timeout directives which >>>>> can help me. >>>>> >>>>> Client side is FreeBSD 7.2 amd64 bacula 5.2.6 >>>> >>>> HEADS UP. FreeBSD 7.2 was end-of-life'd in 2010 which means no >>>> security patches will be issued for it. Upgrading is recommended. >>>> >>>> see http://www.freebsd.org/security/#unsup >>>> >>>> That that matter, that's an old version of FreeBSD too. :) >>>> >>> I know, but upgrade in this case is impossible and this is not >>> related to our problem with bacula I think >>>>> Server side (DIR and SD on the same host) is FreeBSD 9.0 amd64 >>>>> bacula 5.2.6 (also tried 5.2.12 with the same result) >>>> >>>> Have you looked at trying the 'Heartbeat Interval' settings? >>>> >>>> You should try setting on the SD. There is also a 'Heartbeat >>>> Interval' on the FD, but that doesn't seem to be the error you're >>>> getting. >>> >>> yes, I have tried to play with "Heartbeat Interval". Unfortunately >>> doesn't help. >> > > this is not related to NIC in my case, the same issue with various > cards, OS. The same behaviour with FreeBSD, CentOS. Broadcom and > Intel cards. > I have no idea what could be a cause, anybody have have any thoughts? > >> Try with another NIC. We first had problems with our Bacula Server >> failing two clients out of ~20 with connection failures randomly. >> After ditching the Onboard GE (Marvell PHY) and using a PCIe NIC on >> the Server the problem went away. >>
It is a problem outside Bacula which simply uses a standard long-lived TCP connection. The amount of data transfered over a single connection for a long time does reveal subtile bugs in hardware and OS/drivers from time to time. So your only chance to solve is too swap related parts (NIcs, Switches, Router, cable ...) until the culprit is found. Regards Andreas ------------------------------------------------------------------------------ LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial Remotely access PCs and mobile devices and provide instant support Improve your efficiency, and focus on delivering more value-add services Discover what IT Professionals Know. Rescue delivers http://p.sf.net/sfu/logmein_12329d2d _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
