Hi, 25.02.2008 15:37, HAWKER, Dan 2 (external) wrote: > > Hi All, > > Have a dedicated Bacula server (CentOS5.1) which has a HP MSA60 SAS > based drive array and a SAS based HP MSL2024 autoloader attached. All > the data that needs to be backed up is on the MSA and needs to go to > tape on the MSL2024 (LTO3), there is no network Bacula access. > > Have installed the contributed 2.2.7 EL5 RPMS from the site. The > Director, SD, FD and database (MySQL) are all local. Everything seems to > work as normal however after around an hour or so, the server > spontaneously reboots for no apparent reason.
Have you checked if, due to resource startvation, the server reboots itself? In such a case, you might find a hint in the machines log files - not necessarily at the OS level, but more likely in the firmware logs (IPMI or Lights-out management). I have seen such issues when it came to memory starvation - if activated, the hardware watchdog timer will be triggered in such circumstances. > I have tested the changer, etc with btape and all worked fine and due to > the cartridge size, etc, filled quite happily for considerably longer > than above (it took around 3-4hrs IIRC). Equally I have backed up around > 350GB directly to the drive using tar, which also took a few hours and > this was again fine (restored successfully too). > > As you can appreciate, am hence looking into Bacula to see if this is > cause of the problem, however the log file (/var/lib/bacula/log) is > light on details (it just says it started the job, but there was an > error) and there is nothing of real interest in syslog (again it just > notes that syslog has restarted). You could try to log the memory usage of the server, perhaps cpu usage as well. A simple script that prints a time stamp plus 'free' output, called by cron, is a good first step. Adding process-specific memory usage is also simple. And if you feed that data into a RRD file you can even create a nice graph of it :-) > Is there any way of upping the log verbosity somewhere to try and get > some more meaningful information??? Probably at the system level or even below that... > Also, does any of this sound familiar to anyone, is it a known bug for > instance??? Yes and no... see above. My experiences come from a (somewhat underpowered) HP server, by the way, with many jobs running in parallel. As that's probably not identical to your situation, I'd recommend to also observe the database - these can consume lots of memory, especially in conjunction with temporary table space in a RAM disk... but finding where the problem comes from should be done first. Hope this gets you started, Arno > TIA > > Dan > > -- > > Dan Hawker > Linux System Administrator > Astrium > http://www.astrium.eads.net > -- Arno Lehmann IT-Service Lehmann www.its-lehmann.de ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
