>>>>> On Thu, 13 Apr 2006 15:04:41 -0400, Ian Levesque <[EMAIL PROTECTED]> said: > > Hi, > > On Apr 13, 2006, at 12:34 PM, Martin Simmons wrote: > > > >>> Did you try using spooling? > >>> > >>> > >> > >> Yes, it puts an enormous load on the backup server; I'm not sure why. > >> I do utilize spooling for incrementals, but I expect them to be > >> slower (which they are). > >> > > > > Spooling splits the backup into phases of collecting from the > > client, writing > > to the tape (repeated) and then writing to the database. > > > > It could be worthwhile investigating the spooling slowness to see > > if it occurs > > in one particular phase (look at the console output for spooling > > messages). > > Without spooling it might be difficult to detect which phase is > > slow because > > they are interleaved for each file sent by the client. > > > > At the risk of veering this discussion off-topic, I've seen that > during the spooling to disk phase, my server becomes very slow to > respond, though the CPU usage is still relatively light. I've put > some of the cacti graphs from the server online if anyone feels like > checking them out: http://sbgrid.org/bacula/ > > Last night our daily incrementals ran as usual, and one can see in > the graphs that CPU usage never exceeded 50%, whereas the load > average peaked at around 5. > > The incrementals that ran last night: > > 13-Apr 00:00 sbgrid-dir: Start Backup JobId 84, > Job=EML2.2006-04-13_00.00.00 > 12-Apr 23:51 eml2-fd: DIR and FD clocks differ by -630 seconds, FD > automatically adjusting. > 13-Apr 00:02 sbgrid-sd: Spooling data ... > 13-Apr 00:08 sbgrid-sd: Committing spooled data to Volume "000003L2". > Despooling 27,165,575 bytes ... > 13-Apr 00:08 sbgrid-sd: Sending spooled attrs to the Director. > Despooling 389,684 bytes ... > > 13-Apr 01:00 sbgrid-dir: Start Backup JobId 85, Job=CRYSTAL_HOME. > 2006-04-13_01.00.00 > 13-Apr 00:49 home-fd: DIR and FD clocks differ by -635 seconds, FD > automatically adjusting. > 13-Apr 01:00 sbgrid-sd: Spooling data ... > 13-Apr 01:20 sbgrid-sd: Committing spooled data to Volume "000005L2". > Despooling 6,617,884,648 bytes ... > 13-Apr 01:28 sbgrid-sd: Sending spooled attrs to the Director. > Despooling 1,471,546 bytes ... > > 13-Apr 02:00 sbgrid-dir: Start Backup JobId 86, Job=FS4_XRAY. > 2006-04-13_02.00.00 > 13-Apr 01:49 xray-fd: DIR and FD clocks differ by -634 seconds, FD > automatically adjusting. > 13-Apr 02:00 sbgrid-sd: Spooling data ... > 13-Apr 02:42 sbgrid-sd: Committing spooled data to Volume "000005L2". > Despooling 10,782,446,112 bytes ... > 13-Apr 02:53 sbgrid-sd: Sending spooled attrs to the Director. > Despooling 572,734 bytes ... > > 13-Apr 03:00 sbgrid-dir: Start Backup JobId 87, > Job=EML3.2006-04-13_03.00.00 > 13-Apr 02:49 eml3-fd: DIR and FD clocks differ by -637 seconds, FD > automatically adjusting. > 13-Apr 03:00 sbgrid-sd: Spooling data ... > 13-Apr 03:09 sbgrid-sd: Committing spooled data to Volume "000005L2". > Despooling 152,612,206 bytes ... > 13-Apr 03:09 sbgrid-sd: Sending spooled attrs to the Director. > Despooling 97,522 bytes ...
Are these times the same as those in the graphs? It is strange that outbound traffic is high between 00:00 and 00:30. It looks like something else is accessing the server overnight (presumably it was also doing this between 22:45 and 23:15). I assume the spool disk is local. Simultaneuously reading and writing different parts of the same disk is a good way to boost the load average. :-) __Martin ------------------------------------------------------- This SF.Net email is sponsored by xPML, a groundbreaking scripting language that extends applications into web and mobile media. Attend the live webcast and join the prime developer group breaking into this new coding territory! http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642 _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users