From: Sebastian S. <st...@tu...> - 2006-06-29 12:10:37
|
On 26.06.2006, at 13:51, Sebastian Stark wrote: > > On 26.06.2006, at 13:22, Kern Sibbald wrote: > >> On Monday 26 June 2006 10:48, Sebastian Stark wrote: >>> Just a note: I found these entries in the kernel log on my solaris >>> box which are likely to be related to the director hangs: >>> >>> Jun 23 14:48:49 yangtse xntpd[400]: [ID 774427 daemon.notice] time >>> reset (step) 0.135391 s >>> Jun 24 09:10:24 yangtse xntpd[400]: [ID 774427 daemon.notice] time >>> reset (step) 0.132294 s >>> Jun 25 04:00:52 yangtse xntpd[400]: [ID 774427 daemon.notice] time >>> reset (step) 0.138633 s >>> Jun 25 23:13:49 yangtse xntpd[400]: [ID 774427 daemon.notice] time >>> reset (step) 0.139171 s >> >> This is possible, but the Director (scheduler) really should not do >> anything >> unless it notes a shift of more than 10 seconds within any 1 minute >> period. > > Maybe there are more shifts that do not occur in the log for some > reason. Anyway, the director did not only hang sometimes , it also > "forgot" some jobs that should have been run! Since the upgrade to > 1.38.10 not all of the jobs that were supposed to run were actually > showing up in the "Scheduled jobs" listing. > > I applied the patch and wait until tomorrow to see if things changed. Things changed. With this patch my problem went away. Sebastian |