On Tuesday 19 September 2006 16:39, Bill Moran wrote:
> In response to Bill Moran <[EMAIL PROTECTED]>:
> > 
> > My plan at this point:
> > 1) Upgrading to RELENG_6_1 tonight
> > 2) See if the problem repeats this Saturday
> > 3) If it does, grab another backtrace and put it out on freebsd-hackers@
> >    to see if anyone has any suggestions.
> 
> Followup posting.
> 
> I upgraded the system to FreeBSD 6.1p5.  There were some other circumstances
> that interfered with getting any reasonable data from the system, so it
> took a little longer than I expected to do the remaining steps.
> 
> However, it occurred to me that the log rotation occurs at about the same
> time this freeze would occur.  This involves sending Bacula a HUP signal.

Sending Bacula an HUP signal causes it to exit as rapidly as possible.  It 
attempts to cancel and cleanup jobs, but it does so in a rather brutal 
fashion.

Bottom line: make sure no root program sends HUP signals to Bacula while it is 
running a job, and preferably never send such signals unless you want to 
shutdown Bacula.

> 
> I moved the log rotation to later in the day (after the jobs are finished)
> and the whole set of backups ran without a hitch.  It would appear as if
> sending bacula-dir a HUP signal in the middle of a backup causes this
> problem.  Note that -HUPping the director while it is idle doesn't seem
> to cause any trouble.  Note also that I experimented a bit, and hupping
> doesn't seem to cause the problem all the time.
> 
> Can anyone verify this on another platform?  I want to set up some test
> scenarios here, but I'm dogged for time, and it may be a while before I
> can do any more serious work on this issue.
> 
> On a specific note: changing the time that logs rotate is only a workaround
> for us, and one with a dubious future.  We're running backups more and
> more frequently, and a lot of them run rather dynamically (using
> reschedule on error and the like) so it's going to get increasingly more
> difficult to find a time when I can be sure that no backup is running.

I don't understand why any process is sending HUP signals to Bacula when it is 
cycling logs.  This seems to me to be a "configuration" error.


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users

Reply via email to