I am running on 2.6.9-11. All the PID's are different.
Arunav.
----- Original Message -----
From: "Kern Sibbald" <[EMAIL PROTECTED]>
To: "Arunav Mandal" <[EMAIL PROTECTED]>
Cc: <bacula-users@lists.sourceforge.net>
Sent: Friday, September 23, 2005 8:59 AM
Subject: Re: [Bacula-users] Bacula dir crashing
> On Wednesday 21 September 2005 12:08, Arunav Mandal wrote:
> > Hi,
> > You said to attach gdb to bacula-dir but there are 4 bacula-dir
> > running which one to attach to? In the documentation it says if I have
> > /lib/tls it may have some problems. I am running Centos 4.1. Anyway the
DIR
> > server has has only one cpu but was running smp kernel now I switched to
> > normail kernel.
>
> Uh, if there are 4 bacula-dir's running, then you must be running on a 2.4
> kernel, and in that case using /lib/tls will create no end of problems
with
> Bacula mostly bizarre hangs (missed signals). Under a 2.6 kernel, Bacula
> should appear as a single process.
>
> In the case you are using a 2.4 kernel, you either need to move /lib/tls,
or
> disable it with the environment variable as documented in the manual.
When
> using the debugger on a 2.4 kernel, you always attach to the first PID,
but
> if you follow the instructions in the manual, you will run Bacula under
the
> debugger rather than attaching to it later.
>
> In the case of a 2.6 kernel, if there are multiple instances of Bacula,
you
> are most likely using an option that shows all the threads -- as opposed
to
> running under a 2.6 kernel, they should all have the same PID, or you have
> disabled /lib/tls. This /lib/tls workaround is not needed for 2.6
kernels.
>
> >
> > Arunav.
> >
> > > Hello,
> > >
> > > I suspect that you have now set the record for having the most
problems
> >
> > with
> >
> > > Bacula, if not, you are close. Unfortunately, that is a rather
> > > unpleasant distinction :-(
> > >
> > > If you are asking if your FileSet is correct, I don't see any major
> >
> > problems.
> >
> > > However, now that you have moved the wildcards to the Exclude
resource, I
> > > personally would remove the "Exclude = yes" from the Options. It
should
> >
> > do
> >
> > > no harm, but it could make reading the FileSet confusing for someone
who
> > > doesn't know the history ...
> > >
> > > There is one known mutex race condition in 1.36.x that could cause a
> >
> > Director
> >
> > > hang, and perhaps you are more likely to see it than most users
because
> >
> > you
> >
> > > are running a *lot* of jobs every night, and if I am not mistaken, you
> >
> > have a
> >
> > > real smp system, which tends to make race conditions even more
evident.
> > >
> > > Note, I forgot to mention last time I emailed that if either the SD
*or*
> >
> > the
> >
> > > DIR crashes during a backup, the number of files on your tape is
likely
> > > to
> >
> > be
> >
> > > wrong (if the Director goes down the SD cannot update the catalog).
> > >
> > > If this happens again, please attach to the Director with the debugger
> >
> > using
> >
> > > something like:
> > >
> > > gdb bacula-dir <pid>
> > >
> > > where you replace <pid> with the PID of the Director, then produce a
> > > traceback as described in the Kaboom chapter of the manual. At least
I
> >
> > can
> >
> > > verify if you are seeing a known problem. This race bug is fixed in
> > > 1.37, but it was such a substantial fix that there is no patch for
1.36.
> > >
> > > On Friday 16 September 2005 10:50, Arunav Mandal wrote:
> > > > > On Wednesday 14 September 2005 08:45, Arunav Mandal wrote:
> > > > > > Now I got another problem bacula dir crashed without any reason.
> >
> > What
> >
> > > > debug
> > > >
> > > > > > level I should use to see what's going on?
> > > > >
> > > > > You should have gotten a traceback by email. If not, you can
produce
> >
> > one
> >
> > > > by
> > > >
> > > > > running the Director under the debugger as described in the kaboom
> > > > > chapter
> > > >
> > > > of
> > > >
> > > > > the manual.
> > > > >
> > > > > > Arunav.
> > > >
> > > > It happened again yesterday night bacula dir didnt crash it seems
but
> >
> > when
> >
> > > > I tried to log into it in morning via bconsole I can't and there
were
> > > > no backup mails also.I changed nothing in the config file expect the
> >
> > Fileset
> >
> > > > given below. Fileset was correct isnt?
> > > >
> > > > FileSet {
> > > > Name = linux-default
> > > > Ignore Fileset changes = yes
> > > > Include {
> > > > Options {
> > > > signature=SHA1
> > > > verify=pins1
> > > > onefs=no
> > > > sparse=no
> > > > Exclude = yes
> > > > }
> > > > File = /
> > > > }
> > > >
> > > > Exclude {
> > > > File = /sys
> > > > File = /proc
> > > > File = /tmp
> > > > File = /.journal
> > > > File = /.fsck
> > > > File = /mnt
> > > > File = /dev
> > > > File = /var/chroot/hoary-ia32/home
> > > > File = /space
> > > > File = *.mp3
> > > > File = *.m4a
> > > > File = *.o
> > > > File = *.obj
> > > > File = *.vob
> > > > File = *.VOB
> > > > File = *.journal
> > > > File = *.fsck
> > > >
> > > > }
> > > > }
> > > >
> > > >
> > > >
> > > > Arunav.
> > >
> > > --
> > > Best regards,
> > >
> > > Kern
> > >
> > > (">
> > > /\
> > > V_V
>
> --
> Best regards,
>
> Kern
>
> (">
> /\
> V_V
>
-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server.
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
