RE: cfenvd dumping core

[Luke Youngblood]

If you still have the core file on your system, and can install the GNU
debugger, it might help Mark to fix the bug so that other people don't have
to deal with it.

Luke

-----Original Message-----
From: help-cfengine-bounces+lyoungblood=phonechargeinc.com@gnu.org
[mailto:help-cfengine-bounces+lyoungblood=phonechargeinc.com@gnu.org] On
Behalf Of PAUL WILLIAMSON
Sent: Friday, June 03, 2005 8:35 AM
To: help-cfengine@gnu.org
Subject: RE: cfenvd dumping core

Deleting the db files on the policy master seems to have eliminated 
the problem.  cfexecd is no longer dumping a core file while starting 
from the cfagent command on the local client.

Paul

>>> "Mark Burgess" <Mark.Burgess@iu.hio.no> 06/03/05 2:49 AM >>>
Can you do a backtrace?

gdb /var/cfengine/cfexecd core 
gdb> back

M

On Thu, 2005-06-02 at 16:11 -0400, PAUL WILLIAMSON wrote:
> I just tried to delete all my db files on one of the clients...
> 
> I ran cfagent -v on the client (I did not delete the db on the 
> policy master) and it left a big core dump.  
> 
> $ file core
> core:           ELF 32-bit MSB core file SPARC Version 1, from
> 'cfexecd'
> 
> So, it looks like cfexecd is core dumping.  Interestingly, my 
> cfagent file is configured to restart cfenvd, cfservd and cfexecd
> if necessary:
> 
>         "cfenvd" restart "/usr/local/sbin/cfenvd"
>         "cfservd" restart "/usr/local/sbin/cfservd"
>         "cfexecd$" restart "/usr/local/sbin/cfexecd"
> 
> If I remove cfexecd$, I get no core dump.  Is this the correct format

> for starting up cfexecd?
> 
> Paul
> 
> >>> "Mark Burgess" <Mark.Burgess@iu.hio.no> 06/02/05 3:36 PM >>>
> 
> This is a possible explanation for the problem. One might also try
> downloading the latest snapshot from www.cfengine.org where a few
> patches are applied.
> 
> Certainly I cannot understand why cfenvd should cause cfexecd to
> crash.
> They are utterly independent. The only conceivable link between them
> is
> Berkeley DB.
> 
> Mark
> 
> On Thu, 2005-06-02 at 15:07 -0400, Luke Youngblood wrote:
> > Have you used two or more different compiled binaries on the same
> box
> > without deleting DB files in between runs?
> > 
> > Initially, I was using the 2.1.13 version of Cfengine from
> SunFreeware.  I
> > decided to switch to Blastwave's version of 2.1.14, which is
> compiled
> > against a different version of BerkeleyDB.
> > 
> > What I found was that Cfagent and Cfenvd were throwing lots of
> errors
> > related to database issues.  Cfenvd would core dump every time I
> started it.
> > 
> > In order to resolve these issues, I removed all of the databases
> that
> > Cfagent and Cfenvd were complaining about (run cfagent -v), then
let
> cfagent
> > recreate them on its next run.  I didn't have any problems after
> that.
> > 
> > Just to be sure and cleanup properly, you might first kill cfenvd
> and
> > cfservd, then nuke everything under /var/cfengine, and perform a
run
> of
> > cfagent again.  This should reinitialize all the databases and
> everything
> > should work fine from then on.
> > 
> > It seems that different versions of BerkeleyDB create files that
are
> > slightly different in format.
> > 
> > Hope this helps,
> > 
> > Luke
> > 
> > -----Original Message-----
> > From: help-cfengine-bounces+lyoungblood=phonechargeinc.com@gnu.org

> >
[mailto:help-cfengine-bounces+lyoungblood=phonechargeinc.com@gnu.org]

> On
> > Behalf Of PAUL WILLIAMSON
> > Sent: Thursday, June 02, 2005 2:26 PM
> > To: Darryl.Baker@gedas.com; help-cfengine@gnu.org 
> > Subject: Re: cfenvd dumping core
> > 
> > I have almost the same situation.  Although I get a decent core 
> > almost all the time.  I have both binaries I've compiled (with the

> > same versions of tools as you, but only sol8) and ones from 
> > someone on the list.  Either way, running remotely always 
> > cores out cfexec or cfagent.
> > 
> > Paul
> > 
> > >>> "Baker, Darryl" <Darryl.Baker@gedas.com> 06/02/05 1:51 PM >>>
> > Gentlefolk,
> >     I'm having a problem where cfenvd is Seg faulting and sometimes
> > taking cfexecd with it though I don't get a core from cfexecd. It
> > doesn't
> > happen with any consistency or pattern. Some machines will have it
> > happen
> > every other day others it will happen once a week. I have 134
> machines
> > and I
> > can go 16 hours without a failure and then have 5 fail in the next
4
> > hours.
> > 
> > This is with 2.1.14 on Solaris 8 and 9 boxes compiled with gcc
3.3.2
> > and
> > linked with OpenSSL 0.9.7. The same version of OpenSSL is on all
> > machines.
> > 
> >
>
_____________________________________________________________________
> > Darryl Baker
> > Senior Unix Specialist
> > gedas USA, Inc.
> > Operational Services Business Unit
> > 3800 Hamlin Road
> > Auburn Hills, MI 48326
> > US
> > phone       +1-248-754-5341
> > fax +1-248-754-6399
> > Darryl.Baker@gedas.com 
> > http://www.gedasusa.com 
> >
>
_____________________________________________________________________
> > 
> > 
> > 
> > 
> > _______________________________________________
> > Help-cfengine mailing list
> > Help-cfengine@gnu.org 
> > http://lists.gnu.org/mailman/listinfo/help-cfengine 
> > 
> > 
> > 
> > _______________________________________________
> > Help-cfengine mailing list
> > Help-cfengine@gnu.org 
> > http://lists.gnu.org/mailman/listinfo/help-cfengine 
> 
> 
> 
> 
> 
> _______________________________________________
> Help-cfengine mailing list
> Help-cfengine@gnu.org 
> http://lists.gnu.org/mailman/listinfo/help-cfengine 





_______________________________________________
Help-cfengine mailing list
Help-cfengine@gnu.org
http://lists.gnu.org/mailman/listinfo/help-cfengine