[GE users] [OT] qlicserver + lmutil = kernel panic

Olesen, Mark Mark.Olesen at emcontechnologies.com
Tue Nov 25 09:36:14 GMT 2008


> Hi Mark,
> 
> I have more tests to submit :
> - running on the qmaster host / perl 5.8.0 : crashes
> - running on another host / perl 5.8.5 : works like a charm
> 
> I will continue to dig this.

Hi Ionel,

If you don't mind hacking source code a tiny bit, you might try this to
see if Perl threads in 5.8.0 are causing the problem.

Near the top of qlicserver, you'll find this line:

   use threads;    ## REMOVE FOR UNTHREADED PERL

And much further down, in the license query itself, you'll find this
line:

    ## REMOVE REMAINDER FOR UNTHREADED PERL

The remainder being the rest of the sub, not the rest of the file!

See if that localizes the problem. Also check what 'perl -V' reports for
each version.
My version reports the compile-time option: USE_ITHREADS
Your 5.8.0 may have a different (buggier) threads implementation.


/mark


> Olesen, Mark wrote:
> >> I find in qlicserver a friend to link flexlm-secured features to
> SGE.
> >> Last week, the machine running qlicserver crashed in a kernel panic
> >> referencing 'lmutil'.
> >> I restarted the machine, restarted qlicserver and boom, it crashed
> the
> >> next morning with a kernel panic.
> >>
> >> Finally I restarted the machine, let qlicserver down, and the
> machine
> >> is up and running for a few days.
> >>
> >> I suspect the fact that I configured qlicserver to query two Flexlm
> >> servers with different versions :
> >> - one is running FlexLM 9.5.0
> >> - the other is running FlexLM 10.8.2.1
> >> Queries are made with lmutil v9.5.0
> >
> > Our main query (via qlicserver) is with v9.5.0.
> > The various vendor daemons are running v8.4, v9.2, v10.1, v10.8.
> >
> > From the one license server that I can access, the lmgrd version is
> > mostly v10.8.0.7, but one vendor daemon is actually running via
> > v10.8.0.0
> >
> > I would thus conclude that mixing versions is probably okay.
> > I would look for another issue: DNS timeouts and network problems
> are
> > typically what make our life difficult. Check that the reverse host
> > lookup is providing something sensible (we had a windows DHCP server
> > that incorrectly reverse mapped a bunch of our cluster names and fed
> > these to the DNS server!).
> >
> >
> > To exclude that the problem is with qlicserver and/or GridEngine
> itself,
> > you could write a small infinite loop and monitor the results for
> the
> > next day, eg:
> >
> > while :
> > do
> >    date >> lmutil.log 2>&1
> >    lmutil lmstat query ...  >> lmutil.log 2>&1
> >    sleep 60
> > done
> >
> >
> > /mark
> >
> > ------------------------------------------------------
> >
> http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=38&dsMessa
> geId=89214
> >
> > To unsubscribe from this discussion, e-mail: [users-
> unsubscribe at gridengine.sunsource.net].
> 
> ------------------------------------------------------
> http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=38&dsMessa
> geId=89754
> 
> To unsubscribe from this discussion, e-mail: [users-
> unsubscribe at gridengine.sunsource.net].
This e-mail message and any attachments may contain 
legally privileged, confidential or proprietary Information, 
or information otherwise protected by law of EMCON 
Technologies, its affiliates, or third parties. This notice 
serves as marking of its "Confidential" status as defined 
in any confidentiality agreements concerning the sender 
and recipient. If you are not the intended recipient(s), 
or the employee or agent responsible for delivery of this 
message to the intended recipient(s), you are hereby 
notified that any dissemination, distribution or copying 
of this e-mail message is strictly prohibited. 
If you have received this message in error, please 
immediately notify the sender and delete this e-mail 
message from your computer.

------------------------------------------------------
http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=38&dsMessageId=89759

To unsubscribe from this discussion, e-mail: [users-unsubscribe at gridengine.sunsource.net].



More information about the gridengine-users mailing list