[GE users] Troubleshooting NIS errors (SGE 6.1u3 / Linux)

Chris Dagdigian dag at sonsorol.org
Tue Jan 22 13:55:04 GMT 2008

This is really interesting and may tie into an odd symptom we saw ...

A coworker of mine noticed that the NIS problems went away after  
blowing away the SGE install and re-creating it from our autoinstall  
templates. He scripted it so that the entire SGE install could be torn  
down and rebuilt in a few seconds. It reliably "fixed" the NIS issue  
during testing.

What this means to me is that all of the various Linux and NIS tweaks  
we had made (with no positive effect) may have actually worked had we  
been more careful about restarting SGE each time we made a change to a  
NIS or system authentication/authorization setting.

We are still using getent at this point but we have added a new rule  
to our informal body of knowledge:

  - Plan on restarting SGE (or just the sge_execd daemons) after any  
NIS or auth settings are altered

Joe's comment about ncsd is interesting as well. I'll look into that.



On Jan 21, 2008, at 11:40 PM, Christopher Heiny wrote:

> On Thursday 10 January 2008, ground control picked up the following
> transmission from Chris Dagdigian:
>> Ken, Chansup, Reuti -- thanks for all your help
> Chris,
> Did you ever find a permanent solution to the problem, or are you  
> still
> using getent?
> I've encountered it occasionally here, and it appears to relate to
> whether NIS is working correctly when sge_execd starts up.  I've found
> that if ypwhich shows there's no binding, then
>    /etc/init.d/sgeexecd stop
>    /etc/init.d/ypbind start
>    /etc/init.d/sgeexecd start
> will make the problem go away on that particular node.  Oddly enough,
> simply bringing up YP again does not resolve the issue.
> Of course, this may not work for you (especially if you have more  
> than a
> couple of nodes) and it isn't any kind useful solution anyway.  But it
> is a clue as to what might actually be going wrong.

To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net

More information about the gridengine-users mailing list