[GE users] qrsh, openssh & hanging jobs

Kirk Patton kpatton at transmeta.com
Tue Dec 6 20:00:17 GMT 2005


On Tue, Dec 06, 2005 at 08:26:49PM +0100, Daniel Templeton wrote:
> Kirk,
> 
> I think it's the nature of the beast.  The folks who are likely to be
> interested in security are likely to be running N1GE, i.e. the supported
> version.  There has been discussion on the topic, but it's so far been
> internal.
> 
> It is a very odd issue, and it took quite a while to pin it down to
> where we could determine what was going wrong.  I still don't really
> understand what's broken, just that it's not our fault, and it works on
> Solaris 10. :)
> 
> Daniel

The reason I used ssh for interactive jobs is the pseduo terminal support.
Security was a bonus. :-)

Using the default rsh transport, interactive support in SGE was not very
elegant.  

The backport I received from Sun seems to have fixed the issue for me.  I
am just supprised there has not been more public discussions on the list.
Having a terminal hang is a really noticible issue.

Kirk

> 
> Kirk Patton wrote On 12/06/05 17:21,:
> 
> >Currently, we are using Solaris 8.  I am testing Solaris 9 and trying to
> >get it ready for rollout.  In speaking with Sun, there is a backport of
> >the fixes needed from Solaris 10.  But, at the moment I am running into
> >circular patch incompatibilites.
> >
> >I was suprised that I was not able to locate any chatter in the mail
> >lists about the problem...  I have not been able to reproduce the hang
> >outside of SGE/qrsh jobs.  Straight 'ssh command' works without incident.
> >
> >Solaris 10 is still pretty new.  I would expect that there are a few 
> >sites running Solaris8/9 + SSH + SGE.  I wonder why there has not been
> >more dicussion about this issue.  Perhaps I am not using the correct
> >search terms.
> >
> >Kirk
> >
> >On Tue, Dec 06, 2005 at 05:14:21PM +0100, Daniel Templeton wrote:
> >  
> >
> >>Kirk,
> >>
> >>Are you running Solaris 9?  If so, the answer appears to be either to
> >>upgrade to Solaris 10 or to borrow the sshd from Solaris 10.
> >>
> >>Daniel
> >>
> >>Kirk Patton wrote On 12/06/05 17:01,:
> >>
> >>    
> >>
> >>>My interactive jobs are hanging when they are submitted using qrsh. I have my 
> >>>cluster configured to use openssh.  Sun support tells me this is an Openssh
> >>>issue.
> >>>
> >>>I have not been able to reproduce the hang when using ssh outside of sge.  I
> >>>am wondering if this is a common issue when using SGE and openssh, and how
> >>>other are dealing with the problem.
> >>>
> >>>Thanks,
> >>>Kirk
> >>>
> >>>
> >>> 
> >>>
> >>>      
> >>>
> >>-- 
> >>***************************************************
> >>*        Daniel Templeton   UMPK18 x83749         *
> >>*       Staff Engineer, Sun N1 Grid Engine        *
> >>***************************************************
> >>*   An "intellectual" is a man who takes more     *
> >>* words than he needs to say more than he knows.  *
> >>*                       -Dwight D. Eisenhower     *
> >>***************************************************
> >>
> >>
> >>
> >>---------------------------------------------------------------------
> >>To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
> >>For additional commands, e-mail: users-help at gridengine.sunsource.net
> >>
> >>    
> >>
> >
> >  
> >
> 
> -- 
> ***************************************************
> *        Daniel Templeton   UMPK18 x83749         *
> *       Staff Engineer, Sun N1 Grid Engine        *
> ***************************************************
> *   An "intellectual" is a man who takes more     *
> * words than he needs to say more than he knows.  *
> *                       -Dwight D. Eisenhower     *
> ***************************************************
> 
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
> For additional commands, e-mail: users-help at gridengine.sunsource.net
> 

-- 
Kirk Patton
Unix Administrator
Transmeta Inc.
Tel. 408 919-3055

---------------------------------------------------------------------
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