[GE users] transfer queues again: attribute problem

Ivan R. Judson judson at mcs.anl.gov
Tue Dec 13 18:39:07 GMT 2005


Hi Charu,

No problem with respect to the 5.3/6.0 updates. My biggest issue is getting
the "big" picture. Ie, what's supposed to running and configured where. Once
I get this working, I might write up an extension to your howto with a
concrete 2 cluster example (I think that would have helped me a lot).

If you have a quick summary that'd be great. Here's what I have:

Cluster 1
----------
Scripts in /cluster/sge (cell: default)
Transfer queue defined on qmaster node
Qmaster node is submit and admin host for cluster 2
Mpk27jobs (I know, I can change it later ;-) defined as a global attribute

Cluster 2
----------
Scripts in /cluster/sge (cell: default)
Qmaster node is submit and admin host for cluster 1


I've gotten a bit further, now I have this error:

transfer BP    0/20      0.03     darwin        a
$ qstat -j
scheduling info:            queue instance "transfer" dropped because it is
overloaded: no value for complex attribute "mpk27jobs"

Thanks for helping with this.

--Ivan

> -----Original Message-----
> From: Charu.Chaubal at Sun.COM [mailto:Charu.Chaubal at Sun.COM]
> Sent: Tuesday, December 13, 2005 12:17 PM
> To: users at gridengine.sunsource.net
> Subject: Re: [GE users] transfer queues again: attribute problem
> 
> Hi Ivan,
> 
> [ The transfer queue HOWTO is based on GE 5.3, so it's a bit outdated...
> but some simple modifications should be enough to have it working with GE6
> ]
> 
> "mpk27jobs" is a global resource which indicates the number of jobs
> pending on the "remote" site (btw, MPK27 was the building housing the
> remote cluster when I wrote the HOWTO....  obviously you should use a
> name that makes sense for your environment).  You need to create this
> resource as a global consumable, and then modify the given load sensor
> to provide the value for it.
> 
> Regards,
> 	Charu
> 
> 
> Ivan R. Judson wrote On 12/13/05 09:53,:
> >
> >
> > I've been reworking the transfer queues, and I've gotten this far:
> >
> >
> >
> > ------------------------------------------------------------------------
> ----
> >
> > transfer-to-nwu at host1 BP    0/20      0.02     darwin        a
> >
> >
> >
> > # qstat -j
> >
> > scheduling info:            queue instance "transfer-to-nwu at host1"
> > dropped because it is overloaded: no such complex attribute for
> > threshold "mpk27jobs"
> >
> >
> >
> > I have previously made this error go away, but I forget now if I need a
> > global attribute or one on the host1 queue. And once I create it, do I
> > have to restart execd?
> >
> >
> >
> > --Ivan
> >
> 
> --
> ####################################################################
> # Charu V. Chaubal              # Phone: (650) 786-7672 (x87672)   #
> # Grid Computing Technologist   # Fax:   (650) 786-4591            #
> # Sun Microsystems, Inc.        # Email: charu.chaubal at sun.com     #
> ####################################################################
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
> For additional commands, e-mail: users-help at gridengine.sunsource.net


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