Re: [Condor-users] Unable to Flock jobs to a remote sched

[Jaime Frey <jfrey@xxxxxxxxxxx>]

First, we should clear up how condor names things. The matching of jobs to
resources in a condor pool is done by the central manager (which is
composed of the negotiator and collector demons). The schedd (short for
scheduler demon) maintains a queue of jobs and asks the central manager to
find resources for its jobs. A condor pool can have many schedds (each
with its own queue of jobs), but only one central manager.

With flocking, your local schedd talks to the remote central manager to
obtain additional resources for its jobs. Although there may be a schedd
running on the same remote machine, it doesn't enter the picture.

Try running condor_status -p remotecentralmanager -schedd. The name of
your local schedd should in the list if it has idle jobs.

On Tue, 8 Mar 2005, Srirangam Addepalli wrote:

> Hmm when i try to submit jobs to local sched the jobs are waiting in
> queue with status I if no local resource is available. Does this
> essentially mean that local sched is not able to communicate with
> remote sched.
>
> how ever when i do a condor_status -p remotesched i can c all the
> hosts in the remote pool.
>
> Sincerly
>
> Rangam
>
> On Tue, 8 Mar 2005 10:51:21 -0600 (CST), Jaime Frey <jfrey@xxxxxxxxxxx> wrote:
> > On Tue, 8 Mar 2005, Srirangam Addepalli wrote:
> >
> > > I was trying to flock jobs to a different sched on a different machine
> > > on a entirely
> > > differnet network.
> > >
> > > When i do a condor_status -p systemname.edu i am able to c all the
> > > systems in the condor pool on the other side.
> > >
> > > Following were the error in both cases.
> > >
> > > C:\>condor_submit -r FQDN( systemname) name.sub
> > >
> > > ERROR: Can't find address of schedd systemname
> > >
> > > C:\>condor_submit -r System name.sub
> > > condor_submit: unknown host System
> > >
> > > Does something else need to be modified besides flock_to and flock from
> > > in condor_config.
> >
> > With flocking, you still submit your jobs to your local schedd. Then, if
> > your schedd can't find enough resources to run all your jobs in your local
> > pool, it'll start asking for resources in the remote pool as well.
> >
> > +----------------------------------+---------------------------------+
> > |            Jaime Frey            |  Public Split on Whether        |
> > |        jfrey@xxxxxxxxxxx         |  Bush Is a Divider              |
> > |  http://www.cs.wisc.edu/~jfrey/  |         -- CNN Scrolling Banner |
> > +----------------------------------+---------------------------------+
> >
>

+----------------------------------+---------------------------------+
|            Jaime Frey            |  Public Split on Whether        |
|        jfrey@xxxxxxxxxxx         |  Bush Is a Divider              |
|  http://www.cs.wisc.edu/~jfrey/  |         -- CNN Scrolling Banner |
+----------------------------------+---------------------------------+