Re: [Condor-users] setting up condor on multiple machines...

[Alan De Smet <adesmet@xxxxxxxxxxx>]

bruce <bedouglas@xxxxxxxxxxxxx> wrote:
> after i do a 'condor_submit' from my 2nd machine, i do a followup 'condor_q'
> and it states that i have a job in the queue.
> 
> however, when i go to the manager server i don't see anything in the queue
> when i do a 'condor_q'.. how can i see all jobs for all submits on a given
> machine from the manager machine...??

Every condor_schedd in your pool maintains it's own queue of
jobs.  Thus, your central manager doesn't directly see the queue
on machine 2.

Some ways to get the information:

"condor_status -schedd" will report summaries of all of the
schedd queues.

"condor_q -global" will pull results from every schedd queue in
your pool (this can be slow in a big pool).

"condor_q -name $HOSTNAME_YOU_ARE_INTERESTED_IN" will pull
results from a single queue.

> also, it appears that there is a serious delay, between the time a job is
> submitted, and the time it's run. is there some config parameter that can be
> set to decrease the delay....

It's a bit complicated, in short Condor limits how frequently it
will attempt to match jobs.  This reduces the load on the system
for larger pools.  This defaults to 5 minutes.  

The quick option is "condor_reschedule", assuming you have
permission (and you will in the default configuration) this will
ask Condor to try and match your jobs immediately.

The permenant option is to set NEGOTIATOR_INTERVAL to a smaller
value.  For small pools setting it to 30 (it defaults to 300) is
safe.
NEGOTIATOR_INTERVAL:
http://www.cs.wisc.edu/condor/manual/v6.6/3_3Configuration.html#9197

-- 
Alan De Smet                              Condor Project Research
adesmet@xxxxxxxxxxx                 http://www.condorproject.org/