Re: [Condor-users] Not all processors are being utilized ...

[Dave STREET <davey_street@xxxxxxxxxxx>]

I am still new to this myself

try running 

condor_q -analyse 18 (or what ever number the queue you want to look at) 

this will give you an idea why jobs are being rejected. The chances are that the job is unable to run on the other machines because of permissions. you may see "x reject your job because of their own requirements"

What is the excetuable you are trying to run? I had this error running a _vbscript_ as this interacts with the host OS. Jobs you want to run should be completely stand alone (at least during testing and getting it up and running). 

try writing a very simple script file. 

ie. 

#!/bin/bash

#

echo hello world

then make sure you can run it from shell with ./scriptname (may need to make it excatuable) 

make a sub file like bellow 

universe = vanilla

requirements = ( Arch=="Intel")

executable = helloworld

output = hello$(PROCESS).out

error = hello.err

transfer_files = ON_EXIT

log = hello.log

notification = Never

queue 10

and submit it, it will run quick so keep an eye on the queue. 

but you should see it run on all you pool and give 10 output files each containing the string  "hello world"

once you have that working you are ready to start adapting your programs to work on condor. 

Aaron

---

Date: Sat, 1 May 2010 11:57:46 +0530
From: rohit.bioinfo@xxxxxxxxx
To: condor-users@xxxxxxxxxxx
Subject: Re: [Condor-users] Not all processors are being utilized ...

oks ... but i didn't get it completely however i gave the value to queue to 10 and still its using the two processors of the same machine but not the other processors of the machines in the pool and and on viewing the queue its only showing 2 jobs running rest 8 are idle ... can you pleas send me a sample submit file to be used for vanilla universe where the filesystem is not shared ... i will get some idea wots going on in my case

On Fri, Apr 30, 2010 at 9:41 PM, Dave STREET <davey_street@xxxxxxxxxxx> wrote:

Well you ahve only sent one job?
 
you need to set the queue to = some thing like 10
 
this will queue up the job 10 times so it will be handed out to all the PCS and run through untill completition
 
Aaron
 

---

Date: Fri, 30 Apr 2010 17:52:43 +0530
From: rohit.bioinfo@xxxxxxxxx
To: condor-users@xxxxxxxxxxx
Subject: [Condor-users] Not all processors are being utilized ...

Hi everyone i made a small cluster of two computers ... one of which is centramanager and other is compute node with submit and execute permissions .... i am submitting a job from central manager the job is running but only one processor is Claimed rest are idle ... i am not able to understand wots going on ... how to utilize all the processors

My submit file

Universe = Vanilla
Executable = gro
Output = hello.out
Error=hello.err
Log = hello.log
Queue

condor_q

-- Submitter: centralmanager.cbbi.net : <192.168.7.25:59302> : centralmanager.cbbi.net
 ID      OWNER            SUBMITTED     RUN_TIME ST PRI SIZE CMD             
  20.0   rohit           4/30 16:39   0+00:00:00 I  0   0.0  gro             

1 jobs; 0 idle, 1 running, 0 held


and condor_status


Name               OpSys      Arch   State     Activity LoadAv Mem   ActvtyTime

slot1@centralmanag LINUX      INTEL  Claimed   Busy     0.000   244  0+00:00:03
slot2@centralmanag LINUX      INTEL  Unclaimed Idle     0.000   244  0+00:40:05
slot1@xxxxxxxxxxxx LINUX      INTEL  Unclaimed Idle     0.000   497  0+01:00:03
slot2@xxxxxxxxxxxx LINUX      INTEL  Unclaimed Idle     0.000   497  0+01:05:06

                     Total Owner Claimed Unclaimed Matched Preempting Backfill

         INTEL/LINUX     4     0       1         3       0          0        0

               Total     4     0       1         3       0          0        0


and if i am subbmittting job from a computed node its running on that node only ... i am trying to run molecular dynamics using Gromacs mdrun program


please help asap

thanks

rohit

---

Get a free e-mail account with Hotmail. Sign-up now.

_______________________________________________
Condor-users mailing list
To unsubscribe, send a message to condor-users-request@xxxxxxxxxxx with a
subject: Unsubscribe
You can also unsubscribe by visiting
https://lists.cs.wisc.edu/mailman/listinfo/condor-users

The archives can be found at:
https://lists.cs.wisc.edu/archive/condor-users/

---

Get a free e-mail account with Hotmail. Sign-up now.