[GE issues] [Issue 2822] New - tight integration does not work with two queues on one host

joga joga at sun.com
Mon Dec 8 09:53:08 GMT 2008


http://gridengine.sunsource.net/issues/show_bug.cgi?id=2822
                 Issue #|2822
                 Summary|tight integration does not work with two queues on one
                        | host
               Component|gridengine
                 Version|6.1u3
                Platform|All
                     URL|
              OS/Version|All
                  Status|NEW
       Status whiteboard|
                Keywords|
              Resolution|
              Issue type|DEFECT
                Priority|P3
            Subcomponent|execution
             Assigned to|pollinger
             Reported by|joga






------- Additional comments from joga at sunsource.net Mon Dec  8 01:53:07 -0800 2008 -------
It's not possible to start a tight integrated pe job in two queues on one host.
For 6.1 the job error message is:
error: executing task of job 1074943 failed:
in maintrunk the error message is a bit clearer:
error: executing task of job 1 failed: execution daemon on host "host1" didn't
accept task

The setup is quite simple. Add a new pe
pe_name           tight 
slots             100
user_lists        NONE
xuser_lists       NONE
start_proc_args   /bin/true
stop_proc_args    /bin/true
allocation_rule   $round_robin
control_slaves    TRUE
job_is_first_task TRUE
urgency_slots     min

Add two hosts with 2 slots per queue and one host into the hostlist

Submit a pe job:
qsub -pe tight 4 qrsh_inherit_test.sh

The job looks like this
#!/bin/sh
#
#$ -S /bin/sh
#$ -cwd
#set -x

start_tasks()
{
   while read host nproc rest; do
      hosttask=0
      echo "PE_HOSTFILE: $host $nproc $rest"
      while [ $hosttask -lt $nproc ]; do
         echo $host $JOB_ID $SGE_TASK_ID
         $SGE_ROOT/bin/$ARC/qrsh -inherit -cwd $host
$SGE_ROOT/examples/jobs/sleeper.sh 10 &
         hosttask=`expr $hosttask + 1`
      done
   done
   echo "master task submitted all sub tasks"
   wait
}

printf "master task started with job id %6d and pid %8d\n" $JOB_ID $$
cat $PE_HOSTFILE | start_tasks

wait

echo "master task exiting"

------------------------------------------------------
http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=36&dsMessageId=91771

To unsubscribe from this discussion, e-mail: [issues-unsubscribe at gridengine.sunsource.net].



More information about the gridengine-users mailing list