Monitoring Script

The following script checks all the WNs, in order to look for jobs that are not running anymore.

So far only those jobs that have been sent a SIGKILL are deleted; and an email is still sent. If the result of a tracejob is showed in the email, then there could be another reason to mark a job as eligible to be deleted:

 #!/bin/bash
 wn_list=`pbsnodes -a | grep " wn-" | cut -f9 -d' '`                             #List of WNs
 FILE="/tmp/monitoring_script/results/Monitoring_Results_`date +%k%M%d%m%y`"     #Results file, sent by email
 wn_result=""                                                                    #Diagnosis of a WN
 jobidlist=""                                                                    #List of jobs to delete
 i=0                                                                             #Counter of WN with jobs to delete
 jobresult=""                                                                    #Result of tracejob
 
 echo "===============================================================================" >> "$FILE"
 echo "=========================== AFFECTED WN DIAGNOSIS =============================" >> "$FILE"
 echo "===============================================================================" >> "$FILE"
 for wn in ${wn_list[@]}                                                         #Exploring the wn list
 do
         wn_result=`momctl -d2 -h $wn`
         jobidlist=`echo $wn_result | grep "sidlist= " `
         if [ ${#jobidlist} -ne 0 ]                                              #Is there any job to be deleted?
         then
                 echo $wn_result >> "$FILE"
                 echo "_______________________________________________________________________________" >> "$FILE"
                 job_id_list=`echo $jobidlist | sed -e 's/\[/\n/g' | grep "sidlist= " | cut -f1 -d'.'`
                 i=`expr $i + 1`
                 for job_id in ${job_id_list[@]}                                 #Exploring the jobs
                 do
                         echo $job_id
                         jobresult=`tracejob -n 8 -q $job_id`
                         if echo ${jobresult} | awk '/SIGKILL/' > /dev/null      #A SIGKILL signal was sent
                         then
                                 job_id="$job_id.stro.nikhef.nl"
                                 echo "The job $job_id can be deleted" >> "$FILE"
                                 momctl -h $wn -c $job_id                        #Deleting the job...
                                 echo "The job $job_id has been deleted from the queue and the WN $wn" >> "$FILE"
                         else
                                 echo $jobresult >> "$FILE"                      #This could give another reason to delete the job
                         fi
                         echo "_______________________________________________________________________________" >> "$FILE"
                 done
                 echo "_______________________________________________________________________________" >> "$FILE"
                 echo "_______________________________________________________________________________" >> "$FILE"
         fi
 done
 if [ $i -eq 0 ]
 then
         echo "There are no jobs to delete" >> "$FILE"
 fi
mail -s "Monitoring Results" <EMAIL_ADDRESS> < "$FILE"                           #<EMAI_ADDRESS> has to be edited