Re: [Condor-users] Attribute $$(OpSys) not found in the machine ClassAd

[Todd Tannenbaum <tannenba@xxxxxxxxxxx>]

Poornima Pochana wrote:
>    Hi,
>
> I am seeing this same problem as discussed 
> <https://www-auth.cs.wisc.edu/lists/condor-users/2011-June/msg00069.shtml> 
> in the condor-users forum sometime ago, but my case seems to be a bit 
> different (curious), even though the error message is same.

Hi Poornima -

Not 100% certain, but a quick read of your email makes me think you 
stepped on the bug outlined here:
  https://condor-wiki.cs.wisc.edu/index.cgi/tktview?tn=2552

This bug was introduced in Condor v7.7.1 development release and is 
fixed for upcoming v7.7.3, i.e. this bug never made it into any stable 
series.

I think a workaround for Condor v7.7.1 / v7.7.2 could be to add the 
following into the condor_config of your submit machine(s):
  SHADOW_WORKLIFE = 0
The downside of this setting is it can negatively impact throughput of 
that submit machine if it is very busy, i.e. simultaneously running 
thousands of very short jobs.

regards,
Todd


> I am trying to execute a simple DAG workflow (bash script below) which 
> submits 2 condor jobs. The condor setup I have is a Personal Condor 
> (7.7) installed from development condor yum repo.The 'executable' 
> command in the submit file has the macro "simple.$$(OpSys).bat" defined 
> and I have the binary 'simple.LINUX.bat' in the same working tree.
>
> The first job - 1.condor finishes off successfully, but the second job - 
> 2.condor is held with the error msg "Cannot expand $$(OpSys)." Curious 
> thing is the OpSys attribute seems to be set to "LINUX" (condor_status 
> cmd below) and that the first job executes but not the second job.
>
> The log messages and my bash script are below. Any pointers as to what 
> my setup is missing would be a great help.
>
> Thanks,
> Poornima.
>
> Here's what I poked into:
> 1. The Condor system mail msg for the second job: "Attribute $$(OpSys) 
> cannot be expanded because this attribute was not found in the machine 
> ClassAd."
>
> 2. $ condor_status -long |grep -i machine
> Machine = "xxx.xxx.xxx.xxx"
> Unhibernate = MY.MachineLastMatchTime =!= undefined
> MyType = "Machine"
> $ condor_status -long |grep -i opsys
> OpSysAndVer = "LINUX"
> OpSysVer = 206
> OpSys = "LINUX"
>
> 3. Message from SchedLog, where it puts on hold job 81 (the held job 
> 2.condor), but schedules job 80 (1.condor)
> 11/10/11 14:38:39 (pid:17416) Starting add_shadow_birthdate(80.0)
> 11/10/11 14:38:39 (pid:17416) Started shadow for job 80.0 on 
> centos6.lab.ac.uab.edu <10.0.0.26:40521> for ppreddy, (shadow pid = 16301)
> 11/10/11 14:38:39 (pid:17416) Finished negotiating for ppreddy in local 
> pool: 1 matched, 1 rejected
> 11/10/11 14:38:43 (pid:17416) Shadow pid 16301 for job 80.0 reports job 
> exit reason 100.
> 11/10/11 14:38:43 (pid:17416) match (centos6.lab.ac.uab.edu 
> <10.0.0.26:40521> for ppreddy) switching to job 81.0
> 11/10/11 14:38:43 (pid:17416) Shadow pid 16301 switching to job 81.0.
> 11/10/11 14:38:43 (pid:17416) Starting add_shadow_birthdate(81.0)
> 11/10/11 14:38:43 (pid:17416) Putting job 81.0 on hold - cannot expand 
> $$(OpSys)
> 11/10/11 14:38:43 (pid:17416) Job 81.0 put on hold: Cannot expand $$(OpSys).
> 11/10/11 14:38:43 (pid:17416) Failed to expand job ad when switching 
> shadow 16301 to new job 81.0
>
> 4. NegotatorLog and MatchLog show rejection for not having found a match:
> 11/10/11 14:38:39       Rejected 81.0 <x.x.x.x:51994>: no match found
>
> #!/bin/bash
>
> runs=2
> arg1=4
> arg2=10
> for job in `seq $runs`
> do
>     cat > $job.condor << EOF
> Universe   = vanilla
> Executable = simple.\$\$(OpSys).bat
> Arguments  = $arg1 $arg2
> Log        = $job.log
> Output     = $job.out
> Error      = $job.error
> Queue
> EOF
>
>     let arg1=$arg1+1
>     let arg2=$arg2+10
>     JOB_LIST="$JOB_LIST $job"
> done
>
> # generate condor dagman to manage jobs
> for JOB_ID in $JOB_LIST; do
>     echo "JOB    job_$JOB_ID $JOB_ID.condor" >>master.dag
>     echo "SCRIPT PRE   job_$JOB_ID pre-job " >>master.dag
>     echo "SCRIPT POST  job_$JOB_ID post-job " >>master.dag
>     echo "RETRY  job_$JOB_ID 5" >>master.dag
> done
> condor_submit_dag -notification Never master.dag >condor_submit_dag.out
>
>
>
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> 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/


--
Todd Tannenbaum                       University of Wisconsin-Madison
Center for High Throughput Computing  Department of Computer Sciences
tannenba@xxxxxxxxxxx                  1210 W. Dayton St. Rm #4257
Phone: (608) 263-7132                 Madison, WI 53706-1685