Re: [Condor-users] Running a HOOK_JOB_EXIT

[Colin Leavett-Brown <crlb@xxxxxxx>]

Thank you Matt for all your help with this problem. I think, in the end, 
the problem turns out to be a bug in 7.5.1. Here's the output from 761:

[crlb@elephant condor]$ cat x.out

1. Condor version - Command "condor_version" produces:
$CondorVersion: 7.6.1 May 31 2011 BuildID: 339001 $
$CondorPlatform: x86_rhap_5 $

2. Condor configuration - Command "condor_config_val -dump | grep -i 
hook" produces:
MUNIN_HOOK_JOB_EXIT = /usr/local/bin/munin-node-condor-job-exit

3. Job Class Ad - Command "grep -i hook $_CONDOR_JOB_AD" produces:
HookKeyword = "MUNIN"

4. Job hook permissions - Command "ls -l 
/usr/local/bin/munin-node-condor-job-exit" produces:
-rwxr-xr-x 1 root root 126 Jun 21 17:18 
/usr/local/bin/munin-node-condor-job-exit

5. Job hook test run - Command 
"/usr/local/bin/munin-node-condor-job-exit" produces:
Executing munin-node-condor-job-exit

6. Job exit should produce a second line of output from the hook:
Executing munin-node-condor-job-exit
[crlb@elephant condor]$

I found that the following also works, and doesn't require the user to 
add anything to their job file:

2. Condor configuration - Command "condor_config_val -dump | grep -i 
hook" produces:
STARTER_HOOK_JOB_EXIT = /usr/local/bin/munin-node-condor-job-exit
STARTER_JOB_HOOK_KEYWORD = STARTER

Thanks again for your help, Colin.

Matthew Farrellee wrote:
> Colin,
>
> It could very well be a bug fixed between 7.5.1 and 7.6.1 (over a year 
> of development).
>
> If you turn on STARTER_DEBUG = D_FULLDEBUG you should see something 
> along the lines of...
>
> ... HOOK_JOB_EXIT (/usr/local/bin/munin-node-condor-job-exit) invoked 
> with reason: "exit"
> ...
> ... HookClient /usr/local/bin/munin-node-condor-job-exit (pid 8077) 
> exited with status 0
>
> ...in the StarterLog.slotX where the job ran.
>
> Best,
>
>
> matt
>
> On 06/18/2011 11:36 AM, Colin Leavett-Brown wrote:
>   
> > Hi Matt, I think I'm doing everything you suggest, but it still won't run:
> >
> > [crlb@elephant condor]$ cat x.out
> > 1. Condor version - Command "condor_version" produces:
> > $CondorVersion: 7.5.1 Mar 1 2010 BuildID: 220663 $
> > $CondorPlatform: I386-LINUX_RHEL5 $
> >
> > 2. Condor configuration - Command "condor_config_val -dump | grep -i
> > hook" produces:
> > MUNIN_HOOK_JOB_EXIT = /usr/local/bin/munin-node-condor-job-exit
> >
> > 3. Job Class Ad - Command "grep -i hook $_CONDOR_JOB_AD" produces:
> > HookKeyword = "MUNIN"
> >
> > 4. Job hook permissions - Command "ls -l
> > /usr/local/bin/munin-node-condor-job-exit" produces:
> > -rwxr-xr-x 1 root root 102 Jun 16 13:18
> > /usr/local/bin/munin-node-condor-job-exit
> >
> > 5. Job hook test run - Command
> > "/usr/local/bin/munin-node-condor-job-exit" produces:
> > Executing munin-node-condor-job-exit
> >
> > 6. Job exit should produce a second line of output from the hook:
> > [crlb@elephant condor]$
> >
> > Sorry, Colin.
> >
> >
> > On 11-06-17 11:02 AM, Matthew Farrellee wrote:
> >     
> > > From this morning...
> > >
> > > --
> > > $ condor_version
> > > $CondorVersion: 7.6.1 May 31 2011 BuildID: RH-7.6.1-0.9.el6 $
> > > $CondorPlatform: X86_64-Fedora_13 $
> > >
> > > $ condor_config_val -dump | grep HOOK
> > > MUNIN_HOOK_JOB_EXIT = /usr/local/bin/munin-node-condor-job-exit
> > >
> > > $ echo 'cmd =
> > > munin.sh\ntransfer_executable=true\noutput=x.out\nshould_transfer_files=ALWAYS\nwhen_to_transfer_output=ON_EXIT\n+HookKeyword="MUNIN"\nqueue'
> > >
> > > | condor_submit
> > >
> > > $ cat x.out
> > > 1. The job script:
> > > #!/bin/bash
> > >
> > > #echo
> > > #echo 0. pwd
> > > #pwd
> > > #ls -al
> > >
> > > echo 1. The job script:
> > > cat condor_exec.exe
> > >
> > > echo
> > > echo 2. Hook config:
> > > grep -i HOOK /etc/condor/condor_config
> > >
> > > echo
> > > echo 3. Permissions on the hook:
> > > ls -l /usr/local/bin/munin-node-condor-job-exit
> > >
> > > echo
> > > echo 4. The hook:
> > > cat /usr/local/bin/munin-node-condor-job-exit
> > >
> > > echo
> > > echo 5. Test run of the hook:
> > > /usr/local/bin/munin-node-condor-job-exit
> > >
> > > echo
> > > echo 6. Job exit should produce a second line of output from the hook:
> > >
> > > 2. Hook config:
> > >
> > > 3. Permissions on the hook:
> > > -rwxr-xr-x. 1 root root 98 Jun 17 11:04
> > > /usr/local/bin/munin-node-condor-job-exit
> > >
> > > 4. The hook:
> > > #!/usr/bin/perl
> > > open(DD, ">>x.out");
> > > print DD "Executing munin-node-condor-job-exit\n";
> > > close(DD);
> > > 5. Test run of the hook:
> > > Executing munin-node-condor-job-exit
> > >
> > > 6. Job exit should produce a second line of output from the hook:
> > > Executing munin-node-condor-job-exit
> > > --
> > >
> > > Submit again without the HookKeyword and you won't see the message in 6.
> > >
> > > Best,
> > >
> > >
> > > matt
> > >
> > >       
>
>   

--
Colin Leavett-Brown
Department of Physics & Astronomy
University of Victoria
250-721-7728