[gridengine users] shepherd error "unknown variable job_pid"

[Gerard Henry]

On 01/30/12 10:56 AM, Reuti wrote:
> Am 30.01.2012 um 10:11 schrieb Gerard Henry:
>
>> hello all,
>> i have 6.2u5 but sge_execd is now from SGE6.2u5p2. Everything seems ok, but at the end of a job, i got this message:
>> Job 14940 caused action: none
>> User        = webservd
>> Queue       = solqual_dev at charybde.cmi.univ-mrs.fr
>> Start Time  = 01/30/2012 09:22:44
>> End Time    = 01/30/2012 09:36:41
>> failed before epilog:01/30/2012 09:36:41 [1437:28775]: unknown variable "job_pid"
>> Shepherd trace:
>> ...
>> 01/30/2012 09:36:41 [1437:28775]: job exited with exit status 0
>> 01/30/2012 09:36:41 [1437:28775]: reaped "job" with pid 28781
>> 01/30/2012 09:36:41 [1437:28775]: job exited not due to signal
>> 01/30/2012 09:36:41 [1437:28775]: job exited with status 0
>> 01/30/2012 09:36:41 [1437:28775]: now sending signal KILL to pid -28781
>> 01/30/2012 09:36:41 [1437:28775]: writing usage file to "usage"
>> 01/30/2012 09:36:41 [1437:28775]: no tasker to notify
>> 01/30/2012 09:36:41 [1437:28775]: unknown variable "job_pid"
>>
>> Shepherd error:
>> 01/30/2012 09:36:41 [1437:28775]: unknown variable "job_pid"
>>
>> Shepherd pe_hostfile:
>> charybde.cmi.univ-mrs.fr 1 solqual_dev at charybde.cmi.univ-mrs.fr UNDEFINED
>>
>>
>> i don't have epilog script on this queue. Anybody has an idea about this message?
>
> Also no global one? Maybe it slipped in in some way. If I try to define it I get a similar message:
>
> $ qconf -mconf
> denied: parameter "epilog" in configuration: "unknown variable "job_pid""
>
> which makes sense, as of the time of a prolog/epilog the job isn't running and there is no pid at all.
>

sorry, my fault :(
i have several queues, and i was wrong, this queue had an epilog script. 
It's ok now


gerard