On Mon, 8 Aug 2005, Gerard Gadaud wrote:
> Hi
> I have a script doing the following:
> 1- submit my JDL with:
>
> # edg-job-submit --vo dteam -o
> /home/cgg/gadaud/EGG/EGEODE/20050804_101339/edg_job_submit.Jid
> /home/cgg/gadaud/EGG/EGEODE/20050804_101339/egeode.jdl
>
> 2- loop on edg-job-status until I encounter a time in the Done, Cleared,
> Aborted or Canceled phases:
>
> # edg-job-status -v 2 -i
> /home/cgg/gadaud/EGG/EGEODE/20050804_101339/edg_job_submit.Jid
> ....
> ...
> - stateEnterTimes =
> Submitted : Fri Aug 5 11:07:36 2005
> Waiting : Fri Aug 5 11:07:55 2005
> Ready : Fri Aug 5 11:07:58 2005
> Scheduled : Fri Aug 5 11:05:20 2005
> Running : Fri Aug 5 11:07:32 2005
> Done : Fri Aug 5 11:07:54 2005
> Cleared : ---
> Aborted : ---
> Cancelled : ---
>
> *************************************************************
>
> notice: job phases are out of chronological order
Can you check the date (time) on the CE and WNs, compare with the RB?
> or
>
> # edg-job-status -i
> /home/cgg/gadaud/EGG/EGEODE/20050804_101339/edg_job_submit.Jid
> *************************************************************
> BOOKKEEPING INFORMATION:
>
> Status info for the Job :
> https://rb1.egee.fr.cgg.com:9000/b8ZpFpMMbaZ8HQKtopYWBg
> Current Status: Ready
> Status Reason: unavailable
> Destination: ce1.egee.fr.cgg.com:2119/jobmanager-pbs-dteam
> reached on: Fri Aug 5 11:07:58 2005
> *************************************************************
>
>
> 3- if it went through the Running and Done phases and not through any of
> the Cleared, Aborted or Cancelled phases, then I retrieve OutPutSandBag:
You can retrieve the output sandbox only when the state is "Done",
after which the state becomes "Cleared".
The states "Aborted" and "Cancelled" are other terminal states, for which
there is no output sandbox available.
> # edg-job-get-output -i
> /home/cgg/gadaud/EGG/EGEODE/20050804_101339/edg_job_submit.Jid
>
> **** Error: NS_JOB_OUTPUT_NOT_READY ****
> The OutputSandbox files for job
> "https://rb1.egee.fr.cgg.com:9000/5HP8twoNi6OmHKFrmp4Eyw"
> are not yet ready for retrieval. Please wait that the job enters the
> "Done" status.
>
> I suppose that my job went though all the phases until the "Done" , then
No, it must have been aborted and then resubmitted; see below.
> for some reason went back the "Waiting" and "Ready" phases;
> Does anybody has an explanation for this?
> Is there an other way to get a job status an to find out in which phases
> a job went through successfully?
Use "edg-job-get-logging-info -v 1" to find out the exact history of a job.
|