Hi Winnie,
Best I can offer is to grep for exec_host=<node_you_are_interested_in> in /var/spool/pbs/server_priv/accounting/20101028 then try to correlate the start and stop times of the jobs.
Yours,
Chris.
> -----Original Message-----
> From: Testbed Support for GridPP member institutes [mailto:TB-
> [log in to unmask]] On Behalf Of Winnie Lacesso
> Sent: 29 October 2010 09:41
> To: [log in to unmask]
> Subject: what jobs ran on this WN at such+such a time - ask
> torque/maui/mysql?
>
> Happy Friday!
>
> A job yesterday produced an interesting log entry:
>
> Oct 28 16:50:09 bse08 kernel: whizard[17144] trap divide error
> rip:63fd80b
> rsp:7fff05f30f40 error:0
>
> I can't see from a quick scan of torque/pbs admin manual how to ask pbs
> which jobs were running on this node at this time. The Maui admin
> manual
> mentions mprof & profiler but has 0 info, moab manual does not seem to
> have any mprof or profiler...
>
> Does anyone know the recipe of how, as quick+easy as poss, to ask
> torque
> &/or maui which jobs ran on a given WN in a given time-window?
> I'd be very grateful for an example!
> Is it in gLite doc somewhere? Or does one query rgma, or the MON mysql
> db
> directly? Again an example would be much appreciated!
>
> I do have pbswebmon for tht CE+WN but AFAIK pbswebmon only presents an
> instant snapshot of the cluster & keeps 0 historical info.
> Or does anyone know different, if so how to have pbswebmon show what
> jobs
> ran on a given node in a given time-window?
>
> Yours in query-mode!
> wl
|