Hi Mattias ,
The new pbs dynamic plug-in for the generic information provider fixes
the assumption that the pbs server is on the same node as the CE.
If you think the assumption that the queue name is the same as the VO
name is incorrect, then please could you submit a bug.
The reason why this assumption has been made is due to a problem in the
Glue Schema that we are attempting to fix. This is the only way that we
can get accounting to work correctly. We need to know the number of jobs
running and waiting per VO.
Laurence
Mattias Wadenstein wrote:
> On Tue, 18 Jan 2005, Burke, S (Stephen) wrote:
>
> > LHC Computer Grid - Rollout
> >> [mailto:[log in to unmask]] On Behalf Of Mattias Wadenstein
> > said:
> >> With some hackery, it is possible to make lcgpbs.pm submit to only one
> >> queue and have it not verify the queue attribute of the RSL
> >> etc. And also
> >> report on the existing queue. The one problem I've seen with
> >> this is that
> >> every running job turns up 5 times in the IS, so some webpage
> >> overestimate
> >> the total number of running jobs.
> >
> > That seems like a very bad solution to me. The information system is
> > supposed to reflect the situation at the site, hacking it to report
> > something completely different just to get things working is not the
> way to
> > go.
>
> Well, consider it a temporary hack until the software has better support
> for different pbs site setups. For now, the assumption that each vo has
> its own queue seems to be all over the code, and the pbs interface was a
> much smaller "area" to cover than the interface to the information
> system.
>
> /Mattias Wadenstein
>
|