On Tue, 30 Nov 2004, Maarten Litmaath wrote:
> Burke, S (Stephen) wrote:
>
> > LHC Computer Grid - Rollout
> >
> >>[mailto:[log in to unmask]] On Behalf Of Maarten
> >>Litmaath, CERN said:
> >>
> >>>This could usefully be added to the Wiki entry,
> >>>
> >>
> >>http://goc.grid.sinica.edu.tw/gocwiki/Cannot_20read_20JobWrapper_20output_2e_2e_2e
> >
> >>which is just a little *too* terse...
> >>
> >>Being worked on...
> >
> >
> > It's a bit less terse now, but it's still missing other potential causes -
> > at least clock skew, full disks and NFS mount problems. Also diagnosis 1) is
>
> OK, we will add those.
And SSH too of course - which neatly cross-references to several questions
later in the Wiki.
The wall clock issue is one that the end-user can check out themselves,
probably by sending a "Hello World" job to make sure it's not something
else and then checking their JDL requires a realistic value for
other.GlueCEPolicyMaxWallClockTime, taking into account WN speed, possible
connection timeouts, etc.
Thanks
Henry
--
Dr. Henry Nebrensky [log in to unmask]
http://www.brunel.ac.uk/~eesrjjn
"The opossum is a very sophisticated animal.
It doesn't even get up until 5 or 6 p.m."
|