LHC Computer Grid - Rollout
> [mailto:[log in to unmask]] On Behalf Of Steve Traylen
said:
> Is this needed? GRAM already supports passing these requirements and
> the jobmanagers already respect these. Also condor-g jobs can
> be configured with the parameters as demonstrated by Rod's grid. It is
> just the WMS to Condor-G bit that is missing.
It's not so much the WMS as the JDL which is the problem. At the moment
you could in theory have a requirement with a very complex expression,
say (si2k*(wallclocklimit - cputimelimit)/totalcpus)>(freecpus -
runningjobs)/memory - that probably makes no sense because I made it up
at random, but you could still do it ... and it's not that easy to turn
into simple restrictions in the batch system, to say the least.
I think there are broadly four possibilities (at least until someone
thinks of some more :):
1) Keep the current syntax, allow matching against multiple subclusters,
and pass the subcluster name to the batch system.
2) Build a classad parser which does the best it can with the expression
it gets; that parser might be either in the WMS or in the CE.
2a) As 2, but insist that users restrict their requirement expressions
to things the parser can deal with.
3) Add new things in the JDL which match more directly with the kind of
thing you get in batch systems, e.g.:
CPULimit = 3600;
MEMlimit = 511;
DiskLimit = 500;
Priority = 3;
3) is simpler on the batch system side, but you still have to adapt it
to specific systems, you still have to make it match the glue schema (or
vice versa), and you lose a lot of the potential richness of the classad
expressions.
Stephen
|