Re: [Condor-users] Matchmaking in grid universe

[Steven Timm <timm@xxxxxxxx>]

On Fri, 27 Jun 2008, Enol Fernández wrote:

> Hi,
>
> It works! I have added those attributes to the ad and now it is matching 10
> jobs in each cycle (which is the limit I have in the requirements
> expression):
>
> Requirements = (CurMatches < 10) && (TARGET.JobUniverse == 9)
>
> However, the CurMatches attribute in the ad is not updated and whenever I do
> a condor_status its value is always 0. This is not a problem for me, but I
> would expect it to be updated.
>
> Thanks,
> Enol.
CurMatches will always appear to be zero when you do condor_status.
It is only incremented within the negotiation cycle in the
negotiator's memory but then resets to zero for the
next negotiation cycle.  We have the same Requirements in
our matchmaking setup and see the same effect, and Condor staff
have told us this is normal and expected.

Steve Timm



> On Thu, Jun 26, 2008 at 21:59, Dan Bradley <dan@xxxxxxxxxxxx> wrote:
>
> > Jaime and I took a look at the negotiator code.  We cannot explain how
> > this used to work!
> >
> > Anyway, based on our current understanding, the following should allow
> > you to match more than 1 per cycle.  Add/change the following attributes
> > in your site ClassAds that you advertise with condor_advertise:
> >
> > RemoteUser = "fake_user"
> > Rank = 1.0
> > CurrentRank = 0.0
> >
> > Could you please try that and let us know if it solves your problem?
> >
> > Sorry for the confusion on our end about this.  We would like to get
> > this fixed, because it appears that something is not right.
> >
> > --Dan
> >
> > Enol Fernández wrote:
> > > On Wed, Jun 11, 2008 at 18:37, Jaime Frey <jfrey@xxxxxxxxxxx
> > > <mailto:jfrey@xxxxxxxxxxx>> wrote:
> > >
> > >     On Jun 10, 2008, at 2:36 AM, Enol Fernández wrote:
> > >
> > > >     I am trying to do matchmaking with Condor-C advertising to the
> > > >     collector a ClassAd similar to the one found in the Condor manual
> > > >     in
> > http://www.cs.wisc.edu/condor/manual/v7.0/5_3Grid_Universe.html#SECTION00637200000000000000
> > > >     MyType         = "Machine"
> > > >     TargetType     = "Job"
> > > >     Name           = "some_name"
> > > >     Machine        = "some_name"
> > > >     grid_resource_name  = "condor schedd@mymachine mypool"
> > > >     Requirements = (CurMatches < 10) && (TARGET.JobUniverse == 9)
> > > >     Rank = 0.000000
> > > >     CurrentRank = 0.000000
> > > >     WantAdRevaluate = True
> > > >     Activity = "Idle"
> > > >     State = "Unclaimed"
> > > >     CurMatches = 0
> > > >     OpSys = "CondorC"
> > > >     Arch = "CondorC"
> > > >
> > > >     and I am submitting 20 jobs using this description:
> > > >
> > > >     universe = grid
> > > >     Executable = mytest.sh
> > > >     Output     = hello.out
> > > >     Error      = hello.err
> > > >     Log        = hello.log
> > > >     grid_resource = $$(grid_resource_name)
> > > >     Requirements = TARGET.grid_resource_name =!= UNDEFINED
> > > >     +remote_jobuniverse = 5
> > > >     +remote_requirements = (Arch == "INTEL") && (OpSys == "LINUX")
> > > >     +remote_ShouldTransferFiles = "YES"
> > > >     +remote_WhenToTransferOutput = "ON_EXIT"
> > > >     Queue 20
> > > >
> > > >     What I have found is that the CurMatches is never increased after
> > > >     a successful match and that only one job is matched per
> > > >     negotiation cycle (I would expect 10 jobs matched).
> > > >
> > > >     Is CurMatches supposed to be increased automatically or should I
> > > >     update the ad in the Collector? Is there a way to enforce more
> > > >     than one match per cycle?
> > >
> > >     Try adding the following to the configuration file of the machine
> > >     running your negotiator daemon (don't forget to run
> > >     condor_reconfig afterwards):
> > >
> > >     NEGOTIATOR_MATCHLIST_CACHING = false
> > >
> > >     If that fixes the problem, it points out a bug that needs fixing
> > >     in the negotiator.
> > +--------------------------------+-----------------------------------+
> > >     |           Jaime Frey           | I used to be a heavy gambler.
> > |
> > >     |       jfrey@xxxxxxxxxxx <mailto:jfrey@xxxxxxxxxxx>        | But
> > >     now I just make mental bets.  |
> > >     | http://www.cs.wisc.edu/~jfrey/ <http://www.cs.wisc.edu/%7Ejfrey/>
> > >     <http://www.cs.wisc.edu/%7Ejfrey/> | That's how I lost my mind.
> > >         |
> > +--------------------------------+-----------------------------------+
> > > Hello,
> > >
> > > adding this option did not help, still only one match. I attach the
> > > negotiation cycle part of the Negotiator log, maybe this can help to
> > > track down the problem.
> > >
> > > Regards,
> > > Enol.
> > > ------------------------------------------------------------------------
> > >
> > > _______________________________________________
> > > Condor-users mailing list
> > > To unsubscribe, send a message to condor-users-request@xxxxxxxxxxx with
> > a
> > > subject: Unsubscribe
> > > You can also unsubscribe by visiting
> > > https://lists.cs.wisc.edu/mailman/listinfo/condor-users
> > >
> > > The archives can be found at:
> > > https://lists.cs.wisc.edu/archive/condor-users/
> > _______________________________________________
> > Condor-users mailing list
> > To unsubscribe, send a message to condor-users-request@xxxxxxxxxxx with a
> > subject: Unsubscribe
> > You can also unsubscribe by visiting
> > https://lists.cs.wisc.edu/mailman/listinfo/condor-users
> >
> > The archives can be found at:
> > https://lists.cs.wisc.edu/archive/condor-users/

--
------------------------------------------------------------------
Steven C. Timm, Ph.D  (630) 840-8525
timm@xxxxxxxx  http://home.fnal.gov/~timm/
Fermilab Computing Division, Scientific Computing Facilities,
Grid Facilities Department, FermiGrid Services Group, Assistant Group Leader.