[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] Condor negotiator ignoring accounting groups?



Do you have the output of condor_userprio -all
for that same time?

Steve

On Thu, 12 Jan 2012, Sarah Williams wrote:

Hi,

I'm using condor 7.6.0-1 on a cluster with 4252 job slots. I have the
following accounting groups configured:
GROUP_NAMES = group_opport, group_atlasprod, group_dzero,
group_atlasinstall, group_atlasanaly
GROUP_QUOTA_group_dzero = 20
GROUP_QUOTA_group_opport = 200
GROUP_QUOTA_group_atlasprod = 1650
GROUP_QUOTA_group_atlasinstall = 1
GROUP_QUOTA_group_atlasanaly = 2477
GROUP_PRIO_FACTOR_group_opport = 500
GROUP_PRIO_FACTOR_group_atlasprod = 10
GROUP_PRIO_FACTOR_group_atlasinstall = 99
GROUP_PRIO_FACTOR_group_atlasanaly = 9
GROUP_PRIO_FACTOR_group_dzero = 1000
GROUP_ACCEPT_SURPLUS = TRUE
GROUP_AUTOREGROUP_group_dzero = FALSE
GROUP_AUTOREGROUP_group_opport = FALSE

We're currently running ~3600 jobs in the atlasprod group, and ~470 jobs
in the atlasanaly group.  I would expect that the next jobs to run would
be atlasanaly. However, I find that the negotiator is instead starting
atlasprod jobs.  I attached an excerpt from the Negotiator log for one
such cycle.  The four jobs that were started that cycle were all atlasprod.

Any ideas why that's happening? It seems like from the log that is
completely ignoring the accounting group, even though I can see that the
accounting group is set in the job classad.  I am seeing a difference
that the group_opport attribute is quoted, while the atlas ones are not:
AccountingGroup = group_atlasprod.usatlas1
AccountingGroup = "group_opport.engage"


--Sarah


------------------------------------------------------------------
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, Group Leader.
Lead of FermiCloud project.