Re: [HTCondor-users] Stop Vanilla jobs from eviction/restart

[Prem Kumar <prem.it.kumar@xxxxxxxxx>]

hi Todd, thank you for your response.

i matched all of those settings in the link that you shared, and to my surprise they are exactly the same what it needs to be to disable preemption. 

although what i notice is many jobs have been restarted several times, in effect users having to wait for a long time to get their results. here are some details of what i notice of the jobs running and config values, do you notice anything unusual? All running jobs have equal ranks as well, although not equal userprio.....

$ condor_q -constraint '(JobRunCount > 2)' | wc -l

54

$ condor_q -constraint '(JobRunCount > 4)' | wc -l

32

$ condor_q -constraint '(JobRunCount > 5)' | wc -l

32

$ condor_q -constraint '(JobRunCount > 3)' | wc -l

33

$ condor_q -constraint '(RANK == 0)'  |wc -l

65

$ condor_q -constraint '(RANK == 0.0)'  |wc -l

65

$ condor_config_val -dump | grep -e ^CLAIM -e NEGOTIATOR_CONSIDER -e NEGOTIATOR_CONSIDER -e PREEMPT -e START

CLAIM_WORKLIFE = 540

HOSTALLOW_READ_STARTD = $(HOSTALLOW_READ), $(FLOCK_FROM)

HOSTALLOW_WRITE_STARTD = $(HOSTALLOW_WRITE), $(FLOCK_FROM)

MAX_STARTD_LOG = 10000000

MAX_STARTER_LOG = 10000000

NEGOTIATOR_CONSIDER_PREEMPTION = False

PREEMPT = False

PREEMPTION_RANK = 0

PREEMPTION_REQUIREMENTS = False

START = (True) && ( ($(START_SINGLE_CORE_JOB)) || ($(START_WHOLE_MACHINE_JOB)) )

START_SINGLE_CORE_JOB = TARGET.RequiresWholeMachine =!= True && MY.CAN_RUN_WHOLE_MACHINE == False && $(WHOLE_MACHINE_SLOT_STATE) =!= "Claimed"

START_WHOLE_MACHINE_JOB = TARGET.RequiresWholeMachine =?= True && MY.CAN_RUN_WHOLE_MACHINE

STARTD = $(SBIN)/condor_startd

STARTD_ADDRESS_FILE = $(LOG)/.startd_address

STARTD_ATTRS = COLLECTOR_HOST_STRING

STARTD_DEBUG =

STARTD_EXPRS =  CAN_RUN_WHOLE_MACHINE

STARTD_JOB_EXPRS = ImageSize, ExecutableSize, JobUniverse, NiceUser

STARTD_LOG = $(LOG)/StartLog

STARTD_SLOT_EXPRS =  State

STARTER = $(SBIN)/condor_starter

STARTER_DEBUG = D_NODATE

STARTER_LIST = STARTER, STARTER_STANDARD

STARTER_LOCAL = $(SBIN)/condor_starter

STARTER_LOG = $(LOG)/StarterLog

STARTER_STANDARD = $(SBIN)/condor_starter.std

STARTIDLETIME = 15 * $(MINUTE)

TESTINGMODE_PREEMPT = False

TESTINGMODE_PREEMPTION_RANK = 0

TESTINGMODE_PREEMPTION_REQUIREMENTS = False

TESTINGMODE_START = True

UWCS_PREEMPT = ( ((Activity == "Suspended") && ($(ActivityTimer) > $(MaxSuspendTime))) || (SUSPEND && (WANT_SUSPEND == False)) )

UWCS_PREEMPTION_RANK = (RemoteUserPrio * 1000000) - TARGET.ImageSize

UWCS_PREEMPTION_REQUIREMENTS = ( $(StateTimer) > (1 * $(HOUR)) && RemoteUserPrio > SubmittorPrio * 1.2 ) || (MY.NiceUser == True)

UWCS_START = ( (KeyboardIdle > $(StartIdleTime)) && ( $(CPUIdle) || (State != "Unclaimed" && State != "Owner")) )

On Thu, Jun 20, 2013 at 1:13 PM, Todd Tannenbaum <tannenba@xxxxxxxxxxx> wrote:

On 6/20/2013 12:46 PM, Prem Kumar wrote:

i meand this link is where i got the macros from:
https://lists.cs.wisc.edu/archive/htcondor-users/2007-November/msg00091.shtml

There is a section in the Manual that talks all about disabling preemption/eviction.  See

http://research.cs.wisc.edu/htcondor/manual/v8.0/3_5Policy_Configuration.html#SECTION00459500000000000000


Todd

On Thu, Jun 20, 2013 at 12:42 PM, Prem Kumar <prem.it.kumar@xxxxxxxxx

<mailto:prem.it.kumar@gmail.com>> wrote:

    Dear All,

    i have following macros defined for condor pool, but i still cannot
    get to stop the eviction for vanilla jobs. i referred to this link
    https://lists.cs.wisc.edu/archive/htcondor-users/pre-2004-June/msg00061.shtml

    SUSPEND = False
    PREEMPT = False
    CONTINUE = True
    WANT_SUSPEND = False
    WANT_VACATE = False

    these are dedicated rack based executions host managed centrally by
    us, so no individuals are owner of these resources.
    issue is we have users who run vanilla jobs that cannot
    be check-pointed or let us say we don't know how to checkpoint SAS &
    R jobs. any thoughts on how we can get around?

    best!

_______________________________________________
HTCondor-users mailing list
To unsubscribe, send a message to htcondor-users-request@cs.wisc.edu with a
subject: Unsubscribe
You can also unsubscribe by visiting
https://lists.cs.wisc.edu/mailman/listinfo/htcondor-users

The archives can be found at:
https://lists.cs.wisc.edu/archive/htcondor-users/

--
Todd Tannenbaum <tannenba@xxxxxxxxxxx> University of Wisconsin-Madison
Center for High Throughput Computing   Department of Computer Sciences
HTCondor Technical Lead                1210 W. Dayton St. Rm #4257
Phone: (608) 263-7132                  Madison, WI 53706-1685