Re: [Condor-users] rooster on linux, take 3

[Dan Bradley <dan@xxxxxxxxxxxx>]

Hi Ian,

The HIBERNATE expression is supposed to evaluate to a string specifying 
the desired power state.  The expression you specified below does not do 
that.  It evaluates to a boolean true/false value.

--Dan

On 11/29/11 6:55 AM, Ian Cottam wrote:
> Some more info:
>
> I am now testing with a HIBERNATE of just:
> HIBERNATE = (NO_ONE_LOGGED_IN =?= True)
>
> and after a couple of hours with no one logged in, it is still powered up.
>
> I copied the work here
> <https://twiki.grid.iu.edu/bin/view/Tier3/CondorHawkeyeSetup>.
>
> So I have:
>
> STARTD_CRON_JOBLIST = NOONELOGGEDIN
> STARTD_CRON_NOONELOGGEDIN_EXECUTABLE = /etc/condor/local/nooneloggedin.sh
> STARTD_CRON_NOONELOGGEDIN_PERIOD = 30s
> STARTD_CRON_NOONELOGGEDIN_MODE = Periodic
>
> Is that correct for Condor v7.4.4?
>
> nooneloggedin.sh has been tested independently (as described previously,
> below).
> -Ian
>
>
> On 29/11/2011 09:00, "Ian Cottam"<Ian.Cottam@xxxxxxxxxxxxxxxx>  wrote:
>
> > Does anyone have a good working of "hibernation/rooster wake up" across a
> > reasonable sized pool (i.e. bigger than just one test PC, although even
> > that would be interesting)?
> > Condor v7.4.4 or higher (below versions were known to have problems I
> > believe)?
> > Please share configs with us if you do.
> >
> > I recently added the recommended way of not hibernating if anyone was
> > logged in (via a "startd cron" as its sometimes called) and now the test
> > machines don't hibernate at all. I have tested that the script generates
> > NO_USER_LOGGED_IN = True
> > when only Condor has the PC, and test for that in the HIBERNATE
> > expression.
> >
> > Thanks.
> > -Ian
> >
> >
> >
> >
> >
> >
> > On 28/11/2011 18:44, "Dan Bradley"<dan@xxxxxxxxxxxx>  wrote:
> >
> > > On 11/28/11 12:33 PM, Dimitri Maziuk wrote:
> > > > On 11/28/2011 09:18 AM, Dan Bradley wrote:
> > > >
> > > > So the next question is how do I figure out what's up with the
> > > > negotiator?
> > > >
> > > > (E.g.) with 40 cores busy and 4 cores sleeping condor_q -analyze 961082
> > > > says:
> > > >
> > > > -- Submitter: minnow.bmrb.wisc.edu :
> > > > <144.92.167.254:9617?sock=13250_c2fa_3>   : minnow.bmrb.wisc.edu
> > > > ---
> > > > 961082.000:  Run analysis summary.  Of 44 machines,
> > > > ...
> > > >         4 match but are currently offline
> > > >         0 are available to run your job
> > > >           No successful match recorded.
> > > >           Last failed match: Fri Nov 25 18:18:55 2011
> > > >           Reason for last match failure: no match found
> > > > -----------------------------------------------------
> > > >
> > > > NegotiatorLog (on D_FULLDEBUG) is not very informative as to why the "4
> > > > matching but offline" cores are not a "successful match":
> > > >
> > > > 11/25/11 18:17:55     Sending SEND_JOB_INFO/eom
> > > > 11/25/11 18:17:55     Getting reply from schedd ...
> > > > 11/25/11 18:17:55     Got JOB_INFO command; getting classad/eom
> > > > 11/25/11 18:17:55     Request 961082.00000:
> > > > 11/25/11 18:17:55 matchmakingAlgorithm: limit 4.000000 used 0.000000
> > > > pieLeft 4.000000
> > > > 11/25/11 18:17:55       Rejected 961082.0 bbee@xxxxxxxxxxxxx
> > > > <144.92.167.254:9617?sock=13250_c2fa_3>: no match found
> > > > --------------------------------------------------------
> > >
> > > If you add D_JOB and D_MACHINE to NEGOTIATOR_DEBUG, you will get verbose
> > > logging of every machine considered by the negotiator when trying to
> > > match the job.  Is it even considering the offline machine?  If so, and
> > > if it matches, I would expect the following to be logged by the
> > > negotiator:
> > >
> > > "Registering attempt to match offline machine<host.name>  by
> > > <user.name>."
> > >
> > > --Dan
> > >
> > > _______________________________________________
> > > 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/
> >
> > --
> > Ian Cottam
> > ext. 61851
> > IT Services for Research
> > Faculty of Engineering and Physical Sciences
> > The University of Manchester
> > "The only strategy that is guaranteed to fail is not taking risks." Mark
> > Zuckerberg
> >
> >
> >
> >
> > _______________________________________________
> > 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/