le_unknown?
Many thanks,
Richard
--
Richard Ems mail: richard@cape-horn-eng.com
Cape Horn Engineering S.L.
C/ Dr. J.J. Dómine 1, 5º piso
46011 Valencia
Tel : +34 96 3242923 / Fax 924
http://www.cape-horn-eng.com
___
users mailing list
users@grid
On 02/14/2011 06:10 PM, Reuti wrote:
> Hi,
>
> Am 14.02.2011 um 17:15 schrieb Richard Ems:
>
>> If I understand things correctly, if max_unheard time is reached for a
>> host, the queue instances on that host will be set to status unknown.
>
> Yep.
>
>
>
And another question:
Is there a variable - or anything else - that can be checked in the
running script to recognize that it has been started again as a result
of SGE rescheduling it?
Thanks again,
Richard
--
Richard Ems mail: richard@cape-horn-eng.com
Cape Horn Engineering S.L.
C
don't start parallel jobs on that many slots.
Could it be that max_reservation is set too low?
Regards, Richard
--
Richard Ems mail: richard@cape-horn-eng.com
Cape Horn Engineering S.L.
C/ Dr. J.J. Dómine 1, 5º piso
46011 Valencia
Tel : +34 96 3242923 / Fa
nd didn't get the OOM
anymore but only the header lines.
What's going on?
What am I doing wrong?
process-scheduler-log is running and schedule.last is readable by my user.
Thanks,
Richard
--
Richard Ems mail: richard@cape-horn-eng.com
Cape Horn Engineering S.L.
C/ Dr. J.J.
On 03/02/2011 03:58 PM, Mark Dixon wrote:
> On Wed, 2 Mar 2011, Richard Ems wrote:
> ...
>> I tried it once (with -a), and I got an OOM on our server with 12 GB
>> memory.
> ...
>
> Crikey.
>
> 1) How many lines does your schedule.last have?
# wc -l /opt/sge
tes, -q) 819200
real-time priority (-r) 0
stack size (kbytes, -s) unlimited
cpu time (seconds, -t) unlimited
max user processes (-u) 1000
virtual memory (kbytes, -v) unlimited
file locks (-x) unlimited
--
Richard Ems m
annot change and wants
some X output, and it works fine.
Best regards,
Richard
--
Richard Ems mail: richard@cape-horn-eng.com
Cape Horn Engineering S.L.
C/ Dr. J.J. Dómine 1, 5º piso
46011 Valencia
Tel : +34 96 3242923 / Fax 924
http://www.cape-horn-eng.com
ript I pointed $cmd to use my
lmutil_for_flex-grid script, and lmutil_for_flex-grid is just one line with
/opt/FlexLM/bin/v11.8.0.0/lmutil $* | sort | uniq
Richard Ems
--
Richard Ems mail: richard@cape-horn-eng.com
Cape Horn Engineering S.L.
C/ Dr. J.J. Dómine 1, 5º piso
4601
on error? Or is
it a bug? Or a feature?
We are using GE 6.2u5.
Many thanks,
Richard
--
Richard Ems mail: richard@cape-horn-eng.com
Cape Horn Engineering S.L.
C/ Dr. J.J. Dómine 1, 5º piso
46011 Valencia
Tel : +34 96 3242923 / Fax 924
http://www.ca
work?
Or is the mixing of different PEs the culprit of me not understanding it?
Thanks,
Richard
--
Richard Ems mail: richard@cape-horn-eng.com
Cape Horn Engineering S.L.
C/ Dr. J.J. Dómine 1, 5º piso
46011 Valencia
Tel : +34 96 3242923
equal
default_duration - and therefore N2 gets reserved and N1 starts another
Jn_4 4 slots job when one of J[12]_4 finishes ...
Can anyone understand what I am trying to understand and explain in
these two last sentences? 8)
Many thanks again,
Richard
--
Richard Ems mail: richard@cape-h
would just keep getting pushed back.
That was my thought also ...
Richard
--
Richard Ems mail: richard@cape-horn-eng.com
Cape Horn Engineering S.L.
C/ Dr. J.J. Dómine 1, 5º piso
46011 Valencia
Tel : +34 96 3242923 / Fax 924
http://www.cape-horn-eng.com
__
slots job when one of J[12]_4 finishes ...
Can anyone understand what I am trying to understand and explain in
these two last sentences? 8)
...
Thanks,
Richard
PS: I was called 5 minutes ago by a Univa guy ... coincidence? 8)
--
Richard Ems mail: richard@cape-horn-eng.com
Cape H
heir
> expected termination times should be in the same order.
I will try it now !
Thanks,
Richard
--
Richard Ems mail: richard@cape-horn-eng.com
Cape Horn Engineering S.L.
C/ Dr. J.J. Dómine 1, 5º piso
46011 Valencia
Tel : +34 96 3242923 / Fax 924
http://www.cape-horn-eng.com
___
shouldn't (in my opinion), since the other occupied 8
slots should get free in short time also and so that node should be
reserved for the waiting higher priority 12 slots job.
This does not happen.
Any ideas?
Thanks,
Richard
--
Richard Ems mail: richard@cape-horn-eng.com
Cape
,
Richard
--
Richard Ems mail: richard@cape-horn-eng.com
Cape Horn Engineering S.L.
C/ Dr. J.J. Dómine 1, 5º piso
46011 Valencia
Tel : +34 96 3242923 / Fax 924
http://www.cape-horn-eng.com
___
users mailing list
users@gridengine.org
https
you will probably want to dig deeper into exclusive node usage and
how to define the PE to not distribute the slots across nodes ... See
allocation_rule, man sge_pe .
cheers,
Richard
--
Richard Ems mail: richard@cape-horn-eng.com
Cape Horn Engineering S.L.
C/ Dr. J.J. Dómine 1, 5º
er:/data/... /scratch/dir/...
# run your program here
my_nice_program
# copy results back to server
{scp / rsync} /scratch/dir/... server:/data/...
exit 0
=========
--
Richard Ems mail: richard@cape-horn-eng.com
Cape Horn Engineering S.L.
not work, at least not in 6.2u5.
cheers,
Richard
--
Richard Ems mail: richard@cape-horn-eng.com
Cape Horn Engineering S.L.
C/ Dr. J.J. Dómine 1, 5º piso
46011 Valencia
Tel : +34 96 3242923 / Fax 924
http://www.cape-horn-eng.com
rque/Moab due to poor performance.
>
>
> Anyone else get this? I'm wondering if an over zealous sales rep is
> scraping our mailing list archives or something ...
>
> -dag
>
>
> ___
> users mailing list
> us
21 matches
Mail list logo