[SCore-users-jp] [SCore-users] Questions about Score scheduling scheme

Jure Jerman jure.jerman @ rzs-hm.si
2003年 4月 30日 (水) 01:12:40 JST


Dear Score-users,

we are running score in a quite specific environment of operational
meteorological service. In practice this means that we have some "research jobs" and twice
a day a serie of "operational jobs". We do not use any additional queueing system (like PBS, ...).

We are trying to solve any queueing problems by using score priorities, any "operational"
job goes into the queue 0. However we have noticed several problems:

1. It would be natural if another job with priority 0 would go to a node where there
is no job with priority 0 running. This is not the case with Score-5.4.0. For example,
if the execution of job0 running with priority 0 depends on the output of job1 starting
later with priority 0 and the job1 is scheduled to the node where job0 is already running we
have a deadlock problem.

2.  If we further play with renicing (sc_console nice command) the command becomes
effective only when the process is suspended/resumed. Is this a feature or a bug?

3. For some time we were considering checkpointing/aborting/restarting but the operation fails
with the messages:

SCORE: Checkpointing ... done.
FEP:WARNING Job is aborted by operator.
FEP: [29/Apr/2003 16:04:42] Waiting for job restarted ...
FEP: [29/Apr/2003 16:04:51] Restarted.
SCore-D 5.4.0 connected (jid=66).
<0> SCORE WARNING: Physical memory might be exhausted.
<0:0> SCORE:WARNING (Program will be started from the beginning)
<0:0> SCORE: One node ready.
<0:0> SCORE:WARNING Restarting failed: sc_rstrt_open() failed: 16 (Device or resource busy)
<0:0> SCORE:WARNING (Program will be started from the beginning)
<0:0> SCORE: One node ready.

Any clues on that?

We would appreciate any hint about the problems.

With best regards, Jure Jerman
-- 
--------------------------------------------------------------
Jure Jerman                       Email: jure.jerman @ rzs-hm.si
Environmental Agency of Slovenia
Meteorological office
Vojkova 1b
SI-1001 Ljubljana
SLOVENIA
--------------------------------------------------------------

_______________________________________________
SCore-users mailing list
SCore-users @ pccluster.org
http://www.pccluster.org/mailman/listinfo/score-users



SCore-users-jp メーリングリストの案内