Changes between Version 5 and Version 6 of hexagon

Timestamp:

03/23/16 03:07:23 (9 years ago)

Author:

bdef

Comment:

--

hexagon

@@ -108 +108 @@
 TO BE UPDATED
 
-On Hexagon, you can use up to 30 cores per node. The more nodes and the longer the requested time, the more you will have to wait in the queue. So choose your settings wisely:
+Hexagon is aimed to large jobs. As on Vilje, the accounting is done as if you are using 32 processes per nodes even if you are using less. So if you don't need large ammount of memory try to restrict the number of nodes you are using. The scheduler also prioritize jobs with a high CPU counts. More on the queue priorities here: 
+https://docs.hpc.uib.no/wiki/Queue_priorities_%28Hexagon%29
 
- {{{
-#!m
-md.cluster=hexagon('numnodes',1,'cpuspernode',8);
-md.cluster.time=10;
-}}}
+There is a batch queue that redirect the jobs in specific queues if they respond to certain requirments:
 
-to have a maximum job time of 10 minutes and 8 cores on one node. If the run lasts longer than 10 minutes, it will be killed and you will not be able to retrieve your results.
+|| normal || queue for normal jobs                                  || limited only by job limitations      ||
+|| small  || queue for small jobs, jobs will get higher priority    || max 512 CPUs, max 1 hour walltime    ||
+|| debug  || queue for debugging, jobs will get higher priority     || max 64 CPUs, max 20 minutes walltime ||
 
-Now if you want to check the status of your job and the queue you are using, type in the bash with the Greeplanet session:
+There are some limitations to the ressource that one single user can ask for :
+ * 4096 cpu cores, total number used by all running jobs
+ * 8-22 running (active) jobs (depending on load)
+ * 2 idle jobs
 
- {{{
-#!sh
-squeue -u username
-}}}
-
-You can delete your job manually by typing:
-
+If you want to check the ordering of the queue and your jobs position you can use in your active Hexagon session:
 {{{
 #!sh
-scancel JOBID
+showq          # display jobs
+showq -u $USER # display jobs for $USER
+showq -i      # display only jobs in eligible (idle) queue waiting for execution
 }}}
-
-where JOBID is the ID of your job (indicated in the Matlab session). Matlab indicates too the directory of your job where you can find the files `JOBNAME.outlog` and `JOBNAME.errlog`. The outlog file contains the informations that would appear if you were running your job on your local machine and the errlog file contains the error information in case the job encounters an error. 
-
-If you want to load results from the cluster manually (for example if you have an error due to an internet interruption), you find in the informations Matlab gave you `/home/srebuffi/trunk-jpl/execution//SOMETHING/JOBNAME.lock `, you copy the SOMETHING and you type in Matlab:
-
-{{{
-#!m
-md=loadresultsfromcluster(md,'SOMETHING');
-}}}
-
-A comparision of PBS to slurm commands can be found here: http://slurm.schedmd.com/rosetta.pdf