Differences

This shows you the differences between two versions of the page.

--- public:lofar_processing_juelich [2015-08-10 14:31] – [SRM Copy] Stefan Froehlich
+++ public:lofar_processing_juelich [2017-03-08 15:27] (current) – external edit 127.0.0.1
@@ Line 32: / Line 32: @@
 ==== SRM Copy ====
-For data acquisition with grid-tools please use the Judac system. Your account for Jureca will also work for Judac.
+For data acquisition with grid-tools please use the Judac system. Your account for Jureca will also work for Judac.\\
 [[http://www.fz-juelich.de/ias/jsc/EN/Expertise/Datamanagement/JUDAC/JUDAC_node.html]]
@@ Line 41: / Line 41: @@
 Execute:\\
 <code>
-chmod 400 $HOME/.globus/userkey.pem \\
+chmod 400 $HOME/.globus/userkey.pem
 chmod 600 $HOME/.globus/usercert.pem
 </code>
@@ Line 73: / Line 73: @@
 "gsm" and pass "msss"\\
 \\
+To run your jobs on the compute nodes you first have to setup and submit
+a job via the batch system. A detailed description can be found on the
+Jureca homepage \\
+[[http://www.fz-juelich.de/ias/jsc/EN/Expertise/Supercomputers/JURECA/UserInfo/QuickIntroduction.html?nn=1803700]]
+When configuring the resources on the system please use only the '--nnodes' option. Ignore '--ntasks', '--ntasks-per-node' and '--cpus-per-task'. A (for the moment) working configuration is implemented in the framework itself.\\
+To run a generic pipeline on multiple nodes configure the pipeline as follows:
+In your 'pipeline.cfg' configure
+<code>
+[remote]
+method = slurm_srun
+max_per_node = 1
+</code>
+Set 'max_per_node' individually for every step in your parset to the number tasks you want to (can) run per node.