Differences

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

--- start:working_on_mogon:io_odds_and_ends:slurm_localscratch [2020/10/19 13:48]
meesters [Sending signals to jobs withing SLURM]
+++ start:working_on_mogon:io_odds_and_ends:slurm_localscratch [2020/10/19 13:57]
meesters [Copy files to multiple nodes via job script]
@@ Line 12: / Line 12: @@
 </callout>
-**Attention:** This is //not// a shared filesystem!\\
 If your job runs on multiple nodes, you cannot use the local scratch space on one node from the other nodes.\\
 If you need your input data on every node, please refer to the section [[slurm_localscratch#Copy_files_via_job_script|"Copy files to multiple nodes via job script"]].
@@ Line 21: / Line 20: @@
 Assume you would normally start the program in the current working directory where it will read and write its data like this:
 <code bash>
-$ sbatch -N1 -p nodeshort ./my_program # mogonI
+$ sbatch -N1 -p parallel ./my_program
-#
-$ sbatch -N1 -p parallel ./my_program # mogonII
 </code>
 Now to get the performance of local disk access, you want to use the aforementioned local scratch space on the compute node.
@@ Line 33: / Line 29: @@
-===== Copy files via job script =====
+===== Copy files via job script and signalling batch scripts with SLURM =====
-This methods requires you to wrap your program in a small shell script like this:
+The following example will submit a jobscript, where SLURM will send a signal to the job script prior to ending. This will enable the jobscript to collect data written to the local scratch director(ies).
 <file bash job.sh>
@@ Line 103: / Line 99: @@
 </code>
-withing a submission script to signal the batch-job (instead of all the children of but not the batch job itselft). The reason is: If using a submission script like the one above, you trap the signal within the script, not the application. In case an application accepts a specific signal and you want to use this functionality, you can send the signal from within the script:
+withing a submission script to signal the batch-job (instead of all the children of but not the batch job itselft). The reason is: If using a submission script like the one above, you trap the signal within the script, not the application.
-<code bash>
-# list of process IDs (PIDs) to signal
-QUEUE=""
-function queue {
-  QUEUE="$QUEUE $1"
-}
-function forward_signal() {
-  # this function might fulfill additional purposes, like
-  # forwarding the signal, waiting a checkpoint to be written
-  # and then copying the last checkpoint back to the parallel file system
-  # just send the desired signal, e.g. SIGUSR2
-  kill -s SIGUSR2 $1
-}
-# trap the signal within the bash script
-# it is possible to connect several functions with a signal
-trap 'forward_signal' SIGUSR2
-# start the desired application(s) - note the &
-eval "my command and its parameters &"
-# store the PID of the desired application(s)
-queue $!
-# The sequence above needs to be carried out for every application instance
-# you want to be signalled.
-</code>
 </callout>
 ===== Copy files to multiple nodes via job script =====
 The following script can be used to ensure that input files are present in the job directory on **all** nodes.\\
-This is required for e.g. [[software:namd2|NAMD2]], which in some cases reads input data on other nodes than the starting node.
-This script is very verbose, you might want to delete or comment out the ''echo'' lines.
-Also, this script copies data from **all** nodes back into separate directories named ''${SAVEDPWD}/${SLURM_JOB_NODELIST}/${host}''.
+The demonstrated ''sbcast'' command can also be used for the one-node example above.
-\\If your application //only// needs to read on every node but does not write on every node, you want to use the cleanup function from the script posted above. The demonstrated ''sbcast'' command can also be used for the one-node example above.
 <file bash job_multinode.sh>
 #!/bin/bash
-#SBATCH -N 2 # assuming mogon I 'bulldozer' := 64 cores
+#SBATCH -N 2
-#SBATCH -J 'namd2_128'
+# use other parameterization as appropriate
-#SBATCH -p nodeshort
-#SBATCH -mem 1800M
 JOBDIR="/localscratch/${SLURM_JOB_ID}"
-HOSTLIST=$(scontrol show hostname $SLURM_JOB_NODELIST | paste -d, -s | tr ',' ' ')
-echo $HOSTLIST
-# Store working directory to be safe
-SAVEDPWD=$(pwd)
-# We define a bash function to do the cleaning when the signal is caught
-cleanup() {
-     sbcast ${JOBDIR}/resultfile ${SAVEDPWD}/${SLURM_JOB_ID}/resultfile
-     exit 0
-}
-# Register the cleanup function when SIGUSR2 is sent,
-# ten minutes before the job gets killed
-trap 'cleanup' SIGUSR2
 # copy the input file on all nodes
-sbcast ${HOME}/NAMD_2.9/apoa1.namd $JOBDIR/apoa1.namd
+sbcast <somefile> $JOBDIR/<somefile>
-# some applications only need the file on the 'master' node
-# in this case you can restrict yourself to:
-cp ${HOME}/NAMD_2.9/apoa1.namd $JOBDIR/apoa1.namd
-# Go to jobdir and start the program
+# NOTE: Unlike 'cp' which accepts a directory and would assume that
-cd "${JOBDIR}"
+#       the destination file carries the same name, 'sbcast'
+#       requires that a filename is given for the destination.
-$@ "${JOBDIR}/apoa1.namd"
-# Call the cleanup function when everything went fine
-cleanup
 </file>
-This script is used as follows:
-<code bash>
-$ chmod +x ./job_multinode.sh
-$ namd2 # after loading the appropriate module
-</code>