For some reason the allocation code I use in the defs.local.ini isn’t being picked when the SubmitScript is created.
The value in the script is set to NO_ALLOCATION. When I set it correctly then I can submit the job, and it runs nicely.
If you can point me to the area in the code where allocation value is being read and set I could probably debug the issue.
Joe Hellmers
On Sep 3, 2018, at 12:08 PM, Roland Haas rhaas@illinois.edu wrote:
Hello Joe,
Formaline: Optimising git repository (slow only the first time)... Formaline: ERROR during: Garbage collecting git repo Command was: /usr/bin/git --git-dir='/home/jhellmer/etk/Cactus/configs/sim/configjar.git/.git' gc at /home/jhellmer/etk/Cactus/configs/sim/scratch/Formaline/bin/update-git-repo.pl line 60, <STDIN> line 5090. make[1]: *** [/home/jhellmer/etk/Cactus/configs/sim/build/Formaline/git-store-source] Error 2 make: *** [sim] Error 2
Yes that looks like some error in cleanup. This error will not prevent you from compiling Cactus, as the repository is not used for anything else but keeping track of the code you compiled. Oddly I do not remember having seen that error on Comet before (and executing that line in my own checkout I do not see the same error which may be due to subtle differences in our environment). It may be another fallout of the failed attempts before (you could run the command it gives "/usr/bin/git --git-dir='/home/jhellmer/etk/Cactus/configs/sim/configjar.git/.git' gc" manually and see if git produces a useful error message).
sbatch: error: bank_limit plugin: expired user, can't submit job sbatch: error: Batch job submission failed: Invalid account or account/partition combination specified
The error code seems to indicate that your user account is not allowed to submit jobs. Did you make sure to select an existing allocation? Either via an "allocation =" line in simfactory/etc/defs.local.ini (in the [comet] section) or via an --allocation option to the "sim submit" command?
What is the command just to execute without submitting a job? Alternatvely, how do I configure the ETK to submit successfully on Slurm for the Cluster (Comet) I’m using?
That command would be "simfactory/bin/sim create-run static_tov_test01 --parfile static_tov.par --procs 1 --num-threads 1 --walltime 24:00:00" which you tried and found to fail due to ibrun in there expecting SLURM to be present.
You can do a manual run of the executable (just to see if this even works) like so:
exe/cactus_sim par/static_tov.par 2>&1 | tee static_tov.log
which will run on the *login node* (not a good idea if you want the admins to like you). Note that this may well fail if the admins have set up their MPI stack such that you can only run MPI enabled executables in a job.
You can look at the exact SubmitScript that was passed to sbatch in /oasis/scratch/comet/jhellmer/temp_project/simulations/helloworld/output-0000/SIMFACTORY/SubmitScript
Yours, Roland
-- My email is as private as my paper mail. I therefore support encrypting and signing email messages. Get my PGP key from http://pgp.mit.edu .