Closed jjoets closed 7 years ago
You shouldn't set the gridEngineArrayOption=-t 1-100
, that implies all submissions should have 100 jobs which is not correct. Canu will decide the number of jobs to submit based on your coverage , genome size, and available processes. If you want to limit concurrency you want to use the -tc option (gridEngineArrayOption=`-t ARRAY_JOBS -tc 100
. This will limit all array jobs to at most 100 instances. You could also use a parallel environment which restricts the number of CPUs which can be assigned at one time.
Now for the error, what's the output of one of the failed jobs in (something like correction/asm.ovlStore.BUILDING/logs/2-sort.1070.out
)?
Thanks a lot for your advices. I updated the .spec file and re-started Canu which looks to run fine now. Wait and see.
Unfortunately, the directory ovlStore.BUILDING has disappeared. Sorry, but I cannot copy the file content. However as far as i remember the logs were fine and suddenly were interrupted by a SGE message similar to what I reported yesterday.
Johann
Time limit or memory would be the likeliest reason for the jobs being terminated. You could try specifying a longer time limit using gridOptions
which gets passed through to all jobs.
Glad the resume is working for you, if you encounter other errors please open a new issue.
Hi,
I run this command lines
with this .spec file
In a first attempt, I ran Canu without the gridEngineArrayOption=-t 1-100 option but the cluster was close to fall due to high io . So we killed the job, set the gridEngineArrayOption option and run Canu again (same command line).
But now I get :
Thanks for your help
Johann