Closed sunnycqcn closed 7 years ago
Your grid does not support array jobs which Canu requires to run. It is normally a standard feature of grid systems, yours is the first we've seen where the array jobs are administratively disabled.
I would ask if the admins can allow your jobs to run with array jobs as it would be non-trivial for you to modify Canu to run without array support. Otherwise, you would have to run Canu with useGrid=remote. Every time you get to a submit command like the above:
qsub \
-l mem=8g -l nodes=1:ppn=4 \
-d `pwd` -N "meryl_ecoli" \
-t 1-1 \
-j oe -o /scratch/snyder/f/fu115/Genome_assembly/PBonly/canutest/ecoli-autoa/correction/0-mercounts/meryl.\$PBS_ARRAYID.out \
/scratch/snyder/f/fu115/Genome_assembly/PBonly/canutest/ecoli-autoa/correction/0-mercounts/meryl.sh
It would stop and you would need to manually edit the command as directed on your grid's support page: https://www.rcac.purdue.edu/news/detail.cfm?NewsID=616 Then, once that job is done re-run your Canu command which will pick up at the next step and stop again when it reaches the next array job to submit. Otherwise, you can submit the Canu command to a single node and run it with useGrid=false which will mean it will run on only a single instance which is OK for smaller genomes (<500mb).
Hi, Thanks, If I set the command as
cd $PBS_O_WORKDIR module purge module load bioinfo module load canu /home/fu115/DIRECTORY/canu/canu-1.4/Linux-amd64/bin/canu \ -p asm -d strigaC \ genomeSize=1638.1m \ errorRate=0.035 \ -pacbio-raw /scratch/snyder/f/fu115/Genome_assembly/fastq/seq/filtered_subreads.fastq \ maxMemory=80g maxThreads=20 \ useGrid=true gridEngine="pbs" \ gridEngineThreadsOption="-pe smp THREADS" \ gridEngineMemoryOption="-l h_vmem=MEMORY" \ gridOptions="-V -S /bin/bash" \ gridOptions="-l h=blacklace01.blacklace" \ gridEngineArrayMaxJobs=75000 \ useGrid=remote
Will both 2 nodes run? Because I have right to use 2 nodes. Thanks, Fuyou
On Mon, Dec 26, 2016 at 12:51 PM, Sergey Koren notifications@github.com wrote:
Your grid does not support array jobs which Canu requires to run. It is normally a standard feature of grid systems, yours is the first we've seen where the array jobs are administratively disabled.
I would ask if the admins can allow your jobs to run with array jobs as it would be non-trivial for you to modify Canu to run without array support. Otherwise, you would have to run Canu with useGrid=remote. Every time you get to a submit command like the above:
qsub \ -l mem=8g -l nodes=1:ppn=4 \ -d `pwd` -N "meryl_ecoli" \ -t 1-1 \ -j oe -o /scratch/snyder/f/fu115/Genome_assembly/PBonly/canutest/ecoli-autoa/correction/0-mercounts/meryl.\$PBS_ARRAYID.out \ /scratch/snyder/f/fu115/Genome_assembly/PBonly/canutest/ecoli-autoa/correction/0-mercounts/meryl.sh
It would stop and you would need to manually edit the command as directed on your grid's support page: https://www.rcac.purdue.edu/news/detail.cfm?NewsID=616 Then, once that job is done re-run your Canu command which will pick up at the next step and stop again when it reaches the next array job to submit. Otherwise, you can submit the Canu command to a single node and run it with useGrid=false which will mean it will run on only a single instance which is OK for smaller genomes (<500mb).
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/marbl/canu/issues/317#issuecomment-269234336, or mute the thread https://github.com/notifications/unsubscribe-auth/AXaRKBN7WqSP68PypKAj0jaLdi6VrHjnks5rMAy5gaJpZM4LV4T3 .
-- Fuyou Fu, Ph.D. Department of Botany and Plant Pathology Purdue University USA
No, without array jobs or following my suggestion above, you can only run on a single node and have to use useGrid=false
Thanks. I got it. Fuyou
On Mon, Dec 26, 2016 at 1:06 PM, Sergey Koren notifications@github.com wrote:
No, without array jobs or following my suggestion above, you can only run on a single node and have to use useGrid=false
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/marbl/canu/issues/317#issuecomment-269235257, or mute the thread https://github.com/notifications/unsubscribe-auth/AXaRKFFmbWBfZTs4Xmna0CkhFCK7_NFEks5rMBBPgaJpZM4LV4T3 .
-- Fuyou Fu, Ph.D. Department of Botany and Plant Pathology Purdue University USA
Hi, Thanks, I tried other server. I get the error like as: [fuf@cc03]$ sh canuS.sh -- Canu v0.0 (+0 commits) r0 unknown-hash-tag-no-repository-available. -- Detected Java(TM) Runtime Environment '1.8.0_111' (from 'java'). -- Detected gnuplot version '4.4 patchlevel 0' (from 'gnuplot') and image format 'png'. -- Detected 64 CPUs and 252 gigabytes of memory. -- Detecting PBS/Torque resources. -- Undefined subroutine &canu::Configure::caExit called at /home/u1/fuf/snow/canu-1.4/Linux-amd64/bin/lib/canu/Configure.pm line 192.. Could you help me check it? Thanks, Fuyou
On Mon, Dec 26, 2016 at 1:18 PM, Fuyou Fu fufuyou@gmail.com wrote:
Thanks. I got it. Fuyou
On Mon, Dec 26, 2016 at 1:06 PM, Sergey Koren notifications@github.com wrote:
No, without array jobs or following my suggestion above, you can only run on a single node and have to use useGrid=false
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/marbl/canu/issues/317#issuecomment-269235257, or mute the thread https://github.com/notifications/unsubscribe-auth/AXaRKFFmbWBfZTs4Xmna0CkhFCK7_NFEks5rMBBPgaJpZM4LV4T3 .
-- Fuyou Fu, Ph.D. Department of Botany and Plant Pathology Purdue University USA
-- Fuyou Fu, Ph.D. Department of Botany and Plant Pathology Purdue University USA
What is in your canuS.sh script? It looks like you're setting both a grid engine and useGrid=false
and the machine you are running on is not reporting the grid configuration. Set both useGrid=0 gridEngine=undefined
to make sure it won't poll your grid.
Hi, This is my sh file. Thnks, Fuyou
/home/u1/fuf/snow/canu-1.4/Linux-amd64/bin/canu \ -p asm -d strigaC \ genomeSize=1638.1m \ errorRate=0.035 \ -pacbio-raw p6.25x.fastq \ maxMemory=80g maxThreads=20 \ useGrid=true gridEngine="pbs" \ gridEngineThreadsOption="-pe smp THREADS" \ gridEngineMemoryOption="-l h_vmem=MEMORY" \ gridOptions="-V -S /bin/bash" \ gridOptions="-l h=blacklace01.blacklace" \ gridEngineArrayMaxJobs=75000 \
On Mon, Dec 26, 2016 at 2:00 PM, Sergey Koren notifications@github.com wrote:
What is in your canuS.sh script? It looks like you're setting both a grid engine and ``useGrid=falseand the machine you are running on is not reporting the grid configuration. Set bothuseGrid=0 gridEngine=undefined``` to make sure it won't poll your grid.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/marbl/canu/issues/317#issuecomment-269238350, or mute the thread https://github.com/notifications/unsubscribe-auth/AXaRKMPzSdRkCYzFaskalklCrAk2Lo_7ks5rMBztgaJpZM4LV4T3 .
-- Fuyou Fu, Ph.D. Department of Botany and Plant Pathology Purdue University USA
You are still setting useGrid=true gridEngine="pbs"
.
You want to submit the above script to your grid and let Canu run only on the single scheduled node so you want useGrid=false gridEngine=undefined
as I said above.
Thanks, I got it. Fuyou
On Mon, Dec 26, 2016 at 2:26 PM, Sergey Koren notifications@github.com wrote:
You are still setting useGrid=true gridEngine="pbs".
You want to submit the above script to your grid and let Canu run only on the single scheduled node so you want useGrid=false gridEngine=undefined as I said above.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/marbl/canu/issues/317#issuecomment-269239754, or mute the thread https://github.com/notifications/unsubscribe-auth/AXaRKGPJYTSiTo6XOJZEJqtEXVLWjM4Yks5rMCMPgaJpZM4LV4T3 .
-- Fuyou Fu, Ph.D. Department of Botany and Plant Pathology Purdue University USA
Hi Koren, Thanks for your suggestions. I can run my job. But how can I rerun? For example, I can finish my meryl. jobSubmit.sh and meryl.sh. But if I use the same parameters for rerun, the job still stops as above error. Thanks,
On Mon, Dec 26, 2016 at 2:41 PM, Fuyou Fu fufuyou@gmail.com wrote:
Thanks, I got it. Fuyou
On Mon, Dec 26, 2016 at 2:26 PM, Sergey Koren notifications@github.com wrote:
You are still setting useGrid=true gridEngine="pbs".
You want to submit the above script to your grid and let Canu run only on the single scheduled node so you want useGrid=false gridEngine=undefined as I said above.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/marbl/canu/issues/317#issuecomment-269239754, or mute the thread https://github.com/notifications/unsubscribe-auth/AXaRKGPJYTSiTo6XOJZEJqtEXVLWjM4Yks5rMCMPgaJpZM4LV4T3 .
-- Fuyou Fu, Ph.D. Department of Botany and Plant Pathology Purdue University USA
-- Fuyou Fu, Ph.D. Department of Botany and Plant Pathology Purdue University USA
You can't continue the run, running meryl.jobSubmit.sh will just fail to submit the job again since it relies on arrays. You can run meryl.sh by hand which should take a while after which it will continue to the next step but you have to wait for it to finish before resuming. It would be easiest to start from scratch off grid.
I used the commond as:
Then get the error as: Could you help me check what is wrong? Thanks, Fuyou