search

uchicago-bio / 2016-Autumn-Forum

1 stars 0 forks source link

HW5-1 slurm script giving FAILED, Exit Code 1 #45

Closed anuvedverma closed 7 years ago

anuvedverma commented 7 years ago

I tried running an mpiBLAST job in RCC but the script keeps failing... was wondering if you could point out which step I might have messed up on.

Here's the error code that got emailed to me: SLURM Job_id=20879573 Name=mpi-blast Failed, Run time 00:00:07, FAILED, ExitCode 1 Looking at the error code output, it says APPLICATION TERMINATED WITH THE EXIT STRING: Hangup (signal 1).

Here's what I did so far after logging onto the RCC cluster:

0) Copied the protein1.fasta file into my /home/anuvedverma/ directory

1) Copied the .ncbirc text from https://github.com/uchicago-bio/RCC-Utilities/wiki/Running-mpiBlast-at-RCC into av.ncbirc in my home/anuvedverma/ directory (does the name of the .ncbirc file matter here?)

2) Edited the .ncbirc file to replace all instances of abinkowski with my cnet-id, anuvedverma

3) Setup the /scratch/.../mpiblast/db directories under home/anuvedverma. The complete directory structure now looks like: /home/anuvedverma/scratch/midway/anuvedverma/mpiblast/db

4)Copied the database into my scratch directory using cp /project/mpcs56420/databases/pdb/* scratch/midway/anuvedverma/mpiblast/db

5) Copied the slurm script into /home/anuvedverma and edited it to change the email address

6) Ran the slurm script (Failed)

The only changes I made to either scripts was my CNET ID, am I missing a step somewhere in here? Let me know if it'd be easier to go over this in office hours!

Thanks, Anuved

tabinks commented 7 years ago

See #46.

Yes, the file needs to be named .ncbirc. Nothing before the ..

Keep the paths on the databases to abinkowski. They are located there and should be able to be read by everyone.

tabinks commented 7 years ago

@anuvedverma Any luck?

rdb57 commented 7 years ago

I had the same result, but it was because I didn't make the change to the database -- so that it used the files we copied to scratch. Once I did that, it was fine.

anuvedverma commented 7 years ago

I think it worked! Only thing I had to change was renaming av.ncbirc to just .ncbirc. Output looks right, but the job completed in 7 seconds... is that normal (2 nodes, 8 tasks/node)?

Also I'm still a little unclear which database we should be reading from. The script I used is using Shared=/project/mpcs56420/databases/mpi, but I'm also seeing Shared=/scratch/midway/abinkowski/mpiblast/db and Shared=/project/mpcs56420/databases/mpiblast-nr as options from the RCC-Utilities wiki. Also the homework says to query against /projects/mpcs56420/nr/. Can you clarify what the differences are?