How to compute/Quick start

From MetaCentrum
Jump to: navigation, search

(Česká verze)

Related topics
How to compute
PBS Pro Quick Start [PDF]

}} This Step by Step tutorial shows how to run a job on MetaCentrum machines, see the How to compute topic for more information or follow steps bellow to run an example job. You can also see video tutorial, which covers this topic (works with PuTTY and Windows).

ZarovkaMala.png Note: A terminal is required, Windows users might want to use PuTTY.

1. Log on a frontend

Related topics
Accessing machines

Write the line below into your terminal and fill in your META password:

ssh 'META username'@'frontend'
Password: <Enter META password>

Please type your MetaCentrum username, which you set during registration. You can choose any front end of MetaCentrum, e.g.

2. Prepare a startup script

Related topics
Batch jobs

Create text file (in your favourite text editor) with the following lines:

# sets home directory
# loads the Gaussian's module, version 03
module add g03
# setup SCRATCH cleaning in case of an error
trap 'clean_scratch' TERM EXIT
# enters user's scratch directory
cd $SCRATCHDIR || exit 1
# prepares job's input data
# starts the Gaussian application
g03 < >results.out
# moves the produced (output) data to user's home directory or leave it in SCRATCH if error occured 
cp results.out $DATADIR || export CLEAN_SCRATCH=false

3. Specify job's parameters & submit the job

Related topics
Requesting resources

Type the following line into your terminal with the script's name at the end:

qsub -l select=1:ncpus=2:mem=4gb:scratch_local=10gb -l walltime=1:00:00

You should see an output with JobID is 12345 in this case.

4. Track job's status

Related topics
Tracking job status

You can use jobID and qstat command to track the job. Type line below into your terminal:

qstat 'jobID' # writes info about running and queued jobs
qstat -x 'jobID' # writes infor about finished jobs

You should see an output similar to these lines:

                                                                   Req'd  Req'd   Elap
Job ID          Username        Queue    Jobname    SessID NDS TSK Memory Time  S Time
--------------- --------------- -------- ---------- ------ --- --- ------ ----- - -----
12345.arien-pr 'your username'  q_2h   6866   1   8   2gb    01:00 F 00:00 # F=Finished, R=running, Q=Queued

5. Examine job's standard output and standard error output

Related topics
Examining outputs

When the job is completed you can examine it's outputs in your terminal:

cat ~/results.out   # job's output data
cat ~/   # job's standard output, empty for now (forwarded to results.out)
cat ~/   # job's standard error output