Interactive execution is performed using Slurm's interactive batch function. The tssrun command is provided as a tool for simple execution. You can execute jobs by specifying program and arguments to be executed after the command. Please refer to here for environment variables that are automatically set when a job is executed.
Command | Description | Examples | Notes |
---|---|---|---|
tssrun | Execute the program on the computing node | tssrun ./a.out |
Option | Description | Examples |
---|---|---|
-p QUEUENAME | Specify the queue for batch processing | tssrun -p gr19999b ./a.out |
-t HOURS:MINUTES:SECONDS | Specify the upper limit of elapsed time (unit: hours:minutes:seconds) | tssrun -p gr19999b -t 1:0:0 ./a.out |
--rsc p=PROCS:t=THREADS:c=CORES:m=MEMORY or --rsc g=GPU |
Specify the amount of job-allocated resources. For more details, click here | tssrun -p gr19999b --rsc p=4:t=8:c=8:m=2G ./a.out or tssrun -p gr19999b --rsc g=1 ./a.out |
--x11 | Execute GUI program on computing nodes | tssrun -p gr19999b --x11 xeyes |
Example 1: For MPI program in System B
$ tssrun -p gr19999b --rsc p=4 ./a.out
salloc: Granted job allocation 102362
salloc: Waiting for resource configuration
salloc: Nodes cnode3 are ready for job
My name is cnode3. My rank is 1.
My name is cnode3. My rank is 2.
My name is cnode3. My rank is 3.
My name is cnode3. My rank is 0.
salloc: Relinquishing job allocation 102362
exit code: 0
Example 2: When executing GUI programs
$ tssrun -p gr10001b --x11 xeyes
salloc: Granted job allocation 102366
salloc: Waiting for resource configuration
salloc: Nodes <Computing Node name> are ready for job
(X application starts)
You can confirm the job information with an interactive execution by tssrun command.
Command | Option | Description | Example |
---|---|---|---|
sacct | -j JOBID | Display job information for interactive execution | sacct -j 102362 |
You must specify the job ID followed by the command and options.
The job ID is the XXXXXXX part of salloc: Granted job allocation XXXXXXXX
that is displayed when the tssrun command is executed.
If more than 24 hours elapsed after the interactive execution ends, it can not be displayed.
If an interactive execution terminates abnormally, the reason for the termination is displayed in the job information.
Termination due to memory limitation (due to pressure on overall node memory)
Can't malloc: Cannot allocate memory
Termination due to elapsed time limit
slurmstepd: error: *** STEP 102368.0 ON <computing node name> CANCELLED AT 2022-11-04T15:34:36 DUE TO TIME LIMIT ***