Quick Reference Guide: Difference between revisions
Jump to navigation
Jump to search
(Created page with "=Linux= Below are the minimum Linux commands and shortcuts that you should know to do work on an HPC cluster. ==Commands== <big>'''Directories'''</big> {| class="wikitable" !...") |
No edit summary |
||
| Line 52: | Line 52: | ||
[[#top|Back to Top]] | [[#top|Back to Top]] | ||
=Slurm= | =Slurm= | ||
Below are the minimum Slurm commands that you should know to do work on an HPC cluster. | Below are the minimum Slurm commands that you should know to do work on an HPC cluster. The first table are default Slurm commands. The next table are GACRC pre-formatted wrapper scripts of the Slurm commands to provide more information and flexibility. | ||
<br /> | <br /> | ||
<br /> | <br /> | ||
'''Default Slurm Commands''' | |||
{| class="wikitable" | {| class="wikitable" | ||
! Command !! Description !! Examples | ! Command !! Description !! Examples | ||
| Line 62: | Line 63: | ||
| <code>squeue</code> || Check status currently running jobs (--me for only your jobs) || <code>squeue --me</code> | | <code>squeue</code> || Check status currently running jobs (--me for only your jobs) || <code>squeue --me</code> | ||
|- | |- | ||
| <code>sacct</code> || Check status of your jobs started since midnight of current day (-X for one line per job)|| <code>sacct -X</code> | | <code>sacct -X</code> || Check status of your jobs started since midnight of current day (-X for one line per job)|| <code>sacct -X</code> | ||
|- | |- | ||
| <code>sacct -X -S yyyy-mm-dd</code> || Check status of your jobs since yyyy-mm-dd || <code>sacct -X -S 2021-10-20</code> | | <code>sacct -X -S yyyy-mm-dd</code> || Check status of your jobs since yyyy-mm-dd || <code>sacct -X -S 2021-10-20</code> | ||
| Line 69: | Line 70: | ||
|- | |- | ||
| <code>scancel ''jobID'' || Cancel one of your running jobs || <code>scancel 456781</code> | | <code>scancel ''jobID'' || Cancel one of your running jobs || <code>scancel 456781</code> | ||
|- | |||
| <code>sinfo</code> || Check the status of partitions || <code>sinfo -p batch</code> | |||
|- | |||
|} | |||
'''GACRC Slurm Commands''' | |||
{| class="wikitable" | |||
! Command !! Slurm Equivalent !! Examples | |||
|- | |||
| <code>interact</code> || Same as <code>qlogin</code> but with more flexibility || <code>interact -c 4 --mem 10gb --time 01:00:00</code> | |||
|- | |||
| <code>sq</code> || <code>squeue</code> || <code>sq --me</code> | |||
|- | |||
| <code>sacct-gacrc -X</code> || <code>sacct</code>|| <code>sacct-gacrc -X</code> | |||
|- | |- | ||
|} | |} | ||
Revision as of 17:48, 12 October 2021
Linux
Below are the minimum Linux commands and shortcuts that you should know to do work on an HPC cluster.
Commands
Directories
| Command | Description | Examples |
|---|---|---|
ls |
List the contents of a particular directory (current dir if no path specified) | ls /scratch/MyID
|
mkdir |
Make a new directory at the specified path | mkdir /scratch/MyID/workdir
|
cd |
Change directory to a particular directory (personal /home dir if no path specified) | cd /scratch/MyID
|
rmdir |
Remove an empty directory | rmdir tmpdir
|
Files
| Command | Description | Examples |
|---|---|---|
nano |
Create or edit a file with nano(To save and exit: ctrl + x, y, enter) | nano sub.sh
|
cat |
Print the contents of a file to the terminal (better for shorter files) | cat slurm-42342.out
|
less |
View the contents of a file (better for longer files) | less slurm-42342.out
|
cp |
Copy a file or files (cp sourcepath(s) destinationpath) |
cp /scratch/MyID/workdir/sub.sh /work/labgroup
|
rm |
Remove a file or files | rm oldfile.txt
|
rm -r |
Remove a directory and all of its contents, recursively (starting with the last dir in the path) | rm -r /home/MyID/olddir
|
Shortcuts
| Shortcut | Description |
|---|---|
| ctrl + l (lowercase L) | Clear your terminal screen (scroll up to see previous output) |
| up/down arrow keys | Cycle through previously entered commands |
| esc, . (escape, then period) | Cycle through previous arguments to commands |
| ctrl + w | Backspace typed text on the command line, one word at a time |
| ctrl + c | Cancel typed text on the command line |
Slurm
Below are the minimum Slurm commands that you should know to do work on an HPC cluster. The first table are default Slurm commands. The next table are GACRC pre-formatted wrapper scripts of the Slurm commands to provide more information and flexibility.
Default Slurm Commands
| Command | Description | Examples |
|---|---|---|
sbatch |
Submit a submission script to the cluster | sbatch sub.sh
|
squeue |
Check status currently running jobs (--me for only your jobs) | squeue --me
|
sacct -X |
Check status of your jobs started since midnight of current day (-X for one line per job) | sacct -X
|
sacct -X -S yyyy-mm-dd |
Check status of your jobs since yyyy-mm-dd | sacct -X -S 2021-10-20
|
sacct -j jobID |
Check status of a particular job | sacct -j 234213
|
scancel jobID |
Cancel one of your running jobs | scancel 456781
|
sinfo |
Check the status of partitions | sinfo -p batch
|
GACRC Slurm Commands
| Command | Slurm Equivalent | Examples |
|---|---|---|
interact |
Same as qlogin but with more flexibility |
interact -c 4 --mem 10gb --time 01:00:00
|
sq |
squeue |
sq --me
|
sacct-gacrc -X |
sacct |
sacct-gacrc -X
|