slurm/SLURM-Cheat-Sheet.md

+## sacct
+```shell
+-a # all jobs
+-b # brief 
+-g # specify a group to look at
+-i # specify a node/nodes
+-s # state of jobs PD=pending R=running CP=completed
+```
+example: Will show all jobs PENDING for stat-grad
+```shell
+sacct -a -g stat-grad -s PD
+```
+
+## sacctmgr
+```shell
+list users # will show a list of users and default accounts
+list account # show a list of accounts/groups
+```
+for a list of all users with groups
+```shell
+sacctmgr list users format=user%-20 format=DefaultAccount%-30
+```
+for a single user 
+```shell
+sacctmgr list user $(whoami) format=user%-20 format=DefaultAccount%-30
+```
+list all groups 
+```shell
+sacctmgr list account format=account%-40 format=description%-40
+```
+show users in a given group (account)
+```clike
+sacctmgr show account domain-users -s format=user%-20 format=account%-30
+```
+Adding users
+```shell
+sacctmgr add user name.n account=group-name # -i flag is like assumeyes
+```
+removing users
+```shell
+sacctmgr remove user name.n account=group-name 
+```
+#### Changing accounts and Default accounts
+##### add the account to the user
+```shell 
+sudo sacctmgr add user name.n account=new-group
+```
+##### modify the Default account
+```shell
+sudo sacctmgr modify user where user=name.n set defaultaccount=new-group
+```
+##### (optional) Remove the old entry 
+```shell
+sacctmgr remove user where user=name.n and account=old-group
+```
+##### Running a job as a different account
+```shell
+sinteractive -p stat -A stat-users
+```
+- specify the partition and account and it should work. 
+
+## scontrol
+### extending wall time
+```shell 
+sudo scontrol update jobid=****** TimeLimit=21-00:00:00 #Days-Hours:Minutes:Seconds
+```
+
+## running jobs
+### sinteractive
+```shell
+-p # specify partition to run on
+```
+
+## checking running job status
+#### squeue
+```shell
+-r  # array
+-u  # user
+-A  # account 
+-w  # node or nodelist
+-t  # state code 
+```
+##### [[Slurm Job State Codes]]
+```shell
+watch squeue -u $(whoami)
+```
+##### watch accepts a -n to specify interval, defaults to 2.0 seconds
+```shell
+watch -n 1 squeue -u $(whoami)
+```
+##### see currently running jobs
+```shell
+squeue -t R
+```
+
+
+#### Job Efficiency 
+```shell
+seff $jobID
+```
+
+#### node health
+##### show partition states
+```shell
+sinfo -p batch
+```
+##### show reasons why a node is a certain state
+```shell
+sinfo -n u141 --list-reasons
+```
+```shell
+sinfo -n u141 -N -a -l
+```
+#### sinfo
+```shell
+-a # --all Display information about all partitions.
+-n #  --nodes=<nodes>
+-p # --partition=<partition>
+
+-N # --Node Print information in a node-oriented format 
+-l # --long Print more detailed information.
+-s # --summarize
+```
+```shell
+sinfo -n u021 -N -a -l
+```
+
+## global configuration of defaults
+```clike
+scontrol show config
+```
\ No newline at end of file