strigger

Register a trigger for system events

$ strigger --set --primary_slurmctld_failure -p [path/to/script]

$ strigger --set -j [job_id] -f -p "[path/to/script] [args]"

$ strigger --get

$ strigger --get -j [job_id]

$ strigger --clear [trigger_id]

$ strigger --set --down --flags=PERM -p [path/to/script]

strigger --set [OPTIONS...]
strigger --get [OPTIONS...]
strigger --clear [OPTIONS...]

strigger manages event triggers in Slurm that automatically execute scripts when specific events occur. Triggers can respond to node state changes, job completions, system failures, and other cluster events. The daemon checks triggers approximately every 15 seconds.
Triggered programs execute on the slurmctld node with a 5-minute timeout. By default, triggers are removed after execution unless marked permanent with --flags=PERM.

--set

Register a new trigger

Display registered triggers

Remove a trigger

Target specific job

Target specific node

Script to execute when trigger fires

Timing adjustment (negative = before event)

Trigger flags (PERM = permanent)

Filter by trigger creator

Detailed output

Suppress non-critical errors

Node events: --down, --up, --drained, --draining, --idle, --fail
Job events: --fini, --time
System events: --primaryslurmctldfailure, --primaryslurmdbdfailure, --reconfig

Only SlurmUser (typically root) can set triggers. Programs run on slurmctld node, not compute nodes. Triggers must be re-registered after execution unless permanent. Scripts must be executable and accessible from slurmctld host.

strigger is part of Slurm, developed at Lawrence Livermore National Laboratory starting in 2002. Event triggers allow automated responses to cluster events for monitoring, alerting, and workflow automation. Slurm is now maintained by SchedMD.

scontrol(1), slurmctld(8), slurm.conf(5)