Slurm Configuration Utilities¶
Back up, clean up, and rollback Slurm configuration files using Omnia utility playbooks.
Overview¶
Omnia provides the slurm_config_util.yml utility playbook with three
operations for managing Slurm configuration:
- Backup -- Create timestamped backups of all Slurm configuration files
- Cleanup -- Remove existing Slurm configuration from the live cluster directory
- Rollback -- Restore Slurm configuration from a previous backup
Prerequisites¶
- A working Slurm cluster deployed via Set Up Slurm.
Procedure¶
Backup¶
Create a timestamped backup of all Slurm configuration files:
cd /opt/omnia
ansible-playbook /opt/omnia/utils/slurm_config_util.yml --tags config_backup
Backups are stored at
<client_share_path>/slurm_backups/<backup_name>/<controller_node>/.
Cleanup¶
Remove existing Slurm configuration files from the live cluster directory:
cd /opt/omnia
ansible-playbook /opt/omnia/utils/slurm_config_util.yml --tags slurm_cleanup
Warning
Take a configuration backup before running cleanup. This action
deletes <client_share_path>/slurm/ and cannot be undone.
Rollback¶
Restore Slurm configuration from a previous backup:
cd /opt/omnia
ansible-playbook /opt/omnia/utils/slurm_config_util.yml --tags config_rollback
The utility performs the following steps:
- Lists available backups
- Validates the selected backup
- Optionally creates a safety backup of the current configuration
- Restores files from the selected backup
- Fixes file permissions
- Restarts Slurm services
Verification¶
-
Verify the current configuration is valid after a rollback:
Run on: Slurm controller nodesinfo scontrol ping -
Verify all nodes are healthy:
Run on: Slurm controller nodescontrol show nodes | grep State
Next Steps¶
- Add Slurm Nodes -- Back up before adding nodes
- Remove Slurm Nodes -- Back up before removing nodes
Troubleshooting¶
slurmctld fails to start after config rollback
The backup may reference nodes that no longer exist. Run cleanup and redeploy:
ansible-playbook /opt/omnia/utils/slurm_config_util.yml --tags slurm_cleanup
ansible-playbook provision.yml
For the complete list, see Slurm Issues.