feat(slurmctld, slurmd): make slurm respect memory constraints on com…

…pute nodes * enhancement(slurmctld): enforce job level memory These changes introduce default values for SelectTypeParameters and MemSpecLimit. `SelectTypeParameters=CR_CPU_Memory` to enable memory allocation enforcement. `MemSpecLimit=1024` to ensure systems always have 1G available memory that jobs cannot consume. --------- Co-authored-by: Jason Nucciarone <[email protected]>
charmed-hpc · Nov 25, 2024 · 46b4f17 · 46b4f17
1 parent 70657d6
commit 46b4f17
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 0 deletions.
diff --git a/charms/slurmctld/src/constants.py b/charms/slurmctld/src/constants.py
@@ -18,6 +18,7 @@
     "PluginDir": ["/usr/lib/x86_64-linux-gnu/slurm-wlm"],
     "PlugStackConfig": "/etc/slurm/plugstack.conf.d/plugstack.conf",
     "SelectType": "select/cons_tres",
+    "SelectTypeParameters": "CR_CPU_Memory",
     "SlurmctldPort": "6817",
     "SlurmdPort": "6818",
     "StateSaveLocation": "/var/lib/slurm/checkpoint",

diff --git a/charms/slurmd/src/charm.py b/charms/slurmd/src/charm.py
@@ -326,6 +326,7 @@ def get_node(self) -> Dict[Any, Any]:
         node = {
             "node_parameters": {
                 **machine.get_slurmd_info(),
+                "MemSpecLimit": "1024",
                 **self._user_supplied_node_parameters,
             },
             "new_node": self._new_node,