Merge pull request GoogleCloudPlatform#4486 from alyssa-sm/slice-exclusivelessness

Ignore exclusive jobs for slices

Author: alyssa-sm

community/modules/scheduler/schedmd-slurm-gcp-v6-controller/modules/slurm_files/scripts/resume.py

@@ -507,13 +507,17 @@ def create_placements(nodes: List[str], excl_job_id:Optional[int], lkp: util.Loo
 def _allocate_nodes_to_placements(nodes: List[str], excl_job_id:Optional[int], lkp: util.Lookup) -> List[PlacementAndNodes]:
     # canned result for no placement policies created
     no_pp = [PlacementAndNodes(placement=None, nodes=nodes)]
-    
-    if excl_job_id and len(nodes) < 2:
-        return no_pp # don't create placement_policy for just one node
-    
+
     model = nodes[0]
     nodeset = lkp.node_nodeset(model)
 
+    is_slice = bool(getattr(nodeset, 'accelerator_topology', None))
+
+    excl_job_placement = (excl_job_id is not None) and (not is_slice)
+    
+    if excl_job_placement and len(nodes) < 2:
+        return no_pp # don't create placement_policy for just one node
+
     if lkp.is_flex_node(model):
         return no_pp # TODO(FLEX): Add support for workload policies 
     if lkp.node_is_tpu(model):
@@ -524,7 +528,8 @@ def _allocate_nodes_to_placements(nodes: List[str], excl_job_id:Optional[int], l
     max_count = calculate_chunk_size(nodeset, lkp)
 
     name_prefix = f"{lkp.cfg.slurm_cluster_name}-slurmgcp-managed-{nodeset.nodeset_name}"
-    if excl_job_id: # simply chunk given nodes by max size of placement
+   
+    if excl_job_placement: # simply chunk given nodes by max size of placement
         return [
             PlacementAndNodes(placement=f"{name_prefix}-{excl_job_id}-{i}", nodes=chunk)
             for i, chunk in enumerate(chunked(nodes, n=max_count))

community/modules/scheduler/schedmd-slurm-gcp-v6-controller/modules/slurm_files/scripts/util.py

@@ -1664,7 +1664,7 @@ def node_region(self, node_name=None):
 
     def nodeset_accelerator_topology(self, nodeset_name: str) -> Optional[str]:
         if not self.nodeset_is_tpu(nodeset_name):
-            return self.cfg.nodeset[nodeset_name].get('accelerator_topology')
+            return getattr(self.cfg.nodeset[nodeset_name], 'accelerator_topology', None)
         return None
 
     def nodeset_prefix(self, nodeset_name):