Refactor obtaining "resume_data" and bulk-grouping nodes

GoogleCloudPlatform · Dec 3, 2024 · 6988f56 · 6988f56
1 parent eb8dc53
commit 6988f56
Show file tree

Hide file tree

Showing 7 changed files with 277 additions and 219 deletions.
diff --git a/.../scheduler/schedmd-slurm-gcp-v6-controller/modules/slurm_files/scripts/get_tpu_vmcount.py b/.../scheduler/schedmd-slurm-gcp-v6-controller/modules/slurm_files/scripts/get_tpu_vmcount.py
@@ -57,7 +57,7 @@ def get_vmcount_of_tpu_part(part):
             valid = PART_INVALID
             break
         else:
-            if util.part_is_tpu(part):
+            if util.lookup().partition_is_tpu(part):
                 vmcount = get_vmcount_of_tpu_part(part)
                 if vmcount == -1:
                     valid = DIFF_VMCOUNTS_SAME_PART