Correct Suspend/Resume backoffLimit for Pathways

AI-Hypercomputer · Jul 8, 2024 · c3c5a29 · c3c5a29
1 parent ed4a896
commit c3c5a29
Showing 1 changed file with 2 additions and 1 deletion.
diff --git a/xpk.py b/xpk.py
@@ -237,7 +237,7 @@ class CapacityType(enum.Enum):
  labels:
  xpk.google.com/workload: {args.workload}
  spec:
- backoffLimit: 4
+ backoffLimit: {backoff_limit}
  completions: {system.vms_per_slice}
  parallelism: {system.vms_per_slice}
  template:
@@ -6206,6 +6206,7 @@ def workload_create(args) -> None:
  ].resource_type,
  local_queue_name=_LOCAL_QUEUE_NAME,
  autoprovisioning_args=autoprovisioning_args,
+ backoff_limit=system.vms_per_slice * 4,
  )
  else:
  container, debugging_dashboard_id = get_user_workload_container(