Update/gpu-shell-command (#12)

* Update/gpu-shell-command * Update docs/src/usage/run-workload-examples.md Co-authored-by: Yongun Seong <[email protected]> * Update docs/src/usage/run-workload-examples.md Co-authored-by: Yongun Seong <[email protected]> --------- Co-authored-by: Yongun Seong <[email protected]>
bacchus-snu · Dec 18, 2024 · d64faed · d64faed
1 parent 2e0d022
commit d64faed
Showing 1 changed file with 6 additions and 1 deletion.
diff --git a/docs/src/usage/run-workload-examples.md b/docs/src/usage/run-workload-examples.md
@@ -133,6 +133,11 @@ will be **irrecoverably lost** if the PersistentVolumeClaim is deleted.
 
 Use this example to spawn an ephemeral shell with access to GPU resources.
 
+<div class="warning">
+
+The example GPU shell will **auto terminate after 25 minutes, DO NOT** use for consistent workloads.
+</div>
+
 ```yaml
 # gpu-shell.yaml
 apiVersion: v1
@@ -144,7 +149,7 @@ spec:
   containers:
     - name: app
       image: nvcr.io/nvidia/cuda:12.5.0-base-ubuntu22.04
-      command: ['/bin/bash', '-c', 'sleep inf']
+      command: ['/bin/bash', '-c', 'sleep 1500 && echo "Time expired. Exiting..." && exit']
       resources:
         limits:
           nvidia.com/gpu: 4