Skip to content

Commit

Permalink
when serviceAcountName is gdr add the IPC_LOCK capability (#100)
Browse files Browse the repository at this point in the history
This is a kludge to sidestep file system permission problems in
shared PVs for namespaces where multi-pod jobs run using
roce_gdr (thus as root). In these namespaces, using the gdr
serviceAccount will now always imply running as root.
  • Loading branch information
dgrove-oss authored Nov 19, 2024
1 parent 3fa9a98 commit 4cdd684
Showing 1 changed file with 1 addition and 1 deletion.
2 changes: 1 addition & 1 deletion tools/pytorchjob-generator/chart/templates/_helpers.tpl
Original file line number Diff line number Diff line change
Expand Up @@ -277,7 +277,7 @@ imagePullSecrets: []


{{- define "mlbatch.securityContext" }}
{{- if gt ( int .Values.numRoceGdr ) 0 }}
{{- if or (gt ( int .Values.numRoceGdr ) 0) (eq .Values.serviceAccountName "gdr") }}
securityContext:
capabilities:
add:
Expand Down

0 comments on commit 4cdd684

Please sign in to comment.