Skip to content

Commit 4655179

Browse files
committed
docs(kubernetes): refer to the SDKs collectively in replica routing
Part of NVIDIA#4223 Signed-off-by: divesh <[email protected]>
1 parent 34ad211 commit 4655179

1 file changed

Lines changed: 1 addition & 1 deletion

File tree

‎docs/kubernetes/ingress.mdx‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -257,7 +257,7 @@ helm upgrade <release-name> oci://ghcr.io/nvidia/openshell/helm-chart \
257257
--set grpcRoute.replicaRouting.enabled=true
258258
```
259259

260-
The chart renders one Service per replica and a GRPCRoute rule that matches the `x-openshell-replica` request header. The CLI and the Rust and Go SDKs send `x-openshell-want-owner: 1` on `GetSandbox`, read the owner from the `x-openshell-owner` response header, and set it as `x-openshell-replica`. If the proxy answers `UNAVAILABLE` with an empty message, `no healthy upstream`, or `connection failure`, they retry once without it. Replica routing requires `workload.kind=statefulset`, `workload.allowMultiReplicaStatefulSet=true`, and at most 15 replicas.
260+
The chart renders one Service per replica and a GRPCRoute rule that matches the `x-openshell-replica` request header. The CLI and SDKs send `x-openshell-want-owner: 1` on `GetSandbox`, read the owner from the `x-openshell-owner` response header, and set it as `x-openshell-replica`. If the proxy answers `UNAVAILABLE` with an empty message, `no healthy upstream`, or `connection failure`, they retry once without it. Replica routing requires `workload.kind=statefulset`, `workload.allowMultiReplicaStatefulSet=true`, and at most 15 replicas.
261261

262262
If the release uses `workload.kind=deployment`, this upgrade replaces the Deployment with a StatefulSet, so every gateway pod restarts at once. The StatefulSet also requests a 1Gi PersistentVolumeClaim per replica, which stays `Pending` on clusters without a default StorageClass.
263263

0 commit comments

Comments
 (0)