You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The arm64 variant has its own `app.kubernetes.io/name`, so the chart made
a second Service for it. The HTTPRoute backs only the amd64 Service, so
the arm64 pods never received traffic.
Every pod template now carries `app.kubernetes.io/component`, which holds
the workload name and is the same for all architectures. When arm64 is
enabled, the Service and the PodDisruptionBudget select that label, so
one Service reaches the pods of both architectures and the load balancer
sends traffic to both. When arm64 is off, the selector does not change.
No Deployment selector changes, so no workload is recreated. The label is
added before it is selected, so an environment that enables arm64 later
gets no interval without endpoints.
Bump the chart first, then enable arm64. In one sync the amd64 pods would
still roll to get the label while the selector already moved.
0 commit comments