k8s
k8s copied to clipboard
startup probe to respect status of streams
During a node update or a rolling update, individual nats instances and thus also replicas of a js stream are temporarily offline. When such an instance comes back online, it may not be "in sync" with the other replicas if many new messages have arrived.
It would be desirable if the startup probe would respect this circumstance and only report an OK when all js streams are "in sync" again.
Thanks for the report, I think we may need something like /healthz?js=ready that verifies whether JS is ready and use that as the startup probe.
Healthz Startup probe is the default now and should address this