Closed MerlinDMC closed 7 months ago
any idea why the E2E for 1.27 is failing? is it a transient problem - can we retry the job?
Sometimes single test-run's are failing. This is unrelated to your changes.
I've opened two additional PR to add the flag to the configuration properties and parameter docs.
@MerlinDMC FYI I rebased on top of main and force-pushed to attempt to overcome unexplained E2E errors. Will merge when we get a green signal.
This PR adds a
--drain-delay
flag. The duration given will delay the node draining process to allow for other controllers in the cluster to complete work.We are using
--pre-reboot-node-labels=node.kubernetes.io/exclude-from-external-load-balancers=true
to signal the load balancer controller to start draining connections from the node as it is about to be drained. With the flag added we can delay the draining process to ensure the load balancer controller has enough time to remove the node from active traffic flow before the draining starts.