-
Bug
-
Resolution: Unresolved
-
Normal
-
None
-
4.20.z
-
None
-
None
-
False
-
-
None
-
Important
-
None
-
None
-
None
-
Proposed
-
None
-
Customer Escalated
-
None
-
None
-
None
-
None
-
None
-
None
-
None
Description of problem:
Partner customer is testing product resiliency. After restarting worker nodes with ingress pods, the ingress operator stops accepting connections. Pods are healthy and work. Requests from VIP or Node IP receive the error: [...] SSL_connect: SSL_ERROR_ZERO_RETURN [...]
Version-Release number of selected component (if applicable):
4.20.15
How reproducible:
Always
Steps to Reproduce:
# Console works! 1. Stop worker 1 (with ingress pod). wait for ingress pod to launch in available worker. start node again. # Console works! 2. Stop worker 2 (with newer ingress pod). wait for pod to launch in available worker. start node again. # Console works! 3. Stop the third node (the oldest ingress pod). ingress/console operators stops working. # Console fails!
Actual results:
Ingress operator stops working. Canary errors.
Expected results:
The opposite.
Additional info:
The environment is: Baremetal Hub cluster with: - Hypershift/MCE - Virtualization Hosted Control plane managing 3 worker nodes that are VMs hosted on the Hub Cluster.
