Description of problem:

After an extreme CPU pressure that affected etcd, the sidecar container' script for the leader election stopped working, displaying the following message:

Prerequisites (if any, like setup, operators/versions):

This happened in my OSIA cluster with OpenShift 4.x.

Steps to Reproduce

1. Install the RHODS operator in the Cluster.
2. Force a CPU outage that affects etcd (mine happened due to odh-dashboard issues).
3. Wait until all three sidecar containers fail and display that message.
4. Restart container to check if it still works.

E0828 03:02:40.453210 1 leaderelection.go:253] error retrieving endpoint: Get "https://172.30.0.1:443/api/v1/namespaces/redhat-ods-applications/endpoints/jupyterhub-ha-election": http2: no cached connection was available

Actual results:

After restarting the sidecar container (restarting only the container, not killing the pod), the script worked again.

Expected results:

The Leader Election script should always work. This shouldn't be happening to the container.