slack thread here

essentially, when a readiness probe fails (a check for /etc/cni/net.d/10-ovn-kubernetes.conf)
when ovnkubenode DBs are still coming up it could fail CI if the probe fail
message is seen more than 20 times. The test case that fails is "events should
not repeat pathologically" As of the creation of this bug, it's failing in CI
less than 1% of all failed OVN jobs. The error message is benign and expected
in this scenario.

some solutions to stop this from failing CI could be:

• enhance the test code to understand this specific failure and ignore it
• reduce the polling and/or increase the poll timers so that this message
  would have no chance to occur more than 20 times in a normal scenario

example job here