FIX IS BEING WORKED ON IN OVNK UPSTREAM.
FOR TIME BEING RECOMMENDED SOLUTION: revert and lock ovnk image for microshift rebases, so ART's rebases do not overwrite to faulty ref.


Most recent ovn-k image is causing some troubles for MicroShift.

The root cause is this commit: https://github.com/openshift/ovn-kubernetes/commit/2871cdae138b10a282a3a930c5f5c516f9c523bc

Linked commit changed code to use `RunOVNControllerAppCtl` which has one slight problem: the file with PID is only read once and uses the same PID for 200 retries (every 2 seconds -> almost 7 minutes before it gives up).

For some reason, on MicroShift, the ovn-controller always fail to start cleanly on first boot: it means that after restart it creates a new socket with new PID in its filename.
When ovnkube-master reads file with PID, it doesn't see that the PID changed and there's new socket to use.

Because it tries to use a socket that exists (ovn-c did not delete it), but there's nothing on the other end, it continuously gets "Connection refused" error.

MicroShift becomes healthy again after ovnkube-master runs out of retries, quits, and restarts (so in ~7 minutes).

Every time    

    1. Deploy MicroShift main
    2. Observe ovn-k pods & logs
   
    

    ovn-k takes a long time to start, which prevents all other (non hostnetwork) Pods from starting

    ovnk sees that PID was updated and switches to using correct socket

See that ovnkube-master restarted 7 minutes after ovnkube-node:
# oc get pods -n openshift-ovn-kubernetes
NAME                   READY   STATUS    RESTARTS      AGE
ovnkube-master-l2d7x   4/4     Running   1 (11m ago)   18m
ovnkube-node-r8m7x     1/1     Running   1 (18m ago)   18m



~7 minute log of ovnkube-master continously reading the same socket file: https://drive.google.com/file/d/18MLEV2HkcJClWVn0Dq8Z5X0p5e3Hy_0R/view?usp=sharing

SOS report: https://drive.google.com/file/d/1omPiTfTpZ3oCpiYN4MaPELQoc_V-RE3W/view?usp=sharing