Commits on Jun 16, 2021

1. Fix: egress IP route health check detection state on restart …

   Egress IP initializes a "watcher" which health checks connectivity to egress
   nodes every 5 seconds. This is done in its sync function. However, due to a
   bug in the state initialization in WatchEgressNodes we don't set the
   different egress node fields correctly in our ADD func which screws up the health
   check detection mechanism upon a ovnkube-master restart.
   
   Essentially it does like this:
   
   1) egress setup is made
   2) there's a problem with connectivity to the egress node(s)
   3) ovnkube-master is restarted
   
   Once step 3) is made: it is expected that ovnkube-master will perform
   health checks continously and detect when the connectivity in step 2) is
   back up. However because of this problem that is not the case and even
   if further connectivity is successful, we never flag the node as such in
   our internal data.
   
   Signed-off-by: Alexander Constantinescu <aconstan@redhat.com>

   

   alexanderConstantinescu authored and openshift-cherrypick-robot committed Jun 16, 2021
   Copy the full SHA

   3d669b5 View commit details

   Browse the repository at this point in the history