DaemonSets are special cased during host shutdown and don't get a chance to gracefully fail over. This can cause problems with both connectivity on a kill (non-grace shutdown) and sudden loss of etcd connection.
Switching to a deployment will alleviate symptoms. We original chose a daemonset for spreading and for scaling. For spreading, we now have anti-affinity rules. For scaling, we should wire into something like an HPA anyway.
The migration will require PRs back to 4.3 which remove the deployment on downgrade and in 4.4 to remove the daemonset.
Resetting status to allow follow-on PR to use the same bz.
*** Bug 1790989 has been marked as a duplicate of this bug. ***
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.