We have been experiencing the overlay networking issues in our production environment. For us we see 2 different things happening on a regular basis:
- After redeploying a container, it takes anywhere from 30 seconds to 30 minutes before other containers can connect to it (the containers are very clearly up almost immediately. We experience this for all images… not just our own software)
- When we redeploy more than one container we sometimes will see the entire overlay network bounce up and down for a while causing widespread outages.
I suspect this is all tied back to the well documented DNS issues. Can you give us a rough timeframe to expect a new version to run on our nodes? Alternatively, is there a way that we can install an older version? We brought our own node.