What currently happens is that the runner attempts to connect, but it's killed by kubernetes due to not passing liveness before it can report a context deadline exceeded error. It doesn't log anything - it just crashloops. This gives the user no indication as to what the problem could be.
How does this fix it?
With this change, kubernetes won't kill the pod until after the client's default connection timeout window has passed, giving the runner a chance to log the fact that it couldn't reach the server.
What problem does this address?
Lets say a user runs the following command:
But
10.0.0.10is the wrong address.What currently happens is that the runner attempts to connect, but it's killed by kubernetes due to not passing liveness before it can report a context deadline exceeded error. It doesn't log anything - it just crashloops. This gives the user no indication as to what the problem could be.
How does this fix it?
With this change, kubernetes won't kill the pod until after the client's default connection timeout window has passed, giving the runner a chance to log the fact that it couldn't reach the server.