3 ms·
I switched to k8s about a year and a half ago before deploying a modest frontend/backend pair. although there is some friction, I generally like the approach (
by dekhn 9y ago
I switched to k8s about a year and a half ago before deploying a modest frontend/backend pair. although there is some friction, I generally like the approach (I used to use borg, so it's a pretty low barrier).
The biggest problem I have is debugging all the moving parts when there are ~10+ minute async responses to config changes.
- majewsky 9y ago10 minutes? That sounds a lot. We usually have 1-2 minutes wait time for a changed ConfigMap to reach a running pod, or for a Deployment to roll over to a new ReplicaSet. (That's on k8s 1.4.)