Advice on how to achieve reliable events from the APIServerSource #8843
-
|
Hello, I'm trying to achieve a reliable stream of kubernetes api server events. My setup:
I'm investigating gaps in kubernetes events (e.g. pod and replicaset events that never appear), and they seem to coincide with knative pods getting shut down (logs below). I understand that pods periodically need to be shut down and/or moved, but I had assumed that by using jetstream, the event stream would pick up where it left off and resume sending the events that were missed while the workers were shutdown. I don't believe that's happening and am wondering if there is a way to achieve this. More generally, what's the recommended approach for getting a reliable stream of kube apiserver events (i.e. resilient to pod restarts, etc.)? Thank you in advanced for your help!!
|
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 1 reply
-
|
Update: The missing events did eventually arrive! But about an hour later ... So I guess my next questions are:
|
Beta Was this translation helpful? Give feedback.
I may have found the problem. Even though I'm running a jetstream cluster with 3 replicas, the default NatsJetstreamChannel is only configured with 1 replica. I've configured the NatsJetstreamChannel to use all 3 replicas. Hopefully this fixes the issue. 🤞