The dates of the kubernetes

Question

I would like to see all the different things that happened to the cube cluster on the timeline, including when dead nodes were discovered, when new nodes were added, when containers broke and when they were restarted.

So far, the best we have found, kubectl get eventbut this, apparently, has several limitations:

• he will not return in time so far (I'm not sure how far he will return. Day?)
• he combines such events and orders the list by the time of the last event in each group. This makes it impossible to know what happened over a period of time, as events in this range may have been combined with later events outside the range.

One idea I have is to write a module that will use the API to view the stream of events and write them to a file. This would allow us to control the hold, and it seems that the events that occur during viewing will not be combined, solving the second problem.

What do other people do about this?

Answer 1

• I understand that Kubernetes itself is deduplicating the events described here: https://github.com/kubernetes/kubernetes/blob/master/docs/design/event_compression.md Once this happens, there is no way to return individual events.

• ? " ", -, kube-apiserver --event-ttl, - 1 : https://github.com/kubernetes/kubernetes/blob/da53a247633/cmd/kube-apiserver/app/options/options.go#L71-L72

  . etcd - , github 2015 , TTL 2 , , etcd...

, , kubelet?

• kubectl get event -o yaml --watch . , , pre-dedup.

• Heapster : https://github.com/kubernetes/heapster/blob/master/docs/sink-configuration.md

Answer 2

Dashboard?

, GKE:

Answer 3

kubernetes / heapster can transfer the event to gcl and influxdb, but so far there is no api to access the stored data