Is the moving of pods logged?

[max-rocket-internet]

Because all I see is this:

I0502 14:44:16.294689       1 rescheduler.go:247] No nodes to process.
I0502 14:44:26.443325       1 rescheduler.go:247] No nodes to process.
I0502 14:44:36.574006       1 rescheduler.go:247] No nodes to process.
I0502 14:44:46.724541       1 rescheduler.go:247] No nodes to process.
I0502 14:44:56.940652       1 rescheduler.go:247] No nodes to process.
I0502 14:45:07.118794       1 rescheduler.go:247] No nodes to process.
I0502 14:45:17.301454       1 rescheduler.go:247] No nodes to process.
I0502 14:45:27.464235       1 rescheduler.go:247] No nodes to process.

And sometimes this:

I0502 14:46:28.288639       1 rescheduler.go:205] Waiting for unschedulable pods to be scheduled.

[sercanacar]

@max-rocket-internet did you get to the bottom of this?

[max-rocket-internet]

It's not logged but it should definitely be.

[JoelSpeed]

Hey @max-rocket-internet and @sercanacar, which log setting do you have? The rescheduler uses glog and we typically run with level 2 -v=2 at @pusher. You are right that there is not specific log line for each individual pod being evicted, though i don't see why we couldn't add one, probably somewhere within this method

k8s-spot-rescheduler/scaler/scaler.go

Line 71 in b46cd36

func DrainNode(node *apiv1.Node, pods []*apiv1.Pod, client kube_client.Interface, recorder kube_record.EventRecorder,

[max-rocket-internet]

which log setting do you have?

I'm just using the chart: https://github.com/helm/charts/tree/master/stable/k8s-spot-termination-handler

there is not specific log line for each individual pod being evicted

I would say this is very important to log at default log level. Much more important than No nodes to process. 😄

[JoelSpeed]

Yeah you're probably right... Levelled logging decisions are hard 😅

[aimanparvaiz]

Any update on this please?