/kind feature

CC @linxiulei

/cc @azylinski @linxiulei

/lgtm

New metrics now look even better, @qiutongs can add a release note mentioning that the kubelet now has metrics to expose the duration of the different stages of the node startup?

LGTM label has been added.

/test pull-kubernetes-e2e-gce

unrelated failure

Kubernetes e2e suite: [It] [sig-cli] Kubectl client Simple pod should contain last line of the log expand_less	26s
{ failed [FAILED] Expected
    <string>: failed to create fsnotify watcher: too many open files
to contain substring

/lgtm

New metrics now look even better, @qiutongs can add a release note mentioning that the kubelet now has metrics to expose the duration of the different stages of the node startup?

Updated it in #118568 (comment)

I think that we are good enough for merging, this metrics are important for working on improving the startup process of the kubelet, are you ok with this @dgrisonnet ?

For approvals
/assign @SergeyKanzhelev
for kubelet
/assign @wojtek-t
for kubemark

I would still love to see #118568 (comment) addressed, but am I fine with adding the additional documentation in a follow-up PR.

I doesn't have to be super precise, but I think it would be valuable to give hints about what could be causing increase in latency at these layers.

I don't feel we can provide a good information right now about it, but we must not forget to update that metrics once we do the due diligence

I doesn't have to be super precise, but I think it would be valuable to give hints about what could be causing increase in latency at these layers.

I don't feel we can provide a good information right now about it, but we must not forget to update that metrics once we do the due diligence

I will create a follow-up PR once I can summarize the steps. It will look like:

Duration in seconds of node startup before registration, including X, Y and Z.
(X, Y, Z are some critical steps before node registration)

I don't feel we can provide a good information right now about it, but we must not forget to update that metrics once we do the due diligence

Sounds fair 👍

I will create a follow-up PR once I can summarize the steps. It will look like:
Duration in seconds of node startup before registration, including X, Y and Z.
(X, Y, Z are some critical steps before node registration)

Perfect, that's exactly what I was looking for :)

+1 on addressing #118568 (comment) but ok for the following PR. Looking forward to seeing the reports generated from these metrics. Thanks!

/lgtm
/approve

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: dchen1107, qiutongs

The full list of commands accepted by this bot can be found here.

The pull request process is described here

I was wondering if this can be backported to 1.28 or earlier. After talking to a few folks, it seems such change is not qualified as a bug fix so we generally don't backport.

If someone has different thoughts, please let me know!

/retest-required