chore(deps): bump k8s.io/klog/v2 from v2.130.1 to v2.140.0

[pierluigilenoci]

Summary

Routine vendored dependency update: bumps k8s.io/klog/v2 from v2.130.1 to v2.140.0.

Note on original PR description: The original description claimed this would improve severity-based log filtering. @zerbitx correctly pointed out that devspace already routes all klog output to io.Discard via SetOutput, so the application-level change was a no-op. This updated PR is honest about that — it is a pure routine version bump.

Changes

• go.mod: bump k8s.io/klog/v2 v2.130.1 → v2.140.0
• vendor/: update vendored klog/v2 to v2.140.0

[netlify]

✅ Deploy Preview for devspace-docs canceled.

Built without sensitive environment variables

Name	Link
🔨 Latest commit	02b65d3
🔍 Latest deploy log	https://app.netlify.com/projects/devspace-docs/deploys/6a33875a74a1af0008ea4c7c

[pierluigilenoci]

Hi — friendly follow-up. CI is green and all checks pass. Would you be able to review when you get a chance? Thank you!

[zerbitx]

The state after the function runs looks like:

• logtostderr=false (set last, as before)
• stderrthreshold=INFO (new — added by this PR)
• legacy_stderr_threshold_behavior=false (new — added by this PR)

With logtostderr false, we skip this and make it into the also to stderr bit so we'll have both err and info. I haven't run it, but I think this will make the logs noiser.

[pierluigilenoci]

Thank you for the review and the detailed walkthrough!

You're right. I traced the execution path more carefully:

Since logtostderr=false is set, the output() function enters the else branch (not the else if l.toStderr branch). In that branch, the "also to stderr" condition is:

(!l.alsoToStderr && s >= l.stderrThreshold.get())

With stderrthreshold=INFO and alsoToStderr=false (default), this evaluates to true for every log message — so all klog output would be written to stderr, making things noisier rather than quieter.

The legacy_stderr_threshold_behavior flag only applies inside the else if l.toStderr branch, which is never reached here because logtostderr=false. So it has no effect on this code path.

The fix is to remove the stderrthreshold=INFO line. I'll push an updated version — would you prefer I also drop the legacy_stderr_threshold_behavior=false line (making it a minimal diff: just the klog version bump), or keep it as a no-op signal of awareness of the new flag?

[pierluigilenoci]

Thanks @zerbitx for the detailed walkthrough — you were right.

I traced through the code again: with logtostderr=false, output() enters the else branch (not else if l.toStderr). In that branch the condition to also write to stderr is !l.alsoToStderr && s >= l.stderrThreshold.get(). With stderrthreshold=INFO that evaluates to true for every log level, so everything was being routed to stderr — the opposite of the intended suppression. The legacy_stderr_threshold_behavior flag has no effect because the else if l.toStderr branch is never reached when logtostderr=false.

I've pushed an updated version that drops both flag registrations. The change is now just the klog/v2 bump to v2.140.0 plus logtostderr=false + SetOutput(io.Discard) — consistent with how klogv1 is already initialized in the same function.

[pierluigilenoci]

Hi @zerbitx — thanks for the review that caught the no-op issue. I've simplified the PR to an honest routine dep bump (k8s.io/klog/v2 v2.130.1 → v2.140.0) with no behavioral changes, and updated the description accordingly. Would a pure version bump like this be welcome, or would you prefer I close this? Happy either way — just want to avoid leaving a zombie PR. Thanks!