Fix different timezones in logs about hydrophone

[rjsadow]

We should be able to fix this and I don't think it will be a difficult resolution. By default the conformance image is going to use the tzinfo from the host it's running on. Typically that's going to be UTC for most cloud providers (including KIND).

We can override that by setting the TZ environment variable on the conformance pod when it's created. The slightly more tricky part will be setting the value of that variable. I don't really want timezone to be a flag folks can set. Instead we can use the time package to get the local tzinfo like:

tz, offset := time.Now().Zone()

We'll need to plumb the return value of tz into the environment variable TZ on the conformance image. But that should fix it. @letmerecall are you interested in trying to code this up?

[Bharadwajshivam28]

@Bharadwajshivam28 feel free to give it a shot. Please assign it to yourself when you start working on it.

Thanks....

[letmerecall]

/assign

[rjsadow]

Hey @letmerecall how are you doing on this? Are you still interested in trying to impliment a fix?

[letmerecall]

Hey @rjsadow sorry, and yes I'll find some time and try it this weekend.

[letmerecall]

We should be able to fix this and I don't think it will be a difficult resolution. By default the conformance image is going to use the tzinfo from the host it's running on. Typically that's going to be UTC for most cloud providers (including KIND).

We can override that by setting the TZ environment variable on the conformance pod when it's created. The slightly more tricky part will be setting the value of that variable. I don't really want timezone to be a flag folks can set. Instead we can use the time package to get the local tzinfo like:

tz, offset := time.Now().Zone()

We'll need to plumb the return value of tz into the environment variable TZ on the conformance image. But that should fix it. @letmerecall are you interested in trying to code this up?

@rjsadow Well, it's a bit more complicated than just tz, offset := time.Now().Zone(). time.Now().Zone() returns an abbreviation (like IST), meanwhile TZ environment variable in conformance image expect a full string (like Asia/Kolkata).

Also, we might need different logic for difference OSes 🤦🏻‍♂️ (something like this https://stackoverflow.com/questions/68938751/how-to-get-the-full-name-of-the-local-timezone)

[Bharadwajshivam28]

Hey @rjsadow If no one is working on it then I can work on this one just I some more details

[letmerecall]

@Bharadwajshivam28 feel free to give it a shot. Please assign it to yourself when you start working on it.

[Bharadwajshivam28]

/assign

[xrstf]

meanwhile TZ environment variable in conformance image expect a full string (like Asia/Kolkata).

Are you sure? The image is based on Debian and https://www.gnu.org/software/libc/manual/html_node/TZ-Variable.html mentions that format like "utc+HH" would be valid.

However I think there's no tzdata in the image at all:

$ docker run -e "TZ=UTC" --rm --entrypoint=sh registry.k8s.io/conformance:v1.29.1 -c date
Sun Apr  7 18:31:35 UTC 2024
$ docker run -e "TZ=Asia/Kolkata" --rm --entrypoint=sh registry.k8s.io/conformance:v1.29.1 -c date
Sun Apr  7 18:31:29 Asia 2024

The name changes, but the actual times do not. Which would be super confusing.

Setting the offset does work however, but it's equally confusing (my timezone is Europe/Berlin, which is UTC+2, but TZ would require the opposite sign (the correct output would be 20:31 right now):

$ docker run -e "TZ=UTC+2" --rm --entrypoint=sh registry.k8s.io/conformance:v1.29.1 -c date
Sun Apr  7 16:31:35 UTC 2024

Setting this does not affect the go-runner (times should be 20:34), neither does setting TZ=Europe/Berlin:

$ docker run -e "TZ=UTC-02" --rm registry.k8s.io/conformance:v1.29.1
2024/04/07 18:34:37 The resultsDir /tmp/results does not exist, will create it
2024/04/07 18:34:37 Running command:
Command env: []

[k8s-triage-robot]

The Kubernetes project currently lacks enough contributors to adequately respond to all issues.

This bot triages un-triaged issues according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue as fresh with /remove-lifecycle stale
• Close this issue with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle stale