Bug bash for 2.0?

[apostasie]

What is the problem you're trying to solve

Hey folks,

Really pumped-up about 2.0 getting closer :-) - and I thought I would suggest this:
What about a bug bash and shake-up of the bug tracker?

If you good people are open to the idea and have time for that, what about the following for a starter:

• if you have any open PR, rebase them, dust them up, and let's get them merged
• if you have any open issue, go through them and see if they are still current - close if they are no longer current evidently

Then on top of that, we could start tackling bug reports from the top:

• we have 46 unconfirmed bugs - https://github.com/containerd/nerdctl/issues?q=is%3Aissue+is%3Aopen+label%3Akind%2Funconfirmed-bug-claim
  • we should be able to cut that in half easily - if you can repro, then fine, label it "bug" - if you can't repro, maybe ping the OP?
• we have 61 confirmed bugs - https://github.com/containerd/nerdctl/issues?q=is%3Aissue+is%3Aopen+label%3Abug
  • I have 8 that are pending close with in-flight PRs
  • I will also probably close 5 more with another "auth/login" PR
  • out of the remaining
    • pretty sure a number of them are no longer current
    • there are definitely some low hanging fruits
    • a bunch of issues wrt flaky tests <- any help on these would be nice - they are tough to crack of course
    • some more involved stuff, but maybe it's a good time to have a look

Tagging @AkihiroSuda @ktock @fahedouch @Zheaoli @djdongjin @yankay @sondavidb

Describe the solution you'd like

na

Additional context

No response

[sondavidb]

Thanks for taking the initiative on this! I'll comb through and see if there's anything I can help confirm/fix.

I have one open issue myself (#3156) but it's more of a feature request than a bug, and honestly I can't recall what the ask was when I made this issue. Feel free to close it if we think current tests are sufficient.

FWIW, I work off a Windows machine so feel free to ping me if you want me to test anything.

[apostasie]

FWIW, I work off a Windows machine so feel free to ping me if you want me to test anything.

@sondavidb

That's nice. I still need to setup myself with a windows VM but this has taken a backseat in the todo list :(

I guess on my windows radar, these would be interesting to look into:

• Windows: Ensure image VOLUME temp paths are well-formed. #2160 - is this still a problem / relevant?
• Fixed volume mount issue(759) for Windows #924 - related?

This one is pending next containerd rc:

• Remove leading slash in Windows path #3304

And there is a handful of issues here, where any progress would be lovely:

• https://github.com/containerd/nerdctl/issues?q=is%3Aissue+is%3Aopen+windows+label%3Aplatform%2FWindows%2FNon-WSL2

[TinaMor]

@apostasie

I guess on my windows radar, these would be interesting to look into:

This PR Allow for binding a named pipe on Windows #2661 addresses #759. Related PRs: #2160 and #924

This one is pending next containerd rc:

Published the PR. Awaiting reviews

And there is a handful of issues here, where any progress would be lovely:

cc. @lucillex

[apostasie]

Thanks a lot @TinaMor

[apostasie]

Update:

• confirmed bugs count: 58 (down 3)
• unconfirmed bugs count: 39 (down 7)

Looking good!

We have a somewhat "large", long lasting problem collected in #3425.
Fixing the underlying cause there would likely significantly increase reliability on many operations (chiefly commit and build), and would certainly close half a dozen confirmed issues.

Help is required on this one, as it is quite tricky.

[apostasie]

Windows folks: recent changes in code / or (more likely) tests did break the windows CI significantly (now failing very often) a few weeks ago.

Details are here: #3437

The regression may have been introduced by this PR: #3304 - although this is unconfirmed.

It would be really nice to have windows savvy peeps looking into this :-)

Tagging the windowsians I know of: @sondavidb @TinaMor @jsturtevant

[apostasie]

Windows folks: recent changes in code / or (more likely) tests did break the windows CI significantly (now failing very often) a few weeks ago.

FWIW, I tried adding a sleep inside the test before we call logs (5 seconds), just in case things would be slow / racy - it still failed.