[release-3.5] server,test: refresh cache on each NewAuthStore

[vivekpatani]

• permissions were incorrectly loaded on restarts.
• server/auth: users/authentication data not loaded on member startup #14355
• Backport of server/auth: refresh cache on each NewAuthStore #14358

Signed-off-by: vivekpatani [email protected]

Please read https://github.com/etcd-io/etcd/blob/main/CONTRIBUTING.md#contribution-flow.

[mitake]

TestV3AuthRestartMember is also failing only in linux-amd64-integration-2-cpu. Let me check.

[mitake]

LGTM, the newly added test is a little bit flaky though. If it keeps failing, I think it's valuable to revisit the timeout config.
Minor request: could you revisit the email address for making DCO check success @vivekpatani ? The github one is detected as unacceptable email address.
Defer to @ahrtr

[ahrtr]

Please signoff the commit.

You used the github emal [email protected] to signoff the commit in #14410 as well, but there is no any DCO issue. It's strange.

Based on the feedback from DCO details, it seems Vivek Patani [email protected] is expected.

Anyway, could you please get this sorted out?

[serathius]

This blocks v3.5.5 release. @vivekpatani are you planning to work on this?

[vivekpatani]

@serathius yes, there is an issue with a flaky test. Sorry, was OoO.

@ahrtr @mitake I'm not able to reproduce the issue locally (or inside docker). I can increase the timeout, but I'll have to create another PR to fix that, and then rebase this once that one is merged. But given the fact, I can't reproduce the issue correctly, I'm not sure what timeout needs to be set to. Any recommendations?

@ahrtr I'll also look at fixing the signature. The github one is what github.com provides for privacy reasons but let me fix that.

[serathius]

All tests pass apart of DCO, please sign all commits by running git rebase origin/release-3.5 --signoff, and push the HEAD with force git push origin HEAD:release-3.5 -f

[vivekpatani]

@serathius I think there's a flaky test, but I've re pushed the fix for DCO.

[ahrtr]

LGTM

Thank you @vivekpatani

defer to @mitake to approve & merge this PR.

[ahrtr]

But given the fact, I can't reproduce the issue correctly, I'm not sure what timeout needs to be set to. Any recommendations?

Let's investigate the test failure in #14410

[mitake]

LGTM, thanks a lot @vivekpatani . Let’s discuss about the flaky test in #14410