Retry Build and Functional Test Steps when timeout occurs

[jamesfredley]

Attempts to address: Connect to repository.apache.org:443 [repository.apache.org/65.109.119.155] failed: Connect timed out

max_attempts: 3
retry_wait_seconds: 600

nick-fields/retry is already used in publish

[Copilot]

Pull Request Overview

This pull request updates the Gradle workflow to use a retry action when the build step times out.

• Replaces the simple run command with a retry action from nick-fields/retry
• Sets additional parameters (timeout, max attempts, and retry wait) for improved build reliability

Comments suppressed due to low confidence (1)

.github/workflows/gradle.yml:92

• The comment indicates a normal range of 20-40 minutes, but 3600 seconds represents 60 minutes. Please update the comment or the timeout value to reflect the intended behavior.

          timeout_seconds: 3600 # normal range 20-40m

[matrei]

So, if I understand this correctly, this will retry the step for any failure, not just timeouts?
Like, if there is a genuine test failure, it will still wait 10 minutes and then retry?

[jdaugherty]

So it's my understanding that RAO will be back to normal within the week (https://the-asf.slack.com/archives/CBX4TSBQ8/p1749838588937849). These issues are temporary (but painful https://the-asf.slack.com/archives/CBX4TSBQ8/p1750077904096649). I think we should hold off on this and continue to manually retry. Auto retrying will likely lead to more issues.

[jamesfredley]

nick-fields/retry does support retry_on, although I'm not confident it is passed these failures in a way that will match up.

retry_on
Optional Event to retry on. Currently supports [any (default), timeout, error].

Every CI run failing is very painful at present.

[jamesfredley]

Interestingly on https://github.com/apache/grails-core/actions/runs/15731765021/job/44334333078?pr=14821 there were 0 timeouts on the first attempt, which is the first time I have seen that in a while.