search

zephyrproject-rtos / zephyr

Primary Git Repository for the Zephyr Project. Zephyr is a new generation, scalable, optimized, secure RTOS for multiple hardware architectures.
https://docs.zephyrproject.org
Apache License 2.0
10.93k stars 6.65k forks source link

Documentation workflow stuck blocking PR CI completion #80260

Open dleach02 opened 1 month ago

dleach02 commented 1 month ago

The documentation build workflow is stuck without a path to restart.

Speculation is that it will be stuck if the PR has no reason to do documentation generation.

Behavior shows CI waiting on completion of all tasks but the documentation workflow has the following error:

Invalid workflow file: .github/workflows/doc-build.yml#L264
error parsing called workflow ".github/workflows/doc-build.yml" -> "./.github/workflows/ready-to-merge.yml" : failed to fetch workflow: workflow was not found.

See #80016 as a reference.

dleach02 commented 1 month ago

ping @nashif

kartben commented 1 month ago

Looking at recent jobs it's only ever happened once on another PR https://github.com/zephyrproject-rtos/zephyr/actions/runs/11427402099 This is really odd and while I investigate, or at least keep an eye on future failed doc build jobs, please try retriggering CI on the linked PR. I am almost certain this will do the trick (looks like it did for the other PR I just mentioned since issue seems to have magically fixed itself for @faxe1008 as I didn't hear from him :))

kartben commented 1 month ago

Speculation is that it will be stuck if the PR has no reason to do documentation generation.

Something's off, for sure, but it's definitely not what's happening as many PRs in that case ran just fine, ex. https://github.com/zephyrproject-rtos/zephyr/actions/runs/11469657074.

dleach02 commented 1 month ago

Looking at recent jobs it's only ever happened once on another PR zephyrproject-rtos/zephyr/actions/runs/11427402099 This is really odd and while I investigate, or at least keep an eye on future failed doc build jobs, please try retriggering CI on the linked PR. I am almost certain this will do the trick (looks like it did for the other PR I just mentioned since issue seems to have magically fixed itself for @faxe1008 as I didn't hear from him :))

I don't know how to retrigger the CI in this case. There is no "restart failed jobs" path.

kartben commented 1 month ago

Looking at recent jobs it's only ever happened once on another PR zephyrproject-rtos/zephyr/actions/runs/11427402099 This is really odd and while I investigate, or at least keep an eye on future failed doc build jobs, please try retriggering CI on the linked PR. I am almost certain this will do the trick (looks like it did for the other PR I just mentioned since issue seems to have magically fixed itself for @faxe1008 as I didn't hear from him :))

I don't know how to retrigger the CI in this case. There is no "restart failed jobs" path.

Like I said in the linked PR force-push (after a rebase or, less disruptive, a simple commit amend) will typically do just that (and without dismissing existing reviews)

dleach02 commented 3 weeks ago

Discussed in release meeting that this may not be so broad of an issue. @kartben will continue to monitor.