[NOTICE] Unable to fetch/merge certain pull requests for kubernetes/kubernetes

[Stephen Augustus]

Kubernetes Community,

You may have encountered failing kubernetes/kubernetes presubmits due to git fetch errors.

This issue is tracked in detail here: https://github.com/kubernetes/test-infra/issues/19910

Bob has escalated this to the GitHub team.

Here's the latest from them:

Thanks for your patience while we investigate the errors you encountered.

We've flagged this ticket with our engineering team and they've advised the current workflow you're using is not recommended. Because updates to the API and refs are triggered by post-receive hooks, in certain circumstances they will fail to run. If these failures occur again our engineering team recommends pushing another commit (empty or otherwise) to the branch which should re-trigger the post-receive hooks to run.

Based on the above feedback from our engineering team we recommend reviewing your current workflows, or implement the work around to re-trigger post-receive hooks via another commit.

To help minimise the re-occurrence of the issues seen in this ticket the next time you have a release, I've gone ahead and created an internal issue for our engineering team to check why the repository has behaved the way it has for this release cycle. When we have more information to share, or if we require further details from you, we'll be in touch.

I'm not sure if we're out of the woods yet, but just wanted to send an email to raise visibility and provide you all a tracking issue to subscribe to.

I'll leave Bob to handle any further updates.

-- Stephen

[spi...@gmail.com]

Hi all,

To recap:

- Nov 10 - issue reported, escalated to github

- Nov 11 - updated prow's clone_refs init container to fetch by sha if available (https://github.com/kubernetes/test-infra/pull/19928)

- Nov 12-13 - sporadic reoccurrences of issue

- Nov 16 - github reports removing a high latency replica from infra resoonsible for hosting kubernetes/kubernetes (sometime since Nov 12)

We've received no further reports of this issue reoccurring in the last 7 days, so I am calling this issue closed.  Huge thanks to @alvaroaleman for updating prow, @mrbobbytables for coordinating with Github, and @justaugustus for notifying kubernetes-dev@