dependabot PR merges can result in failed deploy-docs workflows

[noelmcloughlin]

Yesterday I merged a bunch of DependaBot PRs across my repositories and all deploy-docs.yaml jobs failed.

All repositories had similar form of message:

To https://github.com/lmodel/d3fend
 ! [rejected]        gh-pages -> gh-pages (fetch first)
error: failed to push some refs to 'https://github.com/lmodel/d3fend'
hint: Updates were rejected because the remote contains work that you do not
hint: have locally. This is usually caused by another repository pushing to
hint: the same ref. If you want to integrate the remote changes, use
hint: 'git pull' before pushing again.

Fixing this manually is labor intensive since I updated a number of repositories.

It would be better to --force push to gh-pages since fixing these type of failures is not trivial.

[dalito]

I have not yet seen such error and wonder what caused it. The job was this one.
Do you think that it results from dependabot making changes while the publishing pipeline was still running in parallel? Did you try to rerun the failed job?

[noelmcloughlin]

Hi David,

I'm not sure - but something in the flow broke deploy-docs on multiple repos (see screenshots) - all I done was merge some Dependabot PRs. Now I have to update manually. I google the error and see intermittent user reports - some projects applying --force solution, others choosing git pull and a variety of causes were discussed including a github flakiness but I did not have time to dig deep as.

[noelmcloughlin]

I have not yet seen such error and wonder what caused it. The job was this one. Do you think that it results from dependabot making changes while the publishing pipeline was still running in parallel? Did you try to rerun the failed job?

I watched some pipelines yesterday - initial commit to new repo - and two deploy-docs jobs were triggered (one from my commit, once from dependabot PR) and they ran in parallel but completed successfully. I presume gh-pages branch is ok.

[dalito]

GitHub flakiness seem most likely to me.

I am a bit hesitant to add the --force option since it can be destructive. Maybe we should wait a bit if it repeats.

[dalito]

Closed by #154. Using the --force option is the recommended way.