I came across a weird behaviour (at least not as expected) when using the
concurrency field. I’m using a fixed concurrency group like
concurrency: deploy-request. The related workflow creates a new commit whenever the
repository_dispatch event occurs.
If I now trigger two jobs in a row, the first job creates a new commit and the job passes. However, the latter seems to operate on the previous HEAD. The step
cat output.log does not contain the value generated by the first job see and git push fails.
As a workaround I run
git pull origin main right after the checkout
Is this behaviour intentional / by design?