This message was deleted.

I suspect it might have something to do with the default settings for the Fargate container -- I'm seeing stuff like this:

will re-clone repo, could not determine if was at correct commit: git rev-parse HEAD: exit status 128: fatal: detected dubious ownership in repository at '/home/atlantis/.atlantis/repos/REDACTED/2842/default'\nTo add an exception for this directory, call:\n\n\tgit config --global --add safe.directory /home/atlantis/.atlantis/repos/REDACTED/2842/default\n"

This is a git problem inside of the container because the uid:gid probably doesn't match what the ownership/octal permissions are on the mounted filesystem. I have a pre-workflow thing that gets around this in my

repos.yaml

that you may find helpful:

# This hook is to work around <https://github.com/runatlantis/atlantis/issues/2221>
    pre_workflow_hooks:
      - run: git config --global --add safe.directory '*'

It probably needs to be baked into the module itself

I almost put up a PR to mess with the image's entrypoint function but I found that my workaround seems to be ok for me and I also didn't think that forcing everyone to pick up my proposed change would be a good idea.

Weird. I wonder why this doesn't happen in GCP.

I guarantee I am not doing that correctly in EFS

Well, I'm about to find out if that solved the issue. 😛

note that switching to the Atlantis non root user may impact the above workaround - https://github.com/runatlantis/atlantis/pull/3886