Merge strategy when multiple servers are used in the same repository

[SamuelMolling]

Hi folks,

I’m running two Atlantis servers against the same repository. Each one is configured to watch a different folder (so they don’t overlap in scope).

The issue is: when I open a PR that touches files from both folders, both servers run. If I trigger atlantis apply, Atlantis will merge the PR as soon as the first server finishes successfully — even if the other one fails.

This is a problem because the PR gets merged while half of the changes are broken.

Has anyone faced this before? Is there a way to enforce that all Atlantis servers must pass before the PR is merged? Or any best practices/workarounds to handle multi-server setups like this?
Thanks in advance!

[dosubot]

Hi @SamuelMolling! I'm Dosu and I’m helping the atlantis team.

Atlantis doesn't have a built-in way to coordinate multiple servers or enforce that all must pass before merging a PR. Each server acts independently and will trigger a merge as soon as its own apply succeeds, without awareness of the other server's status. This means a PR can be merged even if one server fails, as you've observed.

To prevent premature merges, you'll need to use external orchestration—typically by configuring your VCS (like GitHub) branch protection rules to require all relevant Atlantis status checks to pass before allowing a merge. This way, the PR won't be merged until both servers have reported success. Atlantis itself doesn't aggregate or synchronize results across multiple servers, so external status checks or consolidating to a single Atlantis server (if possible) are the main workarounds. More details on how Atlantis determines mergeability and status checks can be found in the codebase here.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}