Atlantis produce Locks based on unrelated Terraform conditions

[fblgit]

Community Note

• Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request. Searching for pre-existing feature requests helps us consolidate datapoints for identical requirements into a single place, thank you!
• Please do not leave "+1" or other comments that do not add relevant new information or questions, they generate extra noise for issue followers and do not help prioritize the request.
• If you are interested in working on this issue or have submitted a pull request, please leave a comment.

---

Overview of the Issue

Atlantis produces locks based on values that are unrelated to terraform states such as repo/pr/workspace among other factors. That ultimately translates into multiple scenarios such as Atlantis AutoPlan discarding valid plans, caused by a rebase.

If we look at this from a common implementation of monorepo, which is used by other teams to self-service infrastructure resources, ultimately there are several factors taking place which difficult a good consumer experience out of it. Some of the situations faced at a larger scale:

• rebase racing --> I raise a PR, Atlantis plan it, due to the number of resources let's say it takes 7m to plan. By the time I finished the plan, someone else already merged another PR, most of the time, unrelated to the project, workspace, resources, state, or even account but that in any case due to the setup it would require a rebase, per se, a new plan as an outcome.
• concurrent plans with no workspace --> Even if we separate the project, states, etc.. meanwhile this takes place on the same repository and with default workspace, they will not run in parallel. This is mostly caused by the incapacity of Atlantis to identify in a reality where to stale a previous plan or not.

Reproduction Steps

Simple, make a repo, set up Atlantis, make 2 PR, plan both of them (separate projects, so they don't get a chance to get blocked). Merge one and if you require (which is definitively positive) to have a consistently up-to-date diverged branch as a merge requirement, you will have to rebase and that will trigger a new auto plan, discard the previous, etc.

The main problem comes when this is repeated dozens of times per day, requiring a long time and multiple plans to successfully merge.

Logs

No logs are needed, as this details the existing behavior or Atlantis.

Environment details

Latest Atlantis 0.17.3

• AutoPlan
• AutoMerge
• RequireUpToDateBranchToMerge

[fblgit]

I'm keen to work on a PR to solve this, but I would like to know if there are plans in scope related to this?

[jamengual]

no plans in scope

[jamengual]

is this still happening with v0.19.8?

[nitrocode]

I believe this is still an issue so i reopened for now. @fblgit are you still experiencing this?