We are running atlantis as an ECS task and after upgrading f

Question

We are running atlantis as an ECS task and after upgrading from `0 22 3` to `0 24 0` we started seeing sporadic errors like the one below ```dir tmp dev my service infra does not exist``` files in this directory are rendered during `pre workflow hook` step and atlantis yaml looks like ```version 3 projects dir tmp dev my service infra ``` Error usually happens when PR has 20+ plans and works with under 10 plans We ll check version by version but maybe anyone has ideas what might be causing it pray

Ed · Answer

My initial thought was that it might be ephemeral storage space related and has something to do with <https github com runatlantis atlantis pull 2758 files|checkout depth> like it was filling the disk quickly or something But after setting depth to 1 plans stopped rendering altogether using merge checkout strategy

RB · Answer

Sorry to hear this There have been quite a few changes and we re pushing for more frequent version releases to bubble these issues up sooner Have you tried slowly bumping the versions from 0 22 3 to see where the issue may have been introduced

RB · Answer

Also we now push a new tag on commit so its possible to go up and down by commit for more granularity

RB · Answer

One last thing after about 5 or 6 weeks 0 24 0 was released and a panic was fixed rendering 0 24 0 a pre release and 0 24 1 the latest release as of today

Dylan Page · Answer

< Ed> This sounds like some of the race conditions with locks with pre workflow hooks and the filesystem locks when we clone the repo

Dylan Page · Answer

See <https github com runatlantis atlantis pull 3345>

Ed · Answer

thanks we ll proceed updating version by version