This message was deleted.
# atlantis-community
s
This message was deleted.
p
there is an approach here to use the new api for drift detection https://github.com/cresta/atlantis-drift-detection/ it is usable but has a few issues
we tried recruiting people to create a built in drift detection in Atlantis but it never got finished
s
Interesting…might try this out
p
Atlantis uses PRs to runs plans so if it can’t run plan it will not detect diferences
w
Not specifically I’ve played with
driftctl
a little, and we built an in-house drift detection tool at a previous job (which I’m sad never got open sourced). I do find that, since we pin provider updates using Renovate, and since atlantis lets us plan the affected states fairly easily, we (at my current place) do essentially use atlantis as a low-tech drift detector; just one that only runs when provider updates happen. in our case, the environments change infrequently enough, users mostly don’t have permissions to make a lot of out of band changes, and the provider updates are frequent enough, that not having better drift detection hasn’t been a huge issue yet
Driftctl (Snyk) has some things I like and some things I don’t like. it doesn’t support all the resource types we use, and it’s not great at understanding when resources are part of managed services (you have to ignore them explicitly) but, it’s nice to be able to get a rough idea of how many things are not managed by tf vs. just detecting state drift with known objects / resources