#atlantis (2020-06)
| Discuss the Atlantis (<http://runatlantis.io | runatlantis.io>) |
**Archive: ** https://archive.sweetops.com/atlantis/
2020-06-01
2020-06-17
can i get some upvotes && comments plz https://github.com/runatlantis/atlantis/issues/1084
2020-06-30
Follow-up on conversations about Terraform Cloud vs Atlantis. I’m using Terraform Cloud, but bumping to paid tier is high cost and having difficulty with any new software purchases.
Thinking about atlantics and just setting up with the wonderful cloudposse modules we’ve been provided. I like the “plain preview” and all and PR chatops improvements too.
So…. long-term, do you feel atlantis integration is alive and well and will continue to be a robust solution for those willing to spin up some internal infra, or do you feel it’s on life-support/replacement soon by Terraform Cloud due to the primary developers moving to Terraform team? Getting folks to run Terraform Cloud will be tricky I’m realizing due to their reliance on make files to run terraform commands instead of just “pure” terraform. I’m thinking maybe atlantis is that option that’s needed to allow them to run things but eliminate such heavy reliance on Jenkins for running just terraform plans.
I think it will continue going, TF cloud does not support some specific business rules and workflows that people use
there is active development in the project so far even though they moved to hashicorp
very cool. I’ll probably give it a chance
Having played around with the TFCloud workflow I like atlantis better
TFCloud requires apply from master (I like to apply on the branch to make sure it works before merge (this is atlantis’ workflow))
TFCloud also won’t allow you to plan/apply outside the VCS workflow
@Michael Warkentin that’s not true. TFCloud will plan/apply from any branch
Pick a branch and let it rip.
@johncblandii sorry - yes you can pick a branch to apply when you merge to (could be master or some other branch) but I like atlantis’ workflow better where you actually apply from the feature branch before merging into master (or other branch)
I find the other way around leads to a lot of “merge > apply fails (something misconfigured that you don’t know until runtime) > open a PR to fix > repeat…”
You can apply from a feature too if you just use it as a state backend and tie in chatops through your CI. That’s a local vs remote setting for applies. You miss out on good features there, but it does get around the PR fixes.
You can also use the full API to trigger calls directly.
Remote without VCS is possible too.
@Michael Warkentin you can skip to about 12:00 and you can see me remote plan/apply without VCS locally.