This message was deleted.

repos.yaml passing json?

oh duh, yeah that wouldn’t even work, how would i actually set up a repo config then?

just pass the config to it as a whole

https://www.runatlantis.io/docs/server-configuration.html#repo-config oh wait i found it!

yes, that is to pass the file that should be in the container

yeah what i had converted the yaml file to json and then added the config there, but i’m not sure if it was working

ok

oh wait, i’m not sure that will work because the repos.yaml file was never written to disk and that other one is just looking for a path.

so then just pass the json content to the previous flag

ok, so i still get the plan error above, how do i know if it’s actually getting that config?

logs in the container

so it looks like it’s getting it but error out when looking for a plan. my repo looks like this which i think is pretty standard for a cdk repo. it says “plan requires configuration to be present” it doesn’t look like it’s actually able to find the config file and doesn’t look like it even tries to run the pre-workflow hooks here is my repo style for reference

- lib (for custom library stuff)
- scripts
- stacks
  - stack-name
    - main.py
    - cdktf.json
    - cdktf.out
      - stacks/atlantis-ops-us-east-1
        - .terraform
        - .terraform.lock.hcl
        - cdk.tf.json
      - manifest.json
  - stack-name
    - main.py
    - cdktf.json
    - cdktf.out
      - stacks/atlantis-ops-us-east-1
        - .terraform
        - .terraform.lock.hcl
        - cdk.tf.json
      - manifest.json

@RB have run CDK

anybody here have experience working with cdktf? the debug logs don’t seem to show me much and no matter what i get an error that looks like this. logs definitely make it seem like it’s getting the pre-workflow hooks, but that’s only because when it wasn’t, it was giving me an error.

{
  "level": "error",
  "ts": "2024-02-09T19:17:36.977Z",
  "caller": "models/shell_command_runner.go:158",
  "msg": "running \"/root/.tfenv/bin/terraform plan -input=false -refresh -out \\\"/root/.atlantis/repos/rakuten-shopstyle/shopstyle-cdk/1184/default/stacks/atlantis/default.tfplan\\\"\" in \"/root/.atlantis/repos/rakuten-shopstyle/shopstyle-cdk/1184/default/stacks/atlantis\": exit status 1",
  "json": {
    "repo": "rakuten-shopstyle/shopstyle-cdk",
    "pull": "1184",
    "duration": 0.048867454
  },
  "stacktrace": "<http://github.com/runatlantis/atlantis/server/core/runtime/models.(*ShellCommandRunner).RunCommandAsync.func1|github.com/runatlantis/atlantis/server/core/runtime/models.(*ShellCommandRunner).RunCommandAsync.func1>\n\t/home/runner/work/atlantis/atlantis/server/core/runtime/models/shell_command_runner.go:158"
}

I ran it a while ago and it was successful with a custom workflow.

was the custom workflow anything more than just this?

repos:
  - id: /.*cdktf.*/
    pre_workflow_hooks:
      - run: make init && make install && npm i && cdktf synth --output ci-cdktf.out
      #- run: npm i && cdktf get && cdktf synth --output ci-cdktf.out

Could you try simplifying your pre workflow to just

echo "hello world"

or similarly trivial to see if you can avoid the error

yeah i’ll try that, and yep have both installed

- id: /.*cdktf.*/

is for the repo name not the content of the repo

ohhhhh, alright I was understanding it as looking for a folder that contained a file that contained cdktf in it

at the root of the checkout of the PR

hmmm okay, that’s not ideal, is there a way to run it in the directory that you specify? or a directory with changes to them that also has a file containing cdktf.json or something?

you need to have a atlantis.yaml and a custom workflow

we’re only using it for one repo right now, but okay that makes sense, that gets rid of the autodiscovery though right?

not necessarily

hmmm okay, i’ll give it a try

you can run a post_workflow hook to autogenerate the atlantis.yaml file

oh so we would need to write a script for that?

look at the docs