Container agent (container runner) setup

Build Environment
1

I followed container-runner docs to deploy (via Helm) a container agent. I created a resource class and noted the token in the WebUI. I created a values.yaml with the following content (scrubbed).

# values.yaml
---
resourceClasses:
  namespace/my-rc:
    token: MY_TOKEN

Helm deploy shows success:

helm install -f values.yaml container-agent circleci/container-agent -n circleci

However, pod logs show WARN: Missing API Key.. I’ve exec’d into the container and checked out the contents for /etc/container-agent/config.yaml and it shows the following. I don’t see anything representing my resource class and token.

# /etc/container-agent/config.yaml
resourceClasses:

Digging a bit more I think the deployment’s init container is doing some work to prep a config.yaml via the taskpod-config volume. It’s merging the config map container-agent with the secret container-agent. Values for both of these in my setup are equal to resourceClasses:. resourceClasses: is written to config.yaml and copied to the volume taskpod-config which is then mounted to /etc/container-agent/config.yaml in the container-agent container.

In an attempt to just make it work I updated config.yaml with the contents of my values.yaml and while exec’d into the pod ran kill 1; ./container-agent. The container-agent registered and can now successfully run jobs.

2

Looks like values.yaml was off by a property in the docs. Adding agent gets the init container to properly merge the yaml. Not a Helm expert here :rofl:, open to suggestions and explanations. Thanks!

# values.yaml
---
agent:
  resourceClasses:
    namespace/my-rc:
      token: MY_TOKEN
3

Glad you got it figured out! I’m updating the docs to include agent: as the top-line key in that section to avoid confusion going forward. Sorry about that.

Let me know if you have any questions or concerns with the container runner!

4

This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.