Expose terraform CLI logs as part of the Workspace Status

[ytsarev]

What problem are you facing?

Currently, it is hard to debug the terraform issues as the provider-terraform-controller exposes only brief error as the k8s event.

In cases when we need a full terraform CLI output to get more context of the error we need to kubectl exec -it into pod, copy the /tf/${workspace-uuid} directory and reproduce the error in isolation.

It takes time and is generally brittle and inconvenient.

How could Crossplane help solve your problem?

Possible solution:

• Extend Workspace CRD Status with additional Log field
• Expose logs there, maybe in base64 form for brevity( IIUC https://github.com/rancher/terraform-controller does a similar thing)

[bobh66]

@ytsarev - what would you think about using Events for this type of information? I'm seeing some updates to workspaces that I don't expect and I was thinking that it would be useful to have the "diff" output from terraform plan available as an Event on the Workspace object so we can see why it thinks it needs to be updated. I don't know if there are size limitations to the Event text that would cause issues.

[ytsarev]

@bobh66 yes, Events would be ideal, the question is how and what to capture into the Event. Currently it is concise error output, putting the whole log is too much, maybe we can find a way to capture the enough log context around the error - that would be ideal

[ytsarev]

Looks like we should attack #24 first and if we catch enough error context there, the full logs will be redundant.