Support named, ordered services

[skpy]

I'd like to use loadgen to build Honeycomb graphs for training purposes. I'd like to construct scenarios where I know what symptom caused what visible display, so that I can then talk trainees through the logic of digging into and through spans.

Currently, loadgen creates a handful of services but there's no way for users to define the names, or set an order of nesting.

Describe the solution you'd like
I'd like to see either a YAML or JSON file with a simple ordering of services. Something like this YAML:

- toplevel1
  - sub1
    - redis1
  - sub2
    - pgsql2
- toplevel2
  - sub3
    - mssql3
  - sub4
  - sub5
- toplevel3
  - sub1
    - redis1
  - sub6

(I suppose a gnarly CLI option could be supported, too, but I suspect that's a lot more work for very little benefit?)

I would then like to be able to define some variable results for the sub-spans. For my specific use case, I want to generate a bunch of good data over a period of time, and then generate a bunch of data where one of the sub-systems fails completely. This failure may or may not be reported as an error on the higher-up services. In this way, we can simulate "Application sub3 had a bad deploy / fell over / lost its database" and see real-world graphing results of that.

[MikeGoldsmith]

Hi @skpy - sorry for the slow reply.

This is an interesting feature request that we will discuss and add to our backlog. We're unsure when we may consider it for development.

However, we're always open to PRs, so if you feel like working on it and want to submit a PR, we'd be willing to review and give feedback.

Thanks