Convert interval to be an array of intervals

[kaspernissen]

If you want to perform the same query but aggregate at different intervals, it would be handy to provide a string array of intervals instead of just a string right now.

Example now:

queries:
- query: res.statusCode>=200 | count(field=context.userId, distinct=true)
  repo: services
  interval: 24h
  metric_name: active_app_users
- query: res.statusCode>=200 | count(field=context.userId, distinct=true)
  repo: services
  interval: 1h
  metric_name: active_app_users
- query: res.statusCode>=200 | count(field=context.userId, distinct=true)
  repo: services
  interval: 5m
  metric_name: active_app_users
- query:  res.statusCode>=200 | count(field=context.userId, distinct=true)
  repo: services
  interval: 1m
  metric_name: active_app_users

Would produce the same metrics output:

queries:
- query: res.statusCode>=200 | count(field=context.userId, distinct=true)
  repo: services
  interval: [1m, 5m, 1h, 24h]
  metric_name: active_app_users

[Crevil]

Makes good sense. Maybe we should just make it the default? So removing the interval: 1m option all together?

[kaspernissen]

Make it the default that we always output 4 different intervals? We don't always want that many intervals. I think it makes good sense to be able to specify the interval as an array, and interval is required unless we want to fall back to a default interval.

[Crevil]

No, I mean make it the default for interval to be an array. :)

So if you only want one value, you specify it like:

queries:
- query: res.statusCode>=200 | count(field=context.userId, distinct=true)
  repo: services
  interval: [1m]
  metric_name: active_app_users

The reasoning to do so, is to avoid explaining and maintaining functionality that it can either be a string or an array of strings.