Improve handling of SLI queries that don't produce a single value

[grabnerandi]

Currently when a single metric query fails to deliver any data the dynatrace-service returns the message: "Metrics query result has no data". This is correct - but - doesnt tell the user anything which metric query resulted in no data

This is the only message that will be shown in the bridge indicating why the evaluation failed. to troubleshoot the issue the user needs to dig deep into the payload of the get-sli.finished event.

To fix this I suggest to simply list the names of the SLIs that failed as part of the error message. Here is an example payload with the new message:

{
  "data": {
    "get-sli": {
      "indicatorValues": [
        {
          "metric": "Performance_SLO",
          "success": true,
          "value": 99.48424304709904
        },
        {
          "message": "Metrics query result has no data",
          "metric": "problems",
          "success": false,
          "value": 0
        },
        {
          "metric": "response_time_p90",
          "success": true,
          "value": 1238.0154590423788
        },
        {
          "metric": "response_time_p75",
          "success": true,
          "value": 639.2749136467561
        },
        {
          "message": "Metrics query result has no data",
          "metric": "request_error_rate",
          "success": false,
          "value": 0
        }
      ],
    },
    "message": "Metrics query for request_error_rate, problems resulted in no data",
    "project": "dynatrace",
    "result": "fail",
    "service": "myservice",
    "stage": "quality-gate",
    "status": "succeeded"
  },
  "source": "dynatrace-service",
  "specversion": "1.0",
  "type": "sh.keptn.event.get-sli.finished"
}

[arthurpitman]

@grabnerandi I guess the simplest fix would be to include the metric name in the indicator's message field which would at least propagate up to the message of the event, e.g. for this example it would become "Metrics query result for problems has no data, Metrics query result for request_error_rate has no data".

[arthurpitman]

Due to the effect this issue will have on the code paths and tests, I'm broadening the scope to include the following functionality:

• Set the result field of the sh.keptn.event.get-sli.finished to warning if and only if the individual indicators have success set to false only for not returning a single value. This corresponds to extending the code to implement "case B" described in Improve the handling of `sh.keptn.event.get-sli.triggered` events where SLIs don't produce a value #704.
• Harmonize the message fields of individual elements in indicatorValues to make them consistent for the dashboard and sli.yaml use cases.
• Group the indicator messages to form a summarized message in the get-sli event data.

[grabnerandi]

lgtm