[datadog_service_level_objective] Add query_interval_seconds to time-slice SLO condition parameters

[ddjamesfrullo]

https://datadoghq.atlassian.net/browse/SLO-3189

Overview

This adds the query_interval_seconds parameter to the time-slice SLO condition, used when creating and modifying SLOs. This parameter controls the size of a time-slice, and hence the granularity of the uptime measurement. We only want to allow two values for - 60 (1 minute) and 300 (5 minutes). This is done via validation. The corresponding API change was done here.

Motivation

Allow customers to create time-slice SLOs using 1-minute granularity by specifying query_interval_seconds as 60.

Example Terraform

resource "datadog_service_level_objective" "time_slice_slo_1" {
  name              = "Terraform test for Time Slice"
  type              = "time_slice"
  description       = "Created as part of a manual test of Terrform provider changes to support time-slice SLO"
  sli_specification {
    time_slice {
      query {
        formula {
          formula_expression = "(query1-query2)/query1"
        }
        query {
          metric_query {
            name        = "query1"
            data_source = "metrics"
            query       = "sum:trace.grpc.server.hits{service:monitor-history-reader}"
          }
        }
        query {
          metric_query {
            name        = "query2"
            query       = "sum:trace.grpc.server.errors{service:monitor-history-reader}"
          }
        }
      }
      comparator             = ">"
      threshold              = 0.99
      query_interval_seconds = 60
    }
  }

  thresholds {
    timeframe = "7d"
    target    = 99.0
    warning   = 99.5
  }

  timeframe         = "7d"
  target_threshold  = 99.0
  warning_threshold = 99.5

  tags              = ["test:true"]
}

Testing

I tested the following:

1. Applying the above resource to create a time-slice SLO with 1-minute granularity
2. Applying a change from 60 -> 300 for the query interval, modifying the existing SLO
3. Applying a change back to 60, again modifying the SLO

The resulting SLO is here.

I also tested not specifying query_interval_seconds and tested that a 5-minute-granularity time-slice SLO was created. That SLO is here.

Finally, I tested specifying an invalid value for query_interval_seconds, and I get this error message:

╷
│ Error: Invalid enum value
│ 
│   with datadog_service_level_objective.invalid_query_interval,
│   on slo.tf line 111, in resource "datadog_service_level_objective" "invalid_query_interval":
│  111:       query_interval_seconds = 120
│ 
│ `invalid value '120' for SLOTimeSliceInterval: valid values are [60 300]`
╵

[zhengshizhao]

nice!