bug: Group settings unexpectedly overwrite check-level retry and alert configurations

[ehtung]

Terraform version

1.13.0

Provider version

1.15.0

Steps to reproduce

The group

resource "checkly_check_group" "my_group" {
  name        = "My Group"
  activated   = true
  concurrency = 5
  locations   = local.my_locations
}

Plan:

  # checkly_check_group.my_group will be created
+ resource "checkly_check_group" "my_group" {
      + activated    = true
      + concurrency  = 5
      + double_check = false
      + id           = (known after apply)
      + locations    = [
          + "ap-east-1",
          + ... others
          + "ap-southeast-3",
        ]
      + name         = "My Group"

      + alert_settings (known after apply)

      + api_check_defaults (known after apply)

      + retry_strategy {
          + type = "NO_RETRIES"
        }
    }

The checks

resource "checkly_check" "my_checks" {
  for_each = local.my_checks # There are 2 checks

  name      = each.value.name
  type      = each.value.type
  activated = each.value.activated
  frequency = each.value.frequency

  group_id  = checkly_check_group.my_group.id
  locations = local.my_locations

  retry_strategy {
    type        = "FIXED"
    max_retries = 2
    same_region = false
  }

  request {
    url              = each.value.request.url
    follow_redirects = true
    skip_ssl         = false

    # 2xx assertion blocks
  }
}

Plan (of 1 check, for simplicity):

  # checkly_check.my_checks["my-check-1"] will be created
+ resource "checkly_check" "my_checks" {
      + activated              = true
      + degraded_response_time = 15000
      + double_check           = false
      + frequency              = 1
      + group_id               = (known after apply)
      + id                     = (known after apply)
      + locations              = [
          + "ap-east-1",
          + ... others
          + "ap-southeast-3",
        ]
      + max_response_time      = 30000
      + name                   = "My Service 1"
      + type                   = "API"

      + alert_settings (known after apply)

      + request {
          + body_type        = "NONE"
          + follow_redirects = true
          + headers          = (known after apply)
          + ip_family        = "IPv4"
          + method           = "GET"
          + query_parameters = (known after apply)
          + skip_ssl         = false
          + url              = "https://my-service-1.com/healthz"
            # (1 unchanged attribute hidden)

          + assertion {
              + comparison = "GREATER_THAN"
              + source     = "STATUS_CODE"
              + target     = "199"
                # (1 unchanged attribute hidden)
            }
          + assertion {
              + comparison = "LESS_THAN"
              + source     = "STATUS_CODE"
              + target     = "300"
                # (1 unchanged attribute hidden)
            }

          + basic_auth (known after apply)
        }

      + retry_strategy {
          + base_backoff_seconds = 60
          + max_duration_seconds = 600
          + max_retries          = 2
          + same_region          = false
          + type                 = "FIXED"
        }
    }

What is expected?

• Group created, with any values of its own retry and alert settings, except for overriding
• Checks created, with their own retry and alert settings, not being overridden by group's

What is actually happening?

• Group: Has “Group override for retries” and “Group override for alert settings” checked unexpectedly in GUI

  • 

• Checks: Retry strategies and alert settings grayed out due to being overridden by group

  • 

Workaround (failed)

• Manually edit group (GUI): Uncheck “Group override for retries” and “Group override for alert settings” 🔲
  • Terraform plan output:
    • to recreate the group,
    • checks no longer belong to the group, and 1 check with group_order changed (- group_order = 1 -> null)
• Manually edit group (GUI): Check “Group override for retries” and “Group override for alert settings” ☑️
  • Terraform no longer plans to recreate the group. There were still other drifts, but they are "side-effects" and not very important in this bug report.

System Info

Any additional comments?

No response

[sorccu]

I'm sorry but this is not possible right now. The provider has not been updated to use the v2 endpoints of check groups, which is required for this functionality. In the UI, you are always dealing with v2 groups which is why you have the functionality available, and our CLI supports the CheckGroupV2 construct which is also capable of equivalent functionality. There have been some complications with the Terraform provider that have delayed the implementation. Our intention is of course to update the provider as soon as possible, but there are currently some other items that are taking up my time. As a rough estimate, I am hoping to release by the end of next week, but cannot promise for certain right now.

[ehtung]

Thank you, @sorccu, for the prompt answer and the insights. I appreciate that you provided a priority and estimation, and I completely understand if the fix takes longer. Hopefully, a provider update will come eventually.

In the meantime, I will investigate the CLI approach.

Best regards.