newrelic_alert_condition - condition_scope must be application or instance

[mattcanty]

Terraform Version

0.9.1

Affected Resource(s)

Please list the resources as a list, for example:

• newrelic_alert_condition

Terraform Configuration Files

data "newrelic_application" "app" {
  name = "My.Web.Api"
}

resource "newrelic_alert_policy" "foo" {
  name = "foo"
}

resource "newrelic_alert_condition" "foo" {
  policy_id = "${newrelic_alert_policy.foo.id}"

  name        = "foo"
  type        = "apm_app_metric"
  entities    = ["${data.newrelic_application.app.id}"]
  metric      = "apdex"
  runbook_url = "https://www.example.com"

  term {
    duration      = 5
    operator      = "below"
    priority      = "critical"
    threshold     = "0.75"
    time_function = "all"
  }
}

Debug Output

https://gist.github.com/matthewcanty/acbdc5b8247376b8b6803ea733b607d9

Expected Behavior

An alert policy should have been created with conditions on My.Web.Api.

Actual Behavior

An alert policy is created, including notification channel but no condition is made.

Steps to Reproduce

terraform apply

Important Factoids

My configuration is lifted from https://www.terraform.io/docs/providers/newrelic/d/application.html only I've replaced the app name with one which is valid.

[grubernaut]

Hello @matthewcanty, thanks for the issue!
It looks like the newrelic API either changed, or there is an upstream difference between the go-sdk we use and the newrelic API. I searched the New Relic documentation thoroughly, and I couldn't find any hints to what condition_scope is, unfortunately. Paging @paultyng for any thoughts.

[mattcanty]

https://docs.newrelic.com/docs/alerts/new-relic-alerts/configuring-alert-policies/scope-alert-thresholds-specific-instances

Only reference to the property I can find is here.

[paultyng]

I'll try to take a look later today, their admin API is pretty lacking in documentation, looks like maybe we can just at least default it to application for now maybe?

[grubernaut]

@paultyng that'd be okay with me, if we're sure that instance isn't used for anything. Want to make that change in the SDK, and we'll update the vendor?

[paultyng]

Weirdly that field does show up in their API explorer, with zero documentation around it.

Let me contact their support and ask whats up.

[grubernaut]

@paultyng awesome, that'd be best ❤️ thanks!

[paultyng]

Per New Relic support:

After doing a bit of investigation, it looks like your account is enrolled in a few of our new features, specifically "JVM Instance-based alerting".
This is the reason why this account is expecting the condition_scope value. Because of having the feature enabled to select JVM instance or application, you will need to pass in one additional key:value pair to the condition payload to specify whether you would like violations scoped to the application or an individual JVM. This means you need to add the condition_scope to the condition object and pass it the value of either application or instance.

So I'll get this added to the SDK library as an omitempty and we can update the vendor and fix it.

[paultyng]

Same link as above, but at least an explanation as to why its only sometimes an issue.

[grubernaut]

Ah that's awesome, thanks for digging into that!

[paultyng]

PR submitted on #13158 that works with this field now.

[ghost]

I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues.

If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.