After manually deletion of a billing alert, no more plan oder apply will succeed, becuase of an 404 during plan

[ckuetbach]

Terraform CLI and Terraform MongoDB Atlas Provider Version

Terraform v0.12.21
Atlasplugin 0.6.4

Terraform Configuration File

resource "mongodbatlas_alert_configuration" "billing_above_max" {
  project_id   =  <project-id>
  event_type = "PENDING_INVOICE_OVER_THRESHOLD"
  enabled    = true

  threshold = {
    threshold = 10
    operator = "GREATER_THAN"
    unit = "RAW"
  }

  notification {
    type_name     = "EMAIL"
    interval_min  = 60
    delay_min     = 0
    email_address  = "ckuetbach@gmail.com"
  }

}

Steps to Reproduce

• Create a billing alert with terraform
• Delete the alert using the website
• Re Create the Resource with terraform

Expected Behavior

• The Alert is recreated

Actual Behavior

• The plan fails, because of an 404 (alert configuration not found)

Additional Infos

• The statefile stored has to be manually changed in order to get the project up and running again.

• Also, if I change values like event_type or the amount at the website, terraform will not recreate the values from my configurations.

[themantissa]

@ckuetbach in repo steps was a refresh performed to refresh the state from the current system? Could you provide debug logs so we can see the specifics?

[ckuetbach]

I don't have the debug logs ready.

But there was a terraform refresh, before I made the changes, because I made an update from 0.5.1 to 0.6.4 to enable billing alert feature.

I'm not sure, how to create a debug log, do you have any documentation ready?

For me it looks, as if the provider won't check the current state correctly against the atlas API. this would explain the 404, while reading the state and also the fact it woldn't update some of the values, if changed (like amount).

I'll try to make a repro in debug mode.

[nikhil-mongo]

debug.log

Here is the debug file for this issue. The Atlas API responses with 404 all the time if any other service is deleted from UI too. But, this is not expected behaviour for billing as @ckuetbach mentioned. If this happens with Atlas cluster, as an exmaple, we create a new cluster rather than showing 404 error. We definitely need to look into this.

Thanks.

[ckuetbach]

@nikhil-mongo thanks a lot. Some changes in made with the UI aren't reflected by terraform. Lets say, you change the alert in the UI from monthly to daily (with amounts), but terraform says, all is up to date.

[nikhil-mongo]

@ckuetbach As per the eventTypeName in the Condition section of Atlas API, the billing event_type is different for all the 3 options. But as we have seen that the alerts are being managed using the alert_configuration_id and we have seen issues managing it. So, we believe that taking care of the alert section will help us configure this as well.

Attaching the debug.log file as the terraform refresh does finds the change in eventTypeName but does not shows it as a change.

debug.log

[themantissa]

@nikhil-mongo can you create an internal bug ticket for this? PRODTRIAGE and I'll move to the appropriate queue.

[themantissa]

Internal ID: INTMDB-112

[themantissa]

Fix was included in 0.7.0, closing issue.