Getting rate limit from Split API

[bulju]

Hi there,

Terraform Version

Run terraform -v to show the version. If you are not running the latest version of Terraform, please upgrade because your issue may have already been fixed.

HerokuX Provider Version

Run terraform -v to show core and any provider versions. A sample output could be:

Terraform v1.3.2
+ provider.split v0.6.2

Affected Resource(s)

Please list the resources as a list, for example:

• split_environment

• split_user

• we've tested on these 2 kind of resources, but we suspect that the same issue is affecting all resources

If this issue appears to affect multiple resources, it may be an issue with Terraform's core, so please mention this.

Terraform Configuration Files

resource "split_environment" "this" {
  for_each           = var.environments
  name                = each.value.name
  production       = each.value.production
  workspace_id  = each.value.workspace
}
environments = {
  1 = {
    name       = "Production"
    production = "false"
    workspace  = "REDACTED"
  },

  2 = {
    name       = "QA"
    production = "false"
    workspace  = "REDACTED"
  },

  3 = {
    name       = "Staging"
    production = "false"
    workspace  = "REDACTED"
  },

  4 = {
    name       = "Test"
    production = "false"
    workspace  = "REDACTED"
  },

  5 = {
    name       = "staging-mock"
    production = "false"
    workspace  = "REDACTED"
  },

  6 = {
    name       = "staging"
    production = "false"
    workspace  = "REDACTED"
  },

  7 = {
    name       = "Prod"
    production = "true"
    workspace  = "REDACTED"
  },

  8 = {
    name       = "Staging"
    production = "false"
    workspace  = "REDACTED"
  },

  9 = {
    name       = "Builds"
    production = "true"
    workspace  = "REDACTED"
  },

  10 = {
    name       = "Builds"
    production = "false"
    workspace  = "REDACTED"
  },

  11 = {
    name       = "Prod- Auto"
    production = "true"
    workspace  = "REDACTED"
  },

  12 = {
    name       = "Stg- Auto"
    production = "false"
    workspace  = "REDACTED"
  },

  13 = {
    name       = "Prod"
    production = "true"
    workspace  = "REDACTED"
  },

  14 = {
    name       = "Staging"
    production = "false"
    workspace  = "REDACTED"
  },

  15 = {
    name       = "Prod"
    production = "true"
    workspace  = "REDACTED"
  },

  16 = {
    name       = "Stg"
    production = "false"
    workspace  = "REDACTED"
  },

  17 = {
    name       = "Prod"
    production = "true"
    workspace  = "REDACTED"
  },

  18 = {
    name       = "Stg"
    production = "false"
    workspace  = "REDACTED"
  },

  19 = {
    name       = "appstore"
    production = "true"
    workspace  = "REDACTED"
  },

  20 = {
    name       = "dev"
    production = "false"
    workspace  = "REDACTED"
  },

  21 = {
    name       = "staging"
    production = "false"
    workspace  = "REDACTED"
  },
}

Debug Output

Please provider a link to a GitHub Gist containing the complete debug output: https://www.terraform.io/docs/internals/debugging.html. Please do NOT paste the debug output in the issue; just paste a link to the Gist. Please MAKE SURE to mask any sensitive values.

TBD

Panic Output

If Terraform produced a panic, please provide a link to a GitHub Gist containing the output of the crash.log.

Expected Behavior

What should have happened?

Split resources should be created

Actual Behavior

What actually happened?

We got a rate limit and we received a 429 from Split

Steps to Reproduce

Please list the steps required to reproduce the issue, for example:

Terraform apply with more than 20 resources (this number is not precise yet, as the provider doesn't provide actual numbers)

[bulju]

I'll work on this and raise a PR @davidji99

[davidji99]

Hmm, this is interesting. 20 resources at the default parallelization of 10 for terraform seems quite low for a rate limit.

I'm curious, what happens if you set the -parallelism flag to say 5 (terraform apply -parallelism 5)?

In any event, let me raise this issue to my Split contacts.

[bulju]

I did try adding -parallelism=1 and the issue persists.

Based on their documentation, when you receive a 429 HTTP code (which is the rate limit), they return in their headers how much time you should wait to get a new quota. I guess we could add that to the client for doing a retry logic.

Here there is another provider which had the same issue: https://github.com/okta/terraform-provider-okta/issues/186

Let me know what your contacts say about this issue.

I'm planning to work on this by the end of this week, to have a PR and then we can decide if we merge or not :)

[bulju]

Hi @davidji99, I just created a PR -> https://github.com/davidji99/terraform-provider-split/pull/111 Let me know if you like the philosophy that I took and if you believe that something is missing, I can rewrite it if it's needed! As I don't have a sandbox environment I wasn't able to test it through the API with all the functionalities, I just ran a terraform plan for environments and now it's working. But this only covers the "get" use case, not post, patch, etc.

[jramosf]

The HTTP GET use case should be good enough since the rate limiting happens when the statefile is being refreshed in the plan phase.

Only in the case of updating/recreating a very large Split environment with the provider we may hit the limit if the request happen in parallel.