🚧 Frequent GitHub Rate-Limit Blocks — Any Tips?

[thisispr]

Hey everyone,

I’m running into a recurring issue with GitHub API rate limits. About once or twice a day, I get blocked for 15–20 minutes, which interrupts my workflow quite a bit.

Has anyone dealt with this before? I’d love to hear any practical tips or workarounds to avoid hitting the limit so quickly.

Some context:

• It usually happens when I’m doing multiple fetches/requests in a short timeframe (e.g., CI runs).
• I’m using authenticated requests, but I’m still hitting the cap.
• Once the block hits, I generally need to wait it out.

# Check if users exist before creating memberships
# Use try() to handle 404 errors gracefully
# tflint-ignore: terraform_unused_declarations
data "github_user" "check_users" {
  depends_on = [module.entra_id_group]

  for_each = toset(local.existing_github_users)
  username = each.key
}

locals {
  gitflow_bypass_actors = {
    for repo_name, _ in local.all_repos : repo_name => {
      prtarget = local.gitflow_prtarget_default_bypass_actors
      main     = local.gitflow_main_default_bypass_actors
    }
  }
}

Questions:

• Are there settings or tools to help manage or reduce the number of API requests?
• Would switching to a Fine-Grained Personal Access Token or increasing limits with GitHub Enterprise help?

Any ideas, experiences, or suggestions would be greatly appreciated!

Thanks 🙏

[deiga]

One option, especially for CI is to use GitHub App authentication, that has 15k/h limit.
There are some caveats, that it doesn't work for all resources due to API limitations.

In general, I could recommend splitting state into smaller pieces and leveraging remote_state to access data between them.

We've separated resources that need to change rarely into multiple states, thus reducing the rate cost of execution drastically.
For example: repo autolink config changes once in a blue moon, so we can run that "environment" only once a night