Use Idle and Pool Configs Simultaneously - Runners Redeploying in a Loop

[dmatviichuk]

Hi,

I am trying to achieve the following scenario using both idle_config and pool_config in the GitHub Runner module:

1. Maintain 1 idle runner at all times.
2. When a runner becomes busy (e.g., due to a developer pushing code), the pool dynamically adds another idle runner because the first one is busy
3. If 2 runners are idle, the lambda should scale down to only 1 idle runner.

However, this setup results in runners being redeployed in a loop. Here’s the observed behavior:
• When a second runner is deployed (to handle a busy runner), and both runners become idle, the scale-down Lambda terminates one of the runners to maintain only 1 idle runner (as per idle_config).
• The pool Lambda immediately detects the shortfall (as the pool_config expects 1 idle runner) and deploys a new runner.
• This cycle continues indefinitely, creating and terminating runners in a loop.

gh_idle_config = [
  {
    cron             = "* 8-19 * * 1-5 *"
    timeZone         = "America/Phoenix"
    idleCount        = 1
    evictionStrategy = "oldest_first"
  }
]

gh_pool_config = [
  {
    schedule_expression          = "cron(0/5 8-19 ? * 1-5 *)"
    schedule_expression_timezone = "America/Phoenix"
    size                         = 1
  }
]