Problem you have encountered:

I have an App Engine project that I periodically disable while I'm not actively using it. I've had this project well over a year. Since September, I am still being billed continuously for a running app engine instance (i.e. instanceId aef-default-<redacted>), even when the project is disabled.

I'm disabling the application using the Disable Application button in the GCP console. I do not receive any errors when disabling (or later re-enabling) this project, which I have done may times since September. It's plausible I received some error message back in September, but no longer have access to those logs.

This "zombie" instance is still actively receiving and responding to health checks while the project is disabled, per the current logs. e.g.:

{
  "insertId": "<redacted>",
  "jsonPayload": {
    "failReason": null,
    "latencySeconds": "0.000",
    "trace": "-",
    "appLatencySeconds": "-"
  },
  "httpRequest": {
    "requestMethod": "GET",
    "requestUrl": "/liveness_check",
    "status": 200,
    "responseSize": "2",
    "userAgent": "GoogleHC/1.0",
    "remoteIp": "35.191.237.227",
    "referer": "-",
    "latency": "0s",
    "protocol": "HTTP/1.1"
  },
  "resource": {
    "type": "gae_app",
    "labels": {
      "project_id": "<redacted>",
      "version_id": "<redacted>",
      "zone": "",
      "module_id": "default"
    }
  },
  "timestamp": "2024-12-07T18:52:01.248Z",
  "severity": "INFO",
  "labels": {
    "appengine.googleapis.com/trace_id": "-",
    "appengine.googleapis.com/instance_name": "<redacted>",
    "compute.googleapis.com/resource_id": "<redacted>",
    "compute.googleapis.com/zone": "us-west1-b",
    "compute.googleapis.com/resource_name": "<redacted>"
  },
  "logName": "projects/<redacted>/logs/appengine.googleapis.com%2Fnginx.health_check",
  "trace": "projects/prj-dev-epic-roadtrip-planner/traces/-",
  "receiveTimestamp": "2024-12-07T18:52:06.618367699Z"
}

What you expected to happen:

This appears to be an app-engine managed instance I expect to be deleted whenever the application is disabled.

Steps to reproduce:

I have been unable to reproduce the original issue that generated this zombie instance, however, it is still running at this moment.

Other information (workarounds you have tried, documentation consulted, etc):

I've searched the GCP docs, Issue Tracker, Stack Overflow, Reddit, the Google Cloud Community, and a number of plausible sounding but otherwise unhelpful AI answers. I've tried enumerating all services, looked for a VM using the compute API, and a number of various other approaches. I have the instanceId, appId, and versionId of the running instance, and have attempted to stop, disable, or delete it using the GCP console, the gcloud CLI, and the REST API. I am unable to find any reference to this instance at all, outside of the logs and metrics.