Readiness health checks story 5 processes

[mariash]

This PR is part of a larger effort detailed here in this Cloud Controller PR and here in this CFF RFC.

In this story

• readiness health check parameters returned by:
  • GET /v3/processes
  • GET /v3/apps/:guid/processes
  • GET /v3/processes/:guid
  • GET /v3/apps/:guid/processes/:type
  • POST /v3/apps/:guid/processes/:type/actions/scale
• readiness health check parameters can be updated by:
  • PATCH /v3/processes/:guid
  • PATCH /v3/apps/:guid/processes/:type
• API docs are updated to include readiness health check in examples and object definitions.
• We removed readiness health check params from the Process #export_atrributes and #import_attributes methods since they only affect /v2 summary endpoints and we don't need to change them for /v3 endpoints.

Acceptance Setup

1. Deploy capi-release with this PR.

2. Push an app.

3. See that readiness health check type is set to process:

   "readiness_health_check": {
     "type": "process",
      "data": {
       "invocation_timeout": null
      }
   }

   by checking endpoints:

   cf curl /v3/processes
   cf curl /v3/apps/:guid/processes
   cf curl /v3/processes/:guid
   cf curl /v3/apps/:guid/processes/:type
   cf curl -X POST /v3/apps/:guid/processes/:type/actions/scale
   

4. Update readiness health check using process type endpoint:

   cf curl -X PATCH -d '{"readiness_health_check": {"type": "http", "data": {"endpoint": "/ready"}}' /v3/apps/:app-guid/processes/web
   

5. See that readiness health check type is set to http:

   "readiness_health_check": {
     "type": "http",
      "data": {
      "endpoint":  "/ready",
       "invocation_timeout": null
      }
   }

   by checking endpoints:

   cf curl /v3/processes
   cf curl /v3/apps/:guid/processes
   cf curl /v3/processes/:guid
   cf curl /v3/apps/:guid/processes/:type
   cf curl -X POST /v3/apps/:guid/processes/:type/actions/scale
   

6. Update readiness health check using process endpoint:

   cf curl /v3/processes/710eeb42-cb61-4761-bda0-a28e7a0c0d07 -X PATCH -d '{"readiness_health_check": {"type": "port"}}'
   

7. See that readiness health check type is set to port (notice that endpoint is not displayed:

   "readiness_health_check": {
     "type": "port",
      "data": {
      "endpoint":  "/ready",
       "invocation_timeout": null
      }
   }

   by checking endpoints:

   cf curl /v3/processes
   cf curl /v3/apps/:guid/processes
   cf curl /v3/processes/:guid
   cf curl /v3/apps/:guid/processes/:type
   cf curl -X POST /v3/apps/:guid/processes/:type/actions/scale
   

[sethboyles]

why do readiness health checks not have timeouts like the original health check?

[mariash]

@sethboyles health-check-timeout represents start up timeout. When app starts up we run liveness health check for this period to mark it as Running. If it never succeeds for the first time during that timeout period we mark app as crashed and we don't event restart it since it was never running. Readiness health checks don't have that behavior. They start running after app instance is started and marked as Running.