KongIngress Upstream unhealthy.interval value not reflecting

Questions
1

Hello Everyone,

It is regarding Kong Ingress controller and upstream configuration for healthchecks- healthy/unhealthy stats.
During initial testing it is observed that upstream field unhealthy-interval is always 0 (i.e. healthchecks.active.unhealthy.interval ref: https://docs.konghq.com/1.3.x/admin-api/#create-upstream) when configured via declarative approach (i.e. Kong IC).
It is observed that similar fixes are made at fix(upstream) add missing healthchecks.active.unhealthy.interval field by hbagdi · Pull Request #7 · hbagdi/go-kong · GitHub

As a work around we can follow below approach/steps:

  1. create upstreams via declarative approach (Kong IC)
  2. update healthchecks.active.unhealthy.interval field (integer >=1 ) via kong-admin-api

Kindly suggest whether this is a right approach till we get a fix. Also help to understand whether this change is released in Kong-1.3 or planned to be released in future?

# FYI K8s deployment version:
  Kong: 1.2
  Kong Ingress Controller: 0.50

## Ref upstream fields form KongIC; Actual upstream configurations via kong-Ingress resource 
  upstream:
    hash_fallback: none
    hash_on: none
    healthchecks:
      active:
        concurrency: 10
        healthy:
          http_statuses:
          - 200
          interval: 5
          successes: 5
        http_path: /health
        timeout: 1
        unhealthy:
          http_failures: 5
          http_statuses:
          - 500
          interval: 2
          tcp_failures: 5
          timeouts: 5
### Ref upstream fields form Kong admin API;  details of '<<admin-api>>/upstream' configs via kong-manager admin api
{
    "healthchecks": {
        "active": {
            "unhealthy": {
                "http_statuses": [
                    500
                ],
                "tcp_failures": 5,
                "timeouts": 5,
                "http_failures": 5,
                "interval": 0
            },
            "type": "http",
            "http_path": "/health",
            "timeout": 1,
            "healthy": {
                "successes": 5,
                "interval": 5,
                "http_statuses": [
                    200
                ]
            },
            "concurrency": 10
        }
    }
}

Thanks,
Nitesh

2

Hello Nitesh,

Welcome to Kong Nation!

Indeed. You are running into: healthchecks.active.unhealthy.interval field is missing from the usptream · Issue #6 · hbagdi/go-kong · GitHub which the above PR fixes.
Ingress Controller 0.6 will include that fix and is set to be released next month.

3

Hi Harry,
Thank you for warm welcome and quick updates… :slight_smile: