I have an Odoo 14 app running on GKE cluster. The multiprocessor mode has been enabled by setting `workers = 2`. I chose the default GKE ingress to route the requests that match `/longpolling/*` to the longpolling port and the rest to the normal port.

Here's how I configure my GKE ingress.

```

resource "kubernetes_ingress" "ingress_service" {
  metadata {
    name = "ingress-service"
    annotations = {
      "networking.gke.io/managed-certificates": "subdomain-company-com"
      "networking.gke.io/v1beta1.FrontendConfig": "frontend-config" # for https redirection
    }
  }
  spec {
    rule {
      host = "dev.company.com"
      http {
        path {
          path = "/*"
          backend {
            service_name = kubernetes_service.core_service.metadata.0.name
            service_port = kubernetes_service.core_service.spec.0.port.0.port
          }
        }
      }
    }

    rule {
      host = "dev.company.com"
      http {
        path {
          path = "/longpolling/*"
          backend {
            service_name = kubernetes_service.core_service.metadata.0.name
            service_port = kubernetes_service.core_service.spec.0.port.1.port
          }
        }
      }
    }
  }

  wait_for_load_balancer = true
}

```

The Kubernetes Service looks like this.

resource "kubernetes_service" "core_service" {
  metadata {
    name = "core-service"
    annotations = {
      "cloud.google.com/backend-config" = jsonencode({
        "ports" = {
          "longpolling" = "long-polling-be-config"
        }
      })
    }
  }
  spec {
    type = "NodePort"
    selector = {
      app = "core"
    }
    port {
      name = "normal"
      port = 8080
      protocol = "TCP"
      target_port = "8069"
    }
    port {
      name = "longpolling"
      port = 8081
      protocol = "TCP"
      target_port = "8072"
    }
  }
}

```

The custom backend config `long-polling-be-config` is as follows.

```

apiVersion: cloud.google.com/v1
kind: BackendConfig
metadata:
  name: long-polling-be-config
spec:
  healthCheck:
    checkIntervalSec: 30
    healthyThreshold: 5
    type: HTTP
    requestPath: /web/database/selector
    port: 8081

```

The health check for port 8069 is defined in the Kubernetes Deployment as follows.

          readiness_probe {
            http_get {
              path = "/web/database/selector"
              port = "8069"
            }
            initial_delay_seconds = 15
            period_seconds = 30
          }

```

The backend is healthy but I am still seeing this error once every minute or so.

2021-07-10 12:41:24,014 15 INFO odoo werkzeug: 10.2.0.1 - - [10/Jul/2021 12:41:24] "POST /longpolling/poll HTTP/1.1" 200 - 1 0.001 0.010
2021-07-10 12:41:24,739 15 INFO ? werkzeug: 10.2.0.1 - - [10/Jul/2021 12:41:24] "GET /web/database/selector HTTP/1.1" 200 - 6 0.008 0.102
2021-07-10 12:41:29,106 15 INFO ? werkzeug: 10.2.0.1 - - [10/Jul/2021 12:41:29] "GET /web/database/selector HTTP/1.1" 200 - 6 0.009 0.076
2021-07-10 12:41:54,054 14 ERROR odoo odoo.http: Exception during JSON request handling.
Traceback (most recent call last):
  File "/usr/lib/python3/dist-packages/odoo/addons/base/models/ir_http.py", line 237, in _dispatch
    result = request.dispatch()
  File "/usr/lib/python3/dist-packages/odoo/http.py", line 683, in dispatch
    result = self._call_function(**self.params)
  File "/usr/lib/python3/dist-packages/odoo/http.py", line 359, in _call_function
    return checked_call(self.db, *args, **kwargs)
  File "/usr/lib/python3/dist-packages/odoo/service/model.py", line 94, in wrapper
    return f(dbname, *args, **kwargs)
  File "/usr/lib/python3/dist-packages/odoo/http.py", line 347, in checked_call
    result = self.endpoint(*a, **kw)
  File "/usr/lib/python3/dist-packages/odoo/http.py", line 912, in __call__
    return self.method(*args, **kw)
  File "/usr/lib/python3/dist-packages/odoo/http.py", line 531, in response_wrap
    response = f(*args, **kw)
  File "/usr/lib/python3/dist-packages/odoo/addons/bus/controllers/main.py", line 35, in poll
    raise Exception("bus.Bus unavailable")
Exception

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/usr/lib/python3/dist-packages/odoo/http.py", line 639, in _handle_exception
    return super(JsonRequest, self)._handle_exception(exception)
  File "/usr/lib/python3/dist-packages/odoo/http.py", line 315, in _handle_exception
    raise exception.with_traceback(None) from new_cause
Exception: bus.Bus unavailable

```

I basically tried to follow the documentation here - https://www.odoo.com/documentation/14.0/administration/install/deploy.html#https - by implementing a GKE ingress rather than an nginx ingress.

Can somebody point what I have missed or did wrong?