Question

K8S Nginx LoadBalancer constant 503 No server is available to handle this request.

Problem

I receive HTTP 503:

HTTP 503 Service Unavailable. No server is available to handle this request

from my K8S Nginx LoadBalancer all the time, now constantly, previously every 2nd-ish time. All of the K8S droplets are reported healthy. This happens for the 2nd day in a row now (started: 2022-08-02).

See:

curl -H 'host: u00.hix.dev' http://159.89.252.251 --verbose

I use the host header only because of many recreates and DNS domain caching. When dig +short u00.hix.dev resolves only to 159.89.252.251 you can simply:

curl http://u00.hix.dev --verbose

The response:

*   Trying 159.89.252.251:80...
* Connected to 159.89.252.251 (159.89.252.251) port 80 (#0)
> GET /api/health HTTP/1.1
> Host: u00.hix.dev
> User-Agent: curl/7.79.1
> Accept: */*
>
* Mark bundle as not supporting multiuse
< HTTP/1.1 503 Service Unavailable
< cache-control: no-cache
< content-length: 104
< content-type: text/html
< date: Wed, 03 Aug 2022 13:40:01 GMT
<
* Connection #0 to host 159.89.252.251 left intact
<html><body><h1>503 Service Unavailable</h1>No server is available to handle this request.</body></html>

My K8S config

  1. I use kubernetes/ingress-nginx helm chart
  2. Proxy-protocol is both enabled in ingress-nginx’s ConfigMap (use-proxy-protocol: true) and in the Service annotation (service.beta.kubernetes.io/do-loadbalancer-enable-proxy-protocol: "true"), as mentioned here and done in kubernetes/ingress-nginx DO provider
  3. Deployments, Services and Ingress that exposes them are all in the same u00 K8S namespace.
  4. Domain records are handled via DO DNS, and all subdomains have CNAME records pointing to @. The top-level domain hix.dev points to the loadbalancer’s IPV4 address. I also configured ingress-nginx Service’s service.beta.kubernetes.io/do-loadbalancer-hostname annotation to point to my top-level domain hix.dev.
  5. Service selectors does play along with Deployments labels. I confirmed it by executing curl from one service to the other one over internal K8S http. I installed curl on the hix-api Pod, and then executed kubectl exec -it -nu00 hix-api-65994c8448-lg682 -- curl http://hix-web:4000/api/health --verbose, receiving HTTP 200. It also worked with curl http://hix-cms:5000/_health --verbose.
  6. I did not add any extra firewall rules except the defaults generated when K8S cluster is created via terraform resource digitalocean_kubernetes_cluster.

See more detailed descriptions of the K8S resources below.

Service: hix-lb-ingress-nginx-controller

Result of kubectl get svc -nlb hix-lb-ingress-nginx-controller:

NAME                              TYPE           CLUSTER-IP      EXTERNAL-IP   PORT(S)                      AGE
hix-lb-ingress-nginx-controller   LoadBalancer   10.245.217.41   hix.dev       80:30080/TCP,443:30443/TCP   4h4m

Result of kubectl get svc -nlb hix-lb-ingress-nginx-controller -o yaml | xc

apiVersion: v1
kind: Service
metadata:
  annotations:
    kubernetes.digitalocean.com/load-balancer-id: 1a9ffc3f-6389-4d42-992c-1c5f20020b51
    meta.helm.sh/release-name: hix-lb
    meta.helm.sh/release-namespace: lb
    service.beta.kubernetes.io/do-loadbalancer-certificate-id: 96944fa9-164b-4d8d-9cfa-e7713c7e8a7e
    service.beta.kubernetes.io/do-loadbalancer-enable-proxy-protocol: "true"
    service.beta.kubernetes.io/do-loadbalancer-healthcheck-port: "31111"
    service.beta.kubernetes.io/do-loadbalancer-healthcheck-protocol: tcp
    service.beta.kubernetes.io/do-loadbalancer-hostname: hix.dev
    service.beta.kubernetes.io/do-loadbalancer-id: 1a9ffc3f-6389-4d42-992c-1c5f20020b51
    service.beta.kubernetes.io/do-loadbalancer-name: hix-lb
    service.beta.kubernetes.io/do-loadbalancer-size-unit: "1"
    service.beta.kubernetes.io/do-loadbalancer-tls-ports: "443"
  creationTimestamp: "2022-08-03T10:14:20Z"
  finalizers:
  - service.kubernetes.io/load-balancer-cleanup
  labels:
    app.kubernetes.io/component: controller
    app.kubernetes.io/instance: hix-lb
    app.kubernetes.io/managed-by: Helm
    app.kubernetes.io/name: ingress-nginx
    app.kubernetes.io/part-of: ingress-nginx
    app.kubernetes.io/version: 1.3.0
    helm.sh/chart: ingress-nginx-4.2.0
  name: hix-lb-ingress-nginx-controller
  namespace: lb
  resourceVersion: "75175"
  uid: 9a2c35da-d37b-42c9-b9a8-9b3c0fcb0365
spec:
  allocateLoadBalancerNodePorts: true
  clusterIP: 10.245.217.41
  clusterIPs:
  - 10.245.217.41
  externalTrafficPolicy: Local
  healthCheckNodePort: 31917
  internalTrafficPolicy: Cluster
  ipFamilies:
  - IPv4
  ipFamilyPolicy: SingleStack
  ports:
  - appProtocol: http
    name: http
    nodePort: 30080
    port: 80
    protocol: TCP
    targetPort: 80
  - appProtocol: https
    name: https
    nodePort: 30443
    port: 443
    protocol: TCP
    targetPort: 443
  selector:
    app.kubernetes.io/component: controller
    app.kubernetes.io/instance: hix-lb
    app.kubernetes.io/name: ingress-nginx
  sessionAffinity: None
  type: LoadBalancer
status:
  loadBalancer:
    ingress:
    - hostname: hix.dev

Ingress: hix-u00

Result of kubectl get ingress -nu00 hix-u00

NAME      CLASS   HOSTS                                         ADDRESS   PORTS   AGE
hix-u00   nginx   u00.hix.dev,api.u00.hix.dev,cms.u00.hix.dev   hix.dev   80      3h58m

Result of kubectl get ingress -nu00 hix-u00 -o yaml

apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  annotations:
    kubectl.kubernetes.io/last-applied-configuration: |
      {"apiVersion":"networking.k8s.io/v1","kind":"Ingress","metadata":{"annotations":{},"name":"hix-u00","namespace":"u00"},"spec":{"defaultBackend":{"service":{"name":"hix-web","port":{"number":4000}}},"ingressClassName":"nginx","rules":[{"host":"u00.hix.dev","http":{"paths":[{"backend":{"service":{"name":"hix-web","port":{"number":4000}}},"path":"/","pathType":"Prefix"}]}},{"host":"api.u00.hix.dev","http":{"paths":[{"backend":{"service":{"name":"hix-api","port":{"number":3000}}},"path":"/","pathType":"Prefix"}]}},{"host":"cms.u00.hix.dev","http":{"paths":[{"backend":{"service":{"name":"hix-cms","port":{"number":5000}}},"path":"/","pathType":"Prefix"}]}}]}}
  creationTimestamp: "2022-08-03T10:21:56Z"
  generation: 1
  name: hix-u00
  namespace: u00
  resourceVersion: "39240"
  uid: a898b695-7410-4334-9e93-82d9892724c1
spec:
  defaultBackend:
    service:
      name: hix-web
      port:
        number: 4000
  ingressClassName: nginx
  rules:
  - host: u00.hix.dev
    http:
      paths:
      - backend:
          service:
            name: hix-web
            port:
              number: 4000
        path: /
        pathType: Prefix
  - host: api.u00.hix.dev
    http:
      paths:
      - backend:
          service:
            name: hix-api
            port:
              number: 3000
        path: /
        pathType: Prefix
  - host: cms.u00.hix.dev
    http:
      paths:
      - backend:
          service:
            name: hix-cms
            port:
              number: 5000
        path: /
        pathType: Prefix
status:
  loadBalancer:
    ingress:
    - hostname: hix.dev

Configmap hix-ingress-nginx-controller

Result of kubectl describe configmap -nlb hix-lb-ingress-nginx-controller

Name:         hix-lb-ingress-nginx-controller
Namespace:    lb
Labels:       app.kubernetes.io/component=controller
              app.kubernetes.io/instance=hix-lb
              app.kubernetes.io/managed-by=Helm
              app.kubernetes.io/name=ingress-nginx
              app.kubernetes.io/part-of=ingress-nginx
              app.kubernetes.io/version=1.3.0
              helm.sh/chart=ingress-nginx-4.2.0
Annotations:  meta.helm.sh/release-name: hix-lb
              meta.helm.sh/release-namespace: lb

Data
====
use-proxy-protocol:
----
true
allow-snippet-annotations:
----
true
gzip-level:
----
5
use-gzip:
----
true

BinaryData
====

Events:

Rest of the u00 namespace

Result of kubectl get all -nu00

NAME                           READY   STATUS    RESTARTS   AGE
pod/hix-api-65994c8448-lg682   1/1     Running   0          153m
pod/hix-bg-7b64bb7547-84f5g    1/1     Running   0          153m
pod/hix-cms-75d6fd44cf-7n5zf   1/1     Running   0          153m
pod/hix-web-6478ff6fd9-m8vmz   1/1     Running   0          153m

NAME              TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)    AGE
service/hix-api   ClusterIP   10.245.137.85    <none>        3000/TCP   153m
service/hix-cms   ClusterIP   10.245.136.119   <none>        5000/TCP   153m
service/hix-web   ClusterIP   10.245.248.229   <none>        4000/TCP   153m

NAME                      READY   UP-TO-DATE   AVAILABLE   AGE
deployment.apps/hix-api   1/1     1            1           153m
deployment.apps/hix-bg    1/1     1            1           153m
deployment.apps/hix-cms   1/1     1            1           153m
deployment.apps/hix-web   1/1     1            1           153m

NAME                                 DESIRED   CURRENT   READY   AGE
replicaset.apps/hix-api-65994c8448   1         1         1       153m
replicaset.apps/hix-bg-7b64bb7547    1         1         1       153m
replicaset.apps/hix-cms-75d6fd44cf   1         1         1       153m
replicaset.apps/hix-web-6478ff6fd9   1         1         1       153m

As you can see, all is up and running.

Versions

  • Kubernetes 1.23.9-do.0
  • Helm chart kubernetes/ingress-nginx version 4.2.0

More context

If it matters, I created the LoadBalancer using terraform’s digitalocean_loadbalancer and then update it with terraform’s helm_release in order to add DO’s Service annotations and extra ConfigMap data fields.

At some point I observed that it kinda picks up on HTTP 200 after the tf helm_release updates to the annotation tags. I first thought that their misconfiguration is the guilty party, but then realized that the refresh itself improves the availability for a short time.

I also tried following the ingress-nginx Pod logs, but they don’t say much - now, for HTTP 503 responses they display nothing. Semi-interesting stuff that I’ve seen from time to time are 400 BadRequests. Examples:

10.108.16.7 - - [03/Aug/2022:10:28:41 +0000] "PROXY TCP4 128.14.133.58 159.89.252.251 60552 443" 400 150 "-" "-" 0 0.002 [] [] - - - - 15dab2f1b8bea5d879677b0b8405efb6
10.108.16.7 - - [03/Aug/2022:10:47:44 +0000] "PROXY TCP4 163.123.143.71 159.89.252.251 59478 80" 400 150 "-" "-" 0 0.081 [] [] - - - - e468477c2177f06518a36426f61d8000
10.108.16.7 - - [03/Aug/2022:11:40:56 +0000] "PROXY TCP4 108.185.4.210 159.89.252.251 58449 80" 400 150 "-" "-" 0 0.001 [] [] - - - - 2680cfc751060ce7b9e5f431df101d75

Question

What’s the reason of the “HTTP 503 Service Unavailable. No server is available to handle this request” responses that I receive? What’s my mistake?

Thanks in advance.


Submit an answer

This textbox defaults to using Markdown to format your answer.

You can type !ref in this text area to quickly search our full set of tutorials, documentation & marketplace offerings and insert the link!

Sign In or Sign Up to Answer