RachelAmbler
May 13, 2024
175 views
2 votes
3 Answers

The ultimate aim of this is to put nginx in front of my redis-sentinel pods so I can protected them with our corporate SSL certificates.

Redis itself is up and tested. The problem I’m getting is with the nginx deployment and associated service.

Latest update: @Vahid suggested I use an ingress instead. I’m using MicroK8s and ingress was already enabled.

Since I’m using Microk8s I think I need to update the nginx-load-balancer-microk8s-conf file, so this is what I did:

kubectl get configmap --namespace ingress

To get the config maps and then

kubectl edit configmap nginx-load-balancer-microk8s-conf --namespace ingress

to edit it – and I added:

data:
  tcp-services: |-
    6379: "default/redis-service:6379:redis-ssl"
    26379: "default/redis-sentinel-service:26379:redis-ssl"

and the rerolled the daemonset.

Next I added the same secrets to the ingress namespace.

Then the service.yaml:

apiVersion: v1
kind: Service
metadata:
  name: redis-service-lb
  namespace: redis
spec:
  type: LoadBalancer
  loadBalancerIP: 10.250.0.44
  ports:
  - port: 6379
    targetPort: 6379
    protocol: TCP
    name: tcp-redis
  - port: 26379
    targetPort: 26379
    protocol: TCP
    name: tcp-redis-sentinel
  selector:
    app: redis

Looking at kubectl get events I can see all went well:

44s         Normal    IPAllocated        service/redis-sentinel-service-lb   Assigned IP ["10.250.0.41"]

And the services is running:

redis-service-lb          LoadBalancer   10.152.183.56    10.250.0.44   6379:31301/TCP,26379:32502/TCP   5m49s

That said – still can’t get a connection using nc – or redis-cli for that matter…

I’m sure I got something wrong here: this is my first dance with ingress…

Historical post follows

We’ll start off with the nginx.conf file that’s mounted as a configmap:

server {
    listen 6379 ssl;
    server_name ki44.MyDomain.com;

    proxy_set_header Host $host;
    proxy_set_header X-Real-IP $remote_addr;

    sendfile        on;
    keepalive_timeout  65;

    ssl_protocols                               TLSv1.2 TLSv1.3;
    ssl_prefer_server_ciphers                   on;
    ssl_certificate                             /etc/nginx/certs/tls.crt;
    ssl_certificate_key                         /etc/nginx/certs/tls.key;

    location / {
        proxy_pass http://redis-sentinel.redis.svc.cluster.local:6379;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto $scheme;
    }
}

server {
    listen 26379 ssl;
    server_name ki44.MyDomain.com;

    proxy_set_header Host $host;
    proxy_set_header X-Real-IP $remote_addr;

    include       /etc/nginx/mime.types;
    default_type  application/octet-stream;

    sendfile        on;
    keepalive_timeout  65;

    ssl_protocols                               TLSv1.2 TLSv1.3;
    ssl_prefer_server_ciphers                   on;
    ssl_certificate                             /etc/nginx/certs/tls.crt;
    ssl_certificate_key                         /etc/nginx/certs/tls.key;

    location / {
        proxy_pass http://redis-sentinel.redis.svc.cluster.local:23679;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto $scheme;
    }
}

The requisite certificates are mounted as secrets as well. The pod launches with zero errors.

And next my service.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: redis-nginx
  namespace: redis
spec:
  replicas: 1
  selector:
    matchLabels:
      app: nginx-proxy
  template:
    metadata:
      labels:
        app: nginx-proxy
    spec:
      containers:
      - name: nginx
        image: nginx:latest
        ports:
        - containerPort: 6379
          name: redis
          hostPort: 6379
        - containerPort: 26379
          name: sentinel
          hostPort: 26379
        volumeMounts:
        - name: config-volume
          mountPath: /etc/nginx/conf.d
        - name: cert-volume
          mountPath: /etc/nginx/certs
      volumes:
      - name: config-volume
        configMap:
          name: nginx-config
      - name: cert-volume
        secret:
          secretName: redis-ssl
      nodeSelector:
        location: internal
        type: worker
---
apiVersion: v1
kind: Service
metadata:
  name: redis
  namespace: redis
spec:
  type: LoadBalancer
  loadBalancerIP: 10.250.0.44
  ports:
    - port: 6379
      name: redis
      targetPort: 6379
      protocol: TCP
    - port: 26379
      name: sentinel
      targetPort: 26379
      protocol: TCP
  selector:
    app: redis-nginx

I deploy this and the nginx pod fires up just fine, and I’ve confirmed by shelling into it (and installing nc inside it) that I’m able to reach the appropriate ports specified.

Everything seems happy, yet I’m unable to contact either port using nc:

nc -zv ki44.MyDomain.com 6379 and nc -zv ki44.MyDomain.com 26379 get me nowhere.

Here’s the output of kubectl get services

NAME                      TYPE           CLUSTER-IP       EXTERNAL-IP   PORT(S)                          AGE
redis-sentinel-headless   ClusterIP      None             <none>        6379/TCP,26379/TCP               3h30m
redis-sentinel            ClusterIP      10.152.183.130   <none>        6379/TCP,26379/TCP               3h30m
redis                     LoadBalancer   10.152.183.218   10.250.0.44   6379:30420/TCP,26379:30154/TCP   55m

NB: KI44.MyDomain.com does indeed resolve to 10.250.0.44.

Running kubectl describe pod <podName> gives me:

Name:             redis-nginx-7595bdd458-7grll
Namespace:        redis
Priority:         0
Service Account:  default
Node:             kwi01.md.local/10.250.0.147
Start Time:       Sun, 12 May 2024 12:40:16 -0400
Labels:           app=nginx-proxy
                  pod-template-hash=7595bdd458
Annotations:      cni.projectcalico.org/containerID: cd6b7cf612081b0fb62e98f483d97a3f2d25d40b12e0321b70786158d7e8b7a9
                  cni.projectcalico.org/podIP: 10.1.40.48/32
                  cni.projectcalico.org/podIPs: 10.1.40.48/32
Status:           Running
IP:               10.1.40.48
IPs:
  IP:           10.1.40.48
Controlled By:  ReplicaSet/redis-nginx-7595bdd458
Containers:
  nginx:
    Container ID:   containerd://b224262eadc355bfc565a8654f65760f2b2cd5444baab353f15425666e718cad
    Image:          nginx:latest
    Image ID:       docker.io/library/nginx@sha256:32e76d4f34f80e479964a0fbd4c5b4f6967b5322c8d004e9cf0cb81c93510766
    Ports:          6379/TCP, 26379/TCP
    Host Ports:     6379/TCP, 26379/TCP
    State:          Running
      Started:      Sun, 12 May 2024 12:40:17 -0400
    Ready:          True
    Restart Count:  0
    Environment:    <none>
    Mounts:
      /etc/nginx/certs from cert-volume (rw)
      /etc/nginx/conf.d from config-volume (rw)
      /var/run/secrets/kubernetes.io/serviceaccount from kube-api-access-d865w (ro)
Conditions:
  Type              Status
  Initialized       True
  Ready             True
  ContainersReady   True
  PodScheduled      True
Volumes:
  config-volume:
    Type:      ConfigMap (a volume populated by a ConfigMap)
    Name:      nginx-config
    Optional:  false
  cert-volume:
    Type:        Secret (a volume populated by a Secret)
    SecretName:  redis-ssl
    Optional:    false
  kube-api-access-d865w:
    Type:                    Projected (a volume that contains injected data from multiple sources)
    TokenExpirationSeconds:  3607
    ConfigMapName:           kube-root-ca.crt
    ConfigMapOptional:       <nil>
    DownwardAPI:             true
QoS Class:                   BestEffort
Node-Selectors:              location=internal
                             type=worker
Tolerations:                 node.kubernetes.io/not-ready:NoExecute op=Exists for 300s
                             node.kubernetes.io/unreachable:NoExecute op=Exists for 300s
Events:                      <none>

The logs from the pod:

/docker-entrypoint.sh: /docker-entrypoint.d/ is not empty, will attempt to perform configuration
/docker-entrypoint.sh: Looking for shell scripts in /docker-entrypoint.d/
/docker-entrypoint.sh: Launching /docker-entrypoint.d/10-listen-on-ipv6-by-default.sh
10-listen-on-ipv6-by-default.sh: info: can not modify /etc/nginx/conf.d/default.conf (read-only file system?)
/docker-entrypoint.sh: Sourcing /docker-entrypoint.d/15-local-resolvers.envsh
/docker-entrypoint.sh: Launching /docker-entrypoint.d/20-envsubst-on-templates.sh
/docker-entrypoint.sh: Launching /docker-entrypoint.d/30-tune-worker-processes.sh
/docker-entrypoint.sh: Configuration complete; ready for start up
2024/05/12 16:40:17 [notice] 1#1: using the "epoll" event method
2024/05/12 16:40:17 [notice] 1#1: nginx/1.25.5
2024/05/12 16:40:17 [notice] 1#1: built by gcc 12.2.0 (Debian 12.2.0-14)
2024/05/12 16:40:17 [notice] 1#1: OS: Linux 5.15.0-102-generic
2024/05/12 16:40:17 [notice] 1#1: getrlimit(RLIMIT_NOFILE): 65536:65536
2024/05/12 16:40:17 [notice] 1#1: start worker processes
2024/05/12 16:40:17 [notice] 1#1: start worker process 21
2024/05/12 16:40:17 [notice] 1#1: start worker process 22
2024/05/12 16:40:17 [notice] 1#1: start worker process 23
2024/05/12 16:40:17 [notice] 1#1: start worker process 24
2024/05/12 16:40:17 [notice] 1#1: start worker process 25
2024/05/12 16:40:17 [notice] 1#1: start worker process 26
2024/05/12 16:40:17 [notice] 1#1: start worker process 27
2024/05/12 16:40:17 [notice] 1#1: start worker process 28

Running kubectl get events also shows nothing odd:

34s         Normal    Killing             pod/redis-nginx-7595bdd458-7grll    Stopping container nginx
20s         Normal    IPAllocated         service/redis                       Assigned IP ["10.250.0.44"]
20s         Normal    ScalingReplicaSet   deployment/redis-nginx              Scaled up replica set redis-nginx-7595bdd458 to 1
20s         Normal    SuccessfulCreate    replicaset/redis-nginx-7595bdd458   Created pod: redis-nginx-7595bdd458-lwcsg
19s         Normal    Scheduled           pod/redis-nginx-7595bdd458-lwcsg    Successfully assigned redis/redis-nginx-7595bdd458-lwcsg to kwi01.md.local
18s         Normal    Pulling             pod/redis-nginx-7595bdd458-lwcsg    Pulling image "nginx:latest"
18s         Normal    Pulled              pod/redis-nginx-7595bdd458-lwcsg    Successfully pulled image "nginx:latest" in 244.860879ms (244.883758ms including waiting)
18s         Normal    Created             pod/redis-nginx-7595bdd458-lwcsg    Created container nginx
18s         Normal    Started             pod/redis-nginx-7595bdd458-lwcsg    Started container nginx

Things I’ve tried

Changing proxy_pass http://redis-sentinel.redis.svc.cluster.local.* to proxy_pass http://redis-sentinel-headless.redis.svc.cluster.local:*.
Adding and removing the hostPort assignment in my Deployment.
Removing the SSL setting and trying without using encryption.
Just testing with one port – 6379.
Created a new pod in the same namespace, installing curl then using that to connect to the nginx proxy via the internal IP address (e.g. 10.1.40.48). I then saw this attempt in the logs of the nginx pod itself, so I know the nginx proxy is indeed running.

Things I know

This isn’t a firewall issue.
This is not a network connectivity issue.
There is nothing else listening on ki44.MyDomain.com.
From within the nginx pod I can see both 637 and 26379 listening. I am also able to successfully execute nc -zv redis-sentinel.redis.svc.cluster.local:6379 (and all the other derivatives).
I have several other LoadBalancer services running just fine on the cluster – although this is the first one I’ve done with nginx in front of it.
I’m confused. 😕

Answers

Chosen as BEST ANSWER

I ended up redoing the nginx.conf file to use stream as opposed to http and I was then successfully able to connect.

user  nginx;
worker_processes  auto;

error_log  /var/log/nginx/error.log notice;
pid        /var/run/nginx.pid;

events {
    worker_connections  1024;
}


http {
    default_type  application/octet-stream;

    log_format  main  '$remote_addr - $remote_user [$time_local] "$request" '
                      '$status $body_bytes_sent "$http_referer" '
                      '"$http_user_agent" "$http_x_forwarded_for"';

    access_log  /var/log/nginx/access.log  main;

    sendfile        on;

    keepalive_timeout  65;
}

stream {
    upstream redis {
        server redis-sentinel.redis.svc.cluster.local:6379;
    }

    upstream sentinel {
        server redis-sentinel.redis.svc.cluster.local:26379;
    }

    server {
        listen 6379 ssl;
        proxy_pass redis;

        ssl_protocols                               TLSv1.2 TLSv1.3;
        ssl_prefer_server_ciphers                   on;
        ssl_certificate                             /etc/ssl/certs/tls.crt;
        ssl_certificate_key                         /etc/ssl/certs/tls.key;
    }

    server {
        listen 26379 ssl;
        proxy_pass sentinel;

        ssl_protocols                               TLSv1.2 TLSv1.3;
        ssl_prefer_server_ciphers                   on;
        ssl_certificate                             /etc/ssl/certs/tls.crt;
        ssl_certificate_key                         /etc/ssl/certs/tls.key;

    }
}

apiVersion: apps/v1
kind: Deployment
metadata:
  name: redis-nginx
  namespace: redis
spec:
  replicas: 1
  selector:
    matchLabels:
      app: redis-nginx
  template:
    metadata:
      labels:
        app: redis-nginx
    spec:
      containers:
        - name: nginx
          image: nginx:latest
          ports:
            - containerPort: 6379
              name: tcp-redis
            - containerPort: 26379
              name: tcp-sentinel
          volumeMounts:
            - name: config-volume
              mountPath: /etc/nginx
            - name: cert-volume
              mountPath: /etc/ssl/certs
      volumes:
        - name: config-volume
          configMap:
            name: nginx-config
        - name: cert-volume
          secret:
            secretName: redis-ssl
      nodeSelector:
        location: internal
        type: worker
---
apiVersion: v1
kind: Service
metadata:
  name: redis
  namespace: redis
spec:
  type: LoadBalancer
  loadBalancerIP: 10.250.0.44
  ports:
    - port: 6379
      name: tcp-redis
      targetPort: tcp-redis
      protocol: TCP
      nodePort: 32101
    - port: 26379
      name: tcp-sentinel
      targetPort: tcp-sentinel
      protocol: TCP
      nodePort: 32102
  selector:
    app: redis-nginx

(Edit)

- Vahid
- May 12, 2024 at 11:11 pm
- 0 votes
0
Change your service spec.type from LoadBalancer to ClusterIP. Now, your Redis is only exposed within your cluster. Next, create your Ingress resource like
```
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: ingress-redis
spec:
  rules:
  - host: "foo.bar.com"
    http:
      paths:
      - pathType: Prefix
        path: "/redis"
        backend:
          service:
            name: redis
            port:
              number: 6379
      - pathType: Prefix
        path: "/sentinel"
        backend:
          service:
            name: redis
            port:
              number: 26379
```
This will expose your service to the outside.

Note the host. If you want to hit your ingress your request should have this at host header. Assuming your cluster is on your machine (local). You can test it by
```
curl -H "Host: foo.bar.com" http://127.0.0.1/redis
```
You can have any DNS name you want. foo.bar.com is just an example dns record resolving to your cluster.
Login or Signup to reply.

- AbobakrAhmed
- May 13, 2024 at 1:06 am
- 0 votes
0
totally agree 100% with @Vahid, use ingress

Login or Signup to reply.

Please signup or login to give your own answer.

Click here to cancel reply.

Debian – Unable to contact service behind nginx deployment & LoadBalancer service

Historical post follows

Things I’ve tried

Things I know

Answers