Invalid HTTP port

[laoshancun]

Invalid HTTP port: tcp://172.31.14.133:9000 when upgrade from 0.8.5 to 0.9.0.

[moliware]

Could you elaborate a bit more on this?

Not sure if related but we introduced these changes regarding the port:

• Configure cerebro http port using CEREBRO_PORT env var #438
• Support CEREBRO_PORT environment variable cerebro-docker#5

[laoshancun]

Server:Docker
Env:none
Image:lmenezes/cerebro:0.9.0

[laoshancun]

Container Stdout

Pod yaml

apiVersion: v1
kind: Pod
metadata:
  labels:
    app: cerebro
    pod-template-hash: "6954484579"
  name: cerebro-6954484579-kdnhj
spec:
  containers:
  - image: lmenezes/cerebro:0.9.0
    imagePullPolicy: Always
    livenessProbe:
      failureThreshold: 3
      httpGet:
        path: /
        port: 9000
        scheme: HTTP
      initialDelaySeconds: 10
      periodSeconds: 5
      successThreshold: 1
      timeoutSeconds: 2
    name: cerebro
    ports:
    - containerPort: 9000
      name: http
      protocol: TCP
    readinessProbe:
      failureThreshold: 3
      httpGet:
        path: /
        port: 9000
        scheme: HTTP
      initialDelaySeconds: 10
      periodSeconds: 5
      successThreshold: 2
      timeoutSeconds: 2
    resources:
      limits:
        cpu: 500m
        memory: 800Mi
      requests:
        cpu: 100m
        memory: 800Mi
    securityContext:
      capabilities: {}
    terminationMessagePath: /dev/termination-log
    terminationMessagePolicy: File
    volumeMounts:
    - mountPath: /var/run/secrets/kubernetes.io/serviceaccount
      name: default-token-b4655
      readOnly: true
  dnsPolicy: ClusterFirst
  enableServiceLinks: true
  priority: 0
  restartPolicy: Always
  schedulerName: default-scheduler
  securityContext: {}
  serviceAccount: default
  serviceAccountName: default
  terminationGracePeriodSeconds: 30

[BitProcessor]

Can also confirm that 0.8.5 works, while 0.9.1 is giving the Invalid HTTP Port error.
Also running on Kubernetes.

[vmercierfr]

We experienced similar issue during upgrade to Cerebro 9.0
We identified the root cause with a workaround.

Root cause:

Cerebro 9.0 introduced CEREBRO_PORT environment variable to set the daemon port (See 04b5ca8).
This variable could conflicts with Kubernetes deployment.

Kubernetes generates (by default) environment variables for services deployed in same namespace (see https://kubernetes.io/docs/concepts/services-networking/service/#environment-variables).
If you deploy Cerebro with a Kubernetes service called cerebro, the CEREBRO_PORT environment variable will be automatically defined to tcp://<pod_id_address>:9000.

When Cerebro starts, it execute server.http.port = tcp://<pod_id_address>:9000 and crash with the following error:

Invalid HTTP port: tcp://<pod_id_address>:9000

Work-around

We identified 2 workarounds:

1. Override CEREBRO_PORT to force Cerebro port. You may add the following lines to your container:

   env:
   - name: "CEREBRO_PORT"
     value: "9000"

2. Set enableServiceLinks=false in your pod deployment (See https://kubernetes.io/docs/concepts/services-networking/connect-applications-service/#accessing-the-service) to avoid environment variables conflicts

[BitProcessor]

@vmercierfr - tried it with work-around #2 and can confirm this works fine with the 0.9.1 image.
Thanks! 👍
For others encountering this issue:

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: cerebro
  labels:
    app: cerebro
spec:
  selector:
    matchLabels:
      app: cerebro
  replicas: 1
  template:
    metadata:
      labels:
        app: cerebro
    spec:
      enableServiceLinks: false
      containers:
      - name: cerebro   
        image: lmenezes/cerebro:0.9.1
        ports:
        - containerPort: 9000
          name: http
        resources:
          requests:
            cpu: 50m
            memory: 512Mi
          limits:
            memory: 512Mi

[laoshancun]

We experienced similar issue during upgrade to Cerebro 9.0
We identified the root cause with a workaround.

Root cause:

Cerebro 9.0 introduced CEREBRO_PORT environment variable to set the daemon port (See 04b5ca8).
This variable could conflicts with Kubernetes deployment.

Kubernetes generates (by default) environment variables for services deployed in same namespace (see https://kubernetes.io/docs/concepts/services-networking/service/#environment-variables).
If you deploy Cerebro with a Kubernetes service called cerebro, the CEREBRO_PORT environment variable will be automatically defined to tcp://<pod_id_address>:9000.

When Cerebro starts, it execute server.http.port = tcp://<pod_id_address>:9000 and crash with the following error:

Invalid HTTP port: tcp://<pod_id_address>:9000

Work-around

We identified 2 workarounds:

1. Override CEREBRO_PORT to force Cerebro port. You may add the following lines to your container:

   env:
   - name: "CEREBRO_PORT"
     value: "9000"

2. Set enableServiceLinks=false in your pod deployment (See https://kubernetes.io/docs/concepts/services-networking/connect-applications-service/#accessing-the-service) to avoid environment variables conflicts

Thx,it works.

[moliware]

Thank you very much for the investigation @vmercierfr