Remote debug container in kubernetes using intellij - docker

Question

I try to remote debug the application in attached mode with host: 192.168.99.100 and port 5005, but it tells me that it is unable to open the debugger port. The IP is 192.268.99.100 (the cluster is hosted locally via minikube).
Output of kubectl describe service catalogservice
Name: catalogservice
Namespace: default
Labels: <none>
Annotations: <none>
Selector: app=catalogservice
Type: NodePort
IP: 10.98.238.198
Port: web 31003/TCP
TargetPort: 8080/TCP
NodePort: web 31003/TCP
Endpoints: 172.17.0.6:8080
Port: debug 5005/TCP
TargetPort: 5005/TCP
NodePort: debug 32003/TCP
Endpoints: 172.17.0.6:5005
Session Affinity: None
External Traffic Policy: Cluster
Events: <none>
This is the pods service.yml:
apiVersion: v1
kind: Service
metadata:
name: catalogservice
spec:
type: NodePort
selector:
app: catalogservice
ports:
- name: web
protocol: TCP
port: 31003
nodePort: 31003
targetPort: 8080
- name: debug
protocol: TCP
port: 5005
nodePort: 32003
targetPort: 5005
And in here I expose the containers port
spec:
containers:
- name: catalogservice
image: elps/myimage
ports:
- containerPort: 8080
name: app
- containerPort: 5005
name: debug
The way I build the image:
FROM openjdk:11
VOLUME /tmp
EXPOSE 8082
ADD /target/catalogservice-0.0.1-SNAPSHOT.jar catalogservice-0.0.1-SNAPSHOT.jar
ENTRYPOINT ["java", "-agentlib:jdwp=transport=dt_socket,address=5005,server=y,suspend=n", "-jar", "catalogservice-0.0.1-SNAPSHOT.jar"]
When I execute nmap -p 5005 192.168.99.100 I receive
PORT STATE SERVICE
5005/tcp closed avt-profile-2
When I execute nmap -p 32003 192.168.99.100 I receive
PORT STATE SERVICE
32003/tcp closed unknown
When I execute nmap -p 31003 192.168.99.100 I receive
PORT STATE SERVICE
31003/tcp open unknown
When I execute kubectl get services I receive
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
catalogservice NodePort 10.108.195.102 <none> 31003:31003/TCP,5005:32003/TCP 14m
minikube service customerservice --url returns
http://192.168.99.100:32004

Answer 1

As an alternative to using a NodePort in a Service you could also use kubectl port-forward to access the debug port in your Pod.
kubectl port-forward allows using resource name, such as a pod name, to select a matching pod to port forward to since Kubernetes v1.10.
You need to expose the debug port in the Deployment yaml for the Pod
spec:
containers:
...
ports:
...
- containerPort: 5005
Then get the name of your Pod via
kubectl get pods
and then add a port-forwarding to that Pod
kubectl port-forward podname 5005:5005
In IntelliJ you will be able to connect to
Host: localhost
Port: 5005

Answer 2

Alternatively, you can use the Cloud Code Intellij plugin.
Also, if you use Fabric8, it provides the fabric8:debug goal.

Answer 3

There was a slip in the yaml you first posted as:
- containerPort: 5050
name: debug
Should be:
- containerPort: 5005
name: debug
You also need to use the external port of 32003 when configuring the IntelliJ debugger. With those changes it should work.
You may also want to think about how to make it more flexible. In the past when I've done this I've used a different form for the docker start command that allows you to turn remote debug on and off by an environment variable called REMOTE_DEBUG, which for you would be:
CMD if [ "x$REMOTE_DEBUG" = "xfalse" ] ; then java $JAVA_OPTS -jar catalogservice-0.0.1-SNAPSHOT.jar ; else java $JAVA_OPTS -agentlib:jdwp=transport=dt_socket,address=5005,server=y,suspend=n -jar catalogservice-0.0.1-SNAPSHOT.jar ; fi
You'll probably find you want to set the env var $JAVA_OPTS to limit jvm memory use to avoid issues in k8s.