Selenium 4: Chrome Node does not register correctly to the hub - selenium

Question

I have an Openshift 3 Cluster containing the two following containers: selenium-hub and selenium-node-chrome. Please see below the attached deployment and service yaml files.
Hub Deployment:
apiVersion: apps.openshift.io/v1
kind: DeploymentConfig
metadata:
labels:
app: selenium-hub
selenium-hub: master
name: selenium-hub
spec:
replicas: 1
selector:
type: selenium-hub
template:
metadata:
labels:
type: selenium-hub
name: selenium-hub
spec:
containers:
- image: 'selenium/hub:latest'
imagePullPolicy: IfNotPresent
name: master
ports:
- containerPort: 4444
protocol: TCP
- containerPort: 4442
protocol: TCP
- containerPort: 4443
protocol: TCP
triggers:
- type: ConfigChange
Hub Service:
apiVersion: v1
kind: Service
metadata:
labels:
app: selenium-hub
selenium-hub: master
name: selenium-hub
spec:
ports:
- name: selenium-hub
port: 4444
protocol: TCP
targetPort: 4444
- name: publish
port: 4442
protocol: TCP
targetPort: 4442
- name: subscribe
port: 4443
protocol: TCP
targetPort: 4443
selector:
type: selenium-hub
type: ClusterIP
Node Deployment:
apiVersion: apps.openshift.io/v1
kind: DeploymentConfig
metadata:
labels:
app: selenium-node-chrome
name: selenium-node-chrome
spec:
replicas: 1
revisionHistoryLimit: 10
selector:
browser: chrome
template:
metadata:
labels:
app: node-chrome
browser: chrome
name: selenium-node-chrome-master
spec:
containers:
- env:
- name: SE_EVENT_BUS_HOST
value: selenium-hub
- name: SE_EVENT_BUS_PUBLISH_PORT
value: '4442'
- name: SE_EVENT_BUS_SUBSCRIBE_PORT
value: '4443'
- name: SE_NODE_HOST
value: node-chrome
- name: SE_NODE_PORT
value: '5555'
image: 'selenium/node-chrome:4.0.0-20211102'
imagePullPolicy: IfNotPresent
name: master
ports:
- containerPort: 5555
protocol: TCP
triggers:
- type: ConfigChange
Node Service:
apiVersion: v1
kind: Service
metadata:
labels:
app: selenium-node-chrome
name: selenium-node-chrome
spec:
ports:
- name: node-port
port: 5555
protocol: TCP
targetPort: 5555
- name: node-port-grid
port: 4444
protocol: TCP
targetPort: 4444
selector:
browser: chrome
type: ClusterIP
My Issue:
The hub and the node are starting, but the node just keeps sending the registration event and the hub is logging some infos, which i dont really understand. Please see the logs attached below.
Node Log:
Setting up SE_NODE_GRID_URL...
Selenium Grid Node configuration:
[events]
publish = "tcp://selenium-hub:4442"
subscribe = "tcp://selenium-hub:4443"
[server]
host = "node-chrome"
port = "5555"
[node]
session-timeout = "300"
override-max-sessions = false
detect-drivers = false
max-sessions = 1
[[node.driver-configuration]]
display-name = "chrome"
stereotype = '{"browserName": "chrome", "browserVersion": "95.0", "platformName": "Linux"}'
max-sessions = 1
Starting Selenium Grid Node...
11:34:31.635 INFO [LoggingOptions.configureLogEncoding] - Using the system default encoding
11:34:31.643 INFO [OpenTelemetryTracer.createTracer] - Using OpenTelemetry for tracing
11:34:31.774 INFO [UnboundZmqEventBus.<init>] - Connecting to tcp://selenium-hub:4442 and tcp://selenium-hub:4443
11:34:31.843 INFO [UnboundZmqEventBus.<init>] - Sockets created
11:34:32.854 INFO [UnboundZmqEventBus.<init>] - Event bus ready
11:34:33.018 INFO [NodeServer.createHandlers] - Reporting self as: http://node-chrome:5555
11:34:33.044 INFO [NodeOptions.getSessionFactories] - Detected 1 available processors
11:34:33.115 INFO [NodeOptions.report] - Adding chrome for {"browserVersion": "95.0","browserName": "chrome","platformName": "Linux","se:vncEnabled": true} 1 times
11:34:33.130 INFO [Node.<init>] - Binding additional locator mechanisms: name, relative, id
11:34:33.471 INFO [NodeServer$1.start] - Starting registration process for node id 2832e819-cf31-4bd9-afcc-cd2b27578d58
11:34:33.473 INFO [NodeServer.execute] - Started Selenium node 4.0.0 (revision 3a21814679): http://node-chrome:5555
11:34:33.476 INFO [NodeServer$1.lambda$start$1] - Sending registration event...
11:34:43.479 INFO [NodeServer$1.lambda$start$1] - Sending registration event...
11:34:53.481 INFO [NodeServer$1.lambda$start$1] - Sending registration event...
Hub Log:
2021-12-07 11:14:22,663 INFO spawned: 'selenium-grid-hub' with pid 11
2021-12-07 11:14:23,664 INFO success: selenium-grid-hub entered RUNNING state, process has stayed up for > than 0 seconds (startsecs)
11:14:23.953 INFO [LoggingOptions.configureLogEncoding] - Using the system default encoding
11:14:23.961 INFO [OpenTelemetryTracer.createTracer] - Using OpenTelemetry for tracing
11:14:24.136 INFO [BoundZmqEventBus.<init>] - XPUB binding to [binding to tcp://*:4442, advertising as tcp://XXXXXXX:4442], XSUB binding to [binding to tcp://*:4443, advertising as tcp://XXXXXX:4443]
11:14:24.246 INFO [UnboundZmqEventBus.<init>] - Connecting to tcp://XXXXXX:4442 and tcp://XXXXXXX:4443
11:14:24.275 INFO [UnboundZmqEventBus.<init>] - Sockets created
11:14:25.278 INFO [UnboundZmqEventBus.<init>] - Event bus ready
11:14:26.232 INFO [Hub.execute] - Started Selenium Hub 4.1.0 (revision 87802e897b): http://XXXXXXX:4444
11:14:46.965 INFO [Node.<init>] - Binding additional locator mechanisms: name, relative, id
11:15:46.916 INFO [Node.<init>] - Binding additional locator mechanisms: relative, name, id
11:17:52.377 INFO [Node.<init>] - Binding additional locator mechanisms: relative, id, name
Can anyone tell me why the hub wont register the node?
If you need any further informations, let me know.
Thanks alot

Answer 1

So, bit late, but still I had this same issue - the docker-compose example gave me selenium-hub as the host, which is correct in that scenario as it points towards the container defined by the selenium-hub service.
However, in Kubernetes, the inter-pod communication needs to go via a Service. There are multiple kinds of Service, but in order to access it from inside the cluster, it's easiest in this case to use a ClusterIP (docs here for more info).
The way I resolved it was to have a Service for both the ports that the event bus uses:
bus-publisher (port 4442)
bus-subscription (port 4443)
In a manifest yaml, this looks like:
apiVersion: v1
kind: Service
metadata:
labels:
app-name: selenium
name: bus-sub
namespace: selenium
spec:
ports:
- port: 4443
protocol: TCP
targetPort: 4443
selector:
app: selenium-hub
type: ClusterIP

Answer 2

you didn't expose the ports 4443 and 4442 from the hub container (see ports section of spec.containers)

Answer 3

You are in same machine so I think you don't need to use the environment variable: SE_NODE_HOST in the node deployment only use these variables:
SE_EVENT_BUS_HOST=selenium-hub
SE_EVENT_BUS_PUBLISH_PORT=4442
SE_EVENT_BUS_SUBSCRIBE_PORT=4443
If you think you aren't in the same VM, you need to config the node deployment correctly by using these environment variables :
SE_EVENT_BUS_HOST=<ip-of-hub-machine>
SE_EVENT_BUS_PUBLISH_PORT=4442
SE_EVENT_BUS_SUBSCRIBE_PORT=4443
SE_NODE_HOST=<ip-of-node-machine>
Please don't add unused environment variables like:'SE_NODE_PORT' because selenium image doesn't support different environment variables besides the environment variables you can read in the documents in Github 'docker-selenium' project: https://github.com/SeleniumHQ/docker-selenium.
If you are so much want to use your variable. So create your own selenium image (I don't recommend that) I succuss with what I say to you.