Hi,
when I deployed GridGain with Kubernetes recently, the following information appeared in the log every minute:
[ClientListenerNioListener] Unable to perform handshake within timeout [timeout=10000, remoteAddr=/28.4.153.12:45476]
After searching for the reason, it was found that the health check of Kubernetes was turned on when GridGain was deployed in Kubernetes. It tried to establish a connection with GridGain's port 10800, but GridGain would mistake it for a thin client connection request. After the socket connection was established, it would send a handshake message. It reported the above information because it did not receive a response. The previous configuration is as follows:
livenessProbe:
failureThreshold:
initialDelaySeconds:300
periodSeconds:60
successThreshold:1
tcpSocket:
port:10800
timeoutSeconds:30
Now the "Unable to perform handshake within timeout" message is no longer reported after changing the port to 10900, but the container cannot be connected to the application due to the health check, which will cause the container to restart continuously. Is there a recommended port for GridGain to use tcpSocket-based liveness probes health check for Kubernetes? (11211, 47100, 47500, 49112 or still 10800?) so that Kubernetes can successfully perform health checks without affecting the normal business logic of GridGain.
Thanks!
------------------------------
Qiaoqiao Sun
office staff
ASUS Technology
------------------------------