CODEWITHSUNDEEP
jenkins
Liz
Liz

Reputation: 61

Jenkins hudson.TcpSlaveAgentListener$ConnectionHandler run Connection Failed

Here is the setup we have - Jenkins master running on kubernetes cluster. Windows VM connected as a permanent slave. Windows VM connects through port 30502 , exposed by the Jenkins master. The default port 50000 has been modified to 30502 in the Jenkins TCP JNLP port config. The Windows VM connects successfully to the Jenkins Master.

When the Jenkins master starts, tons of these messages keep getting thrown every 2 sec. 

    hudson.TcpSlaveAgentListener$ConnectionHandler run
    WARNING: Connection #788 failed java.io.EOFException
    at java.io.DataInputStream.readFully(DataInputStream.java:197)
    at java.io.DataInputStream.readFully(DataInputStream.java:169) 
    at hudson.TcpSlaveAgentListener$ConnectionHandler.run
    (TcpSlaveAgentListener.java:244)```

Does this require any additional config that I am missing here?

Upvotes: 5

Views: 8115

Answers (1)

Viktor Bogdanov
Viktor Bogdanov

Reputation: 21

This is related to load balancer health probes. https://issues.jenkins-ci.org/browse/JENKINS-48106 We are running Kubernetes on AWS. I manually reconfigured target group health check port to some random number(60000 for example). AWS target groups have interesting behavior: If there is no healthy endpoint, it sends traffic to all endpoints. So AWS "bombards" some random port and mark all endpoints as unhealthy, but still sends traffic to all of them.

Upvotes: 2

Related Questions