Jnlp tcpSlaveAgentListener Issue

Community
1

Hi Folks,
We recently upgraded Jenkins from 2.332.3 to 2.462.2 as part of this upgrade we upgrade all plugins as well as updated Jnlp version from jnlp-agent:3.29-1 to inbound-agent:4.13.3-1-jdk17 to ensure its compatible with Jenkins version 2.462.2.

Post upgrade, we are observing builds are getting aborted randomly with below error. On reviewing Jnlp container is failing with tcpSlaveAgentListener issues

image
image1063×637 118 KB

Observation:

  • Jnlp container try to connect & abort with tcpSlaveAgentListner issue
  • On previous version(jnlp-agent:3.29.1) - It reattempted until pod is created successfully
  • On updated version(inbound-agent:4.13.3-1-jdk17) - If the first attempt fails with tcpSlaveAgentListner issue, the build is aborted.

Environment Details:

  • Jenkins version: 2.462.2
  • Jnlp version: inbound-agent:4.13.3-1-jdk17
  • kubernetes plugin:4253.v7700d91739e5
  • kubernetes-client-api: 6.10.0-240.v57880ce8b_0b_2
  • JDK version: Corretto-17.0.12.7.1

Need help to understand is this issue link to Jnlp image or Kubernetes-client & also possible approach to address this issue

If someone can provide some insights/assistance as this is impacting our production environment

Thanks in advance

2

I’m not a Kubernetes administrator, but I think that the issue is in the DNS provider for the Jenkins agent that is running on Kubernetes.

The log entry:

Caused by java.net.UnknownHostException: jenkinsqe.calypso.com

indicates that the Jenkins agent could not convert the hostname to an IP address. Usually that is because there is an issue with the DNS server on that agent.