-
Type:
Bug
-
Resolution: Unresolved
-
Priority:
Major
-
Component/s: kubernetes-plugin
-
Environment:Official Jenkins Docker image 2.534
Kubernetes plugin 4384.v1b_6367f393d9
Jenkins agent is removed from jenkins somehow, but agent pod is still running on kubernetes cluster, so k8s pod tries to re-connect every 10s.
Jenkins is refusing connection:
2025-11-04 09:52:00.882+0000 [id=2333] Â Â INFO Â Â h.TcpSlaveAgentListener$ConnectionHandler#run: Connection #1099 from /10.123.2.253:29485 failed: null
2025-11-04 09:52:00.883+0000 [id=2334] Â Â INFO Â Â h.TcpSlaveAgentListener$ConnectionHandler#run: Accepted JNLP4-connect connection #1100 from /10.123.2.253:6644
2025-11-04 09:52:00.890+0000 [id=2287] Â Â INFO Â Â o.j.r.p.i.ConnectionHeadersFilterLayer#onRecv: [JNLP4-connect connection from core.prod.publishdrive.com/10.123.2.253:6644] Refusing headers from remote: Unknown client name: livesales-43-storytel-1591-wntv6-0q38w
Connection id is increasing and several threads are open in Java:
Handling GET /tcpSlaveAgentListener/ from 10.123.2.253 : Jetty (winstone)-26
Something seems to be leaking, because after a while, I can see 100% CPU usage on Garbage Collector threads:
1991964 mainten+ 20 0 6099204 2.7g 10604 R 92.4 34.5 215:20.83 GC Thread#1 1991966 mainten+ 20 0 6099204 2.7g 10604 R 88.4 34.5 215:26.16 GC Thread#3 1991311 mainten+ 20 0 6099204 2.7g 10604 R 82.7 34.5 215:21.00 GC Thread#0 1991965 mainten+ 20 0 6099204 2.7g 10604 R 82.7 34.5 215:13.79 GC Thread#2
I sent kill -3, here are the stack trace: java.txt
Can you please:
- fix memory leak
- help to set connection retry limit in kubernetes agent pod
?
Thanks in advice!
Â
BR,
Tamas
Â
Â
