[JIRA] (JENKINS-52958) Slave pod is not removed when a job is finished

[filipbrychta@gmail.com (JIRA)]

Filip Brychta created an issue

Jenkins / JENKINS-52958 Slave pod is not removed when a job is finished Issue Type: Bug Assignee: Carlos Sanchez Attachments: agent.log, master-okhttp3.log, master.log Components: kubernetes-plugin Created: 2018-08-09 10:23 Environment: Jenkins ver. 2.121.2 Kubernetes plugin 1.12.2 OpenShift Master:     v3.9.30 Kubernetes Master:     v1.9.1+a0ce1bc657 OpenShift Web Console:     v3.9.30 Priority: Minor Reporter: Filip Brychta Slave pod is correctly created and agent is connected with jenkins master: Aug 09, 2018 8:49:49 AM hudson.remoting.jnlp.Main$CuiListener status INFO: Connected When the job is done the plugin tries to terminate the slave: Aug 09, 2018 4:50:55 AM INFO org.csanchez.jenkins.plugins.kubernetes.KubernetesSlave _terminate Terminating Kubernetes instance for agent jenkins-slave-centos-base-8ffg2 but it fails with: Aug 09, 2018 4:50:55 AM WARNING jenkins.slaves.DefaultJnlpSlaveReceiver channelClosed Computer.threadPoolForRemoting 760 for jenkins-slave-centos-base-8ffg2 terminated java.nio.channels.ClosedChannelException   Agent tries to re-connect which fails: Aug 09, 2018 8:50:55 AM hudson.remoting.jnlp.Main$CuiListener status INFO: Terminated Aug 09, 2018 8:51:05 AM jenkins.slaves.restarter.JnlpSlaveRestarterInstaller$FindEffectiveRestarters$1 onReconnect ..... INFO: [JNLP4-connect connection to jenkins2.bc.jonqe.lab.eng.bos.redhat.com/10.16.23.71:37355] Local headers refused by remote: Unknown client name: jenkins-slave-centos-base-8ffg2 Aug 09, 2018 8:51:08 AM hudson.remoting.jnlp.Main$CuiListener status INFO: Protocol JNLP4-connect encountered an unexpected exception java.util.concurrent.ExecutionException: org.jenkinsci.remoting.protocol.impl.ConnectionRefusalException: Unknown client name: jenkins-slave-centos-base-8ffg2   Result is that job is successful but the slave pod is not removed and stays there with Error status: Terminated at Aug 9, 2018 10:51:09 AM with exit code 255 (Error)   Note that I'm only running the jnlp container in the pod. Full jenkins master and agent logs are attached. Note that 4 hours difference in master and agent logs are only because master logs are taken from browser so it used local timezone.     Add Comment

This message was sent by Atlassian JIRA (v7.10.1#710002-sha1:6efc396)

[arne-jenkins@kepp.net (JIRA)]

Arne Kepp commented on JENKINS-52958

Re: Slave pod is not removed when a job is finished FWIW: Jenkins 2.134 was working fine until I did a big round of plugin updates, which included kubernetes plugin 1.12.  (Upgrading to Jenkins 2.138 did not make a difference.) After downgrading to kubernetes plugin 1.7.1 everything appears to be fine and dandy again (not tested exhaustively), except the resulting CSRF issue of course.

Add Comment

[arne-jenkins@kepp.net (JIRA)]

Arne Kepp edited a comment on JENKINS-52958

Re: Slave pod is not removed when a job is finished

FWIW: Jenkins 2.134 was working fine until I did a big round of plugin updates, which included kubernetes plugin 1.12.  (Upgrading to Jenkins 2.138 did not make a difference improve the situation .)

After downgrading to kubernetes plugin 1.7.1 everything appears to be fine and dandy again (not tested exhaustively), except the resulting CSRF issue of course.

Add Comment

[jonas.buettner@spring-media.de (JIRA)]

Jonas Büttner commented on JENKINS-52958

Re: Slave pod is not removed when a job is finished

Also with Jenkins ver. 2.150.1 and Kubernetes plugin   1.13.7 Configuration:  Pod Retention = Always Time in minutes to retain agent when idle = 5 Result: Pod Status after Agent remove = Terminated: Completed

Add Comment

This message was sent by Atlassian Jira (v7.11.2#711002-sha1:fdc329d)

[jglick@cloudbees.com (JIRA)]

Jesse Glick assigned an issue to Unassigned

Jenkins / JENKINS-52958

Slave pod is not removed when a job is finished

Change By: Jesse Glick Assignee: Carlos Sanchez

Add Comment