-
Bug
-
Resolution: Duplicate
-
Critical
-
None
-
Jenkins ver. 2.107.3
kubernetes-plugin 1.12.2
Kubernetes 1.8
Recently i've upgraded the plugin to 1.12.2 and started to face issues with idle agents termination.
It looks like when jenkins decides to remove the k8s agent (according to retention policy) it disconnects the computer, but leaves the pod as it is.
Thats why after some time the pod ends up in "Error" state since connection to agent is broken.
Corresponding logs here:
2018-09-05 21:42:58.121+0000 [id=212] INFO o.c.j.p.k.KubernetesSlave#_terminate: Terminating Kubernetes instance for agent test-r4q15 2018-09-05 21:42:58.122+0000 [id=212] FINE o.c.j.p.k.KubernetesCloud#connect: Building connection to Kubernetes kubernetes URL namespace nstest 2018-09-05 21:42:58.124+0000 [id=212] FINE o.c.j.p.k.KubernetesFactoryAdapter#createClient: Configuring Kubernetes client from kubeconfig file 2018-09-05 21:42:58.134+0000 [id=212] FINE o.c.j.p.k.KubernetesFactoryAdapter#createClient: Creating Kubernetes client: KubernetesFactoryAdapter [serviceAddress=, namespace=nstest, caCertData=null, credentials=org.jenkinsci.plugins.plaincredentials.impl.FileCredentialsImpl@123, skipTlsVerify=false, connectTimeout=0, readTimeout=0] 2018-09-05 21:42:58.140+0000 [id=212] FINE o.c.j.p.k.KubernetesCloud#connect: Connected to Kubernetes kubernetes URL
From the code perspective it looks like nothing happens next to this line: https://github.com/jenkinsci/kubernetes-plugin/blob/kubernetes-1.12.2/src/main/java/org/csanchez/jenkins/plugins/kubernetes/KubernetesSlave.java#L245
- duplicates
-
-
- Closed
-
