We're having a significant issue with our Jenkins env. Our Jenkins setup is running in Kubernetes, and sometimes the master is no longer able to provision new nodes, and the existing nodes don't disconnect.
We're using 1.10.1 of Kubernets plugin, and Jenkins version is 2.121.2, and running in Google Cloud. This issue first started at the end of July, and it has been growing worse over the weeks. Each time we think we find the cause and implement a fix, it just comes back to be an issue a few days later. - At first the issue appeared to be based on load, and was impacted by the slave pods not having proper CPU/memory resource requests. They all have proper requests now, and with better monitoring that doesn't seem to be related. - Then it seemed like maybe the jvm heap size wasn't enough, so we increased that and monitored those values with prometheus/grafana. And we don't see much relation to that as that problem either. - If we leave Jenkins untouched and don't start any more builds, it seems to resolve itself of the problem over time. But in a typical working day, with many developers pushing code, that is not a possible solution. - The best solution we have is to restart the Jenkins master when this happens. There is one thing of interest that I want to investigate more, but I'm not sure how to get better logging/details when the issue is happening. As builds are executed in the k8s pods, the Jenkins master lists the pod/job in the "Build Executer Status" table in the left column. Each "stage" of the pipeline is described as the job progresses. When the problem is happening, there will be 2 or more builds in this table, and each of them will show "part" as the stage they are in. It's unclear exactly what is setting that to be "part". And maybe with a better understanding of that, we might have a better idea of what is causing this issue. -- You received this message because you are subscribed to the Google Groups "Jenkins Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to jenkinsci-users+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/jenkinsci-users/4a161cd1-1412-4208-8bd3-eb4fed564ae7%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.