Hello everyone, I just wanted to get an opinion on the approach i took to setting up automation testing in my company - and where better than here. Essentially, I have a repository with all my .spec.js files which contain playwright tests which are represented as jobs in my jenkins server. They are set up in a piepline architecture with an upstream job orchestrating and each test being a downstream job. They are run in parallel sets of 4 jobs. Now the isuse i was having with static agents was that if another pipeline is triggered, it creates a resource contention on my static agent as they were both using the same instance of the project located in a specific path, resulting in both crashing. So my approach was to use docker images and kubernetes clusters to isolate the environement of each job. So i created the docker image which contains a java installation amongst other dependencies so that it will be able to act as the inbound agent as well as the testing environment. I also set up the kubernetes plugin on jenkins by setting up the kubernetes node on the same local VM on which i had set up my static agent. When i tested 1 job in the pipeline (meaning upstream calls 1 downstream job and artifacts are copied from it and uplaoaded to azure blob storage), it worked fine albeit for longer wait times for agent provisioning on the kubernetes pod, and 4 minute wait times for declarative scm checkout (i had stored my jenkinsfile pipeine scripts in another repo and depending on the job and whether it is upstream or downstream it obtained it). This means that my architecture is now each job(upstream and downstream) having its own pod created which contains the dynamic jenkins agent and my project repo in which i run the required tests.
However now i fully ran one of my pipelines with 12 downstream jobs (run in sets of 4 in parallel), the already notably long time for the agent to be provisioned is prolonged as well as the fact it was eventually crashing before the downstream pipeline was intiated: angular-jobs-contract-correspondence-110-k7wp5-3b5mt-8x4rg was marked offline: Connection was broken [Pipeline] // node [Pipeline] } [Pipeline] // podTemplate [Pipeline] End of Pipeline hudson.remoting.ProxyException: java.nio.channels.ClosedChannelException at jenkins.agents.WebSocketAgents$Session.closed(WebSocketAgents.java:157) Also: hudson.remoting.ProxyException: hudson.remoting.Channel$CallSiteStackTrace: Remote call to angular-jobs-contract-correspondence-110-k7wp5-3b5mt-8x4rg at hudson.remoting.Channel.attachCallSiteStackTrace(Channel.java:1826) at hudson.remoting.Request.call(Request.java:199) at hudson.remoting.Channel.call(Channel.java:1041) at hudson.EnvVars.getRemote(EnvVars.java:438) at hudson.model.Computer.getEnvironment(Computer.java:1203) at PluginClassLoader for kubernetes//org.csanchez.jenkins.plugins.kubernetes.KubernetesSlave.checkHomeAndWarnIfNeeded(KubernetesSlave.java:521) at PluginClassLoader for kubernetes//org.csanchez.jenkins.plugins.kubernetes.KubernetesSlave.createLauncher(KubernetesSlave.java:490) at PluginClassLoader for workflow-durable-task-step//org.jenkinsci.plugins.workflow.support.steps.ExecutorStepExecution$PlaceholderTask$PlaceholderExecutable.run(ExecutorStepExecution.java:1051) at hudson.model.ResourceController.execute(ResourceController.java:101) at hudson.model.Executor.run(Executor.java:446) Also: hudson.remoting.ProxyException: org.jenkinsci.plugins.workflow.actions.ErrorAction$ErrorId: 60e0a5e9-7bd2-48fe-b37b-b8f9e13ca408 Caused: hudson.remoting.ProxyException: hudson.remoting.RequestAbortedException: java.nio.channels.ClosedChannelException at hudson.remoting.Request.abort(Request.java:346) given this error in my downstream jobs, i am not sure whether the issue has to do with network bandwidth/connectivity (my controller is hosted on a different server which is notoriously unstable, and the added dynamic jenkins agents communicating with it is causing some type of overload) or whether it is the kubernetes host running out of resources causing the websocket to be closed somehow, because when i ran the one job, it ran successfully, albeit the fact it took longer overall due to the time required to provision the agent and to start running the first shell command in the downstream job, as well as the longer time taken to checkout the declarative scm script. I understand it is confusing what i just explained i am willing to explain further i just need some assistance as i am the only person in my company who has self taught myself to utilise these technologies so i would appreciate any input -- 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/1117400a-ecc0-436c-9cb5-e4e62ef730afn%40googlegroups.com.
