2-3 minute delay starting Jenkins pipelines

[davesteinberg]

I'm not sure that this is strictly a bug, but it's an undesirable behaviour that could be investigated and, hopefully, improved. I've encountered it while doing the MOOC, but @mjperrins indicates that it's the usual behaviour with Jenkins pipelines in OpenShift.

Describe the bug

Whenever I push a change to my GitHub repo, I'm seeing a 2-3 minute delay after the pipeline shows up in OpenShift before it appears in Jenkins. Here's how it looks in the OpenShift console (under Builds > Pipelines) during that time:

Matt says that during this time Jenkins is downloading the images that are used for the build steps and this time is not reflected in the Jenkins UI. He refers to the Jenkins Kubernetes plug-in docs: https://github.com/jenkinsci/kubernetes-plugin

Expected behavior

I would hope to eliminate or at least minimize this delay.

I have experienced Jenkins running in OpenShift in a client environment without a noticeable delay.

IBM Cloud

Select the services and tools affected

• [ ] IBM Kubernetes Service Managed Service
• [X] Red Hat OpenShift Managed Services
• [X] Jenkins
• [ ] Tekton
• [ ] SonarQube
• [ ] Pact
• [ ] Artifactory
• [ ] IBM Image Registry

Additional context

More notes from Matt:

moving docker images around into namespaces and running them is a lot of software moving about I think 3min is pretty fair time for Jenkins to prepare itself to run a pipeline

They do I agree on Dev time , but 15 mins for a UBI based images , UBI are way bigger than standard open source that is probably the difference , check out the sizes of the dependencies , maybe use SysDig to monitor the agents memory CPU and behaviour all the data is inside the cluster , we have found being UBI based does add overhead but then thats for security and compliance reasons

[mjperrins]

This is interesting, on mooc-team-one I had a pipeline kick off nearly instantly , we will have to do some research and investigation on this ,