Jenkins suddenly started failing to provision agents in Amazon EKS


We are using the Kubernetes plugin to provision agents in EKS and around 8:45 pm EST yesterday, with no apparent changes on our end (I’m the only admin, and I certainly wasn’t doing anything then) we started getting issues with provisioning agents. I have rebooted the EKS node and the Jenkins master. I can confirm that kubectl works fine and lists 1 node running.

I’m suspecting something must have changed on the AWS side of things.

What’s odd is that those ALPN errors don’t show up anywhere else in our logs until just before this started happening. Google around, I see people saying to ignore these “info” messages because the Java version doesn’t support ALPN, but the fact that it’s complaining about “HTTP/2” makes me wonder if Amazon changed something on their end to be HTTP/2 only?

I know this might seem too specific for a SO question, but if something did change with AWS that broke compatibility, I think this would be the right place.

From the Jenkins log at around 8:45:


Ran into this today as AWS just pushed the update for the net/http golang CVE for K8s versions 1.12.x. That patch apparently broke the version of the Kubernetes plugin we were on. Updating to the latest version of the plugin 1.18.3 resolved the issue.

Leave a Reply