# Walkthrough k3s/AWS fleet setup through OpenShift Pipelines * the original repo to setup is: https://github.com/stolostron/perf-tool-collection/tree/main/aws-k3s * verified on command line, that we can still provision in AWS, given we can create the 1 VPC * the AMI image exists in the region where we're creating the ec2 instances * **blocker - issue 1**: we can create the managed cluster, but it stays in pending, I don't see manifestwork controller coming up on the k3s managed cluster. This will need to be investigated. * Resolved: registration operator fails at the latest k8s. k3s was deployed with `latest` channel. * Fix is to pin the k3s version back to 1 year ago, `K3S_VERSION=v1.22.2+k3s1` * The bash tool needs to support k3s versioning. * Pipeline: * we can setup a pipeline * trigger by creating a pipelinerun * **issue 2**: Need to modify the original bash to run inside a pod * ps command doesn't work the same * wait pid also doesn't work the same inside a container * ping was added * ssh fails to connect to the ec2 from the tekton job ??? Either the Hub network blocks? Or, tekton blocks? ssh from the laptop to the same ec2 works fine. Not a private key issue. * pipelines does not support matrix stanza out of the box, need to enable a feature flag for this and try. This allows simplified pipeline definition. * AAP: * May need to wrap this inside an ansible playbook, but again, we'll have to deal with Execution Environments * But! someone added a ansible role for k3s :-) or k3sup :+1: