# Infrastructure as Code: Rancher on Harvester

**URL:** <https://forums.suse.com/t/infrastructure-as-code-rancher-on-harvester/38211>\
**Category:** General\
**Created:** [June 19, 2022, 12:45pm UTC](https://forums.suse.com/t/infrastructure-as-code-rancher-on-harvester/38211 "2022-06-19T12:45:59Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![tristan-cottam](https://sea2.discourse-cdn.com/flex022/user_avatar/forums.suse.com/tristan-cottam/32/8818_2.png) [@tristan-cottam](https://forums.suse.com/u/tristan-cottam)\
**Post date:** [June 19, 2022, 12:45pm UTC](https://forums.suse.com/t/infrastructure-as-code-rancher-on-harvester/38211/1 "2022-06-19T12:45:59Z")

</div>

Rancher can currently be provided by Harvester on Elemental with Rancherd (until it gets superseded by system-agent), configured through kernel command line parameters, to auto-deploy a GitRepo and kick off the GitOps pipeline.

However, Rancherd’s configuration and that of its installed components is independent from the cluster’s GitRepo, thus fragmenting the cluster’s configuration.

How could one isolate the cluster’s infrastructure (and platform) code from its application code, and potentially merge Rancherd’s configuration and the GitRepo into a single source of truth ?