This message was deleted.
# harvester
a
This message was deleted.
1
b
also I’m looking forward for a good answer to your question 🙂
in our case, we’ve deployed Rancher using Docker in a Harvester VM (this solution is not HA)
it would be nice if Harvester could come with an addon to deploy Rancher on the K3S cluster used by Harvester, this will make things much easier and also it’s a HA solution
r
A typical use case is to import the Harvester cluster into Rancher outside of that Harvester cluster. And that Rancher can manage multiple Harvester clusters scattered at different sites simultaneously. For a non-production environment, it is possible to spin up VMs running Rancher on the Harvester cluster and then import the Harvester cluster to the Rancher. That’s basically what the sentence means in the Harvester doc:
To use Rancher with Harvester, please install the Rancher and Harvester in two separated servers. If you want to try out the integration features, you can create a VM in Harvester and install Rancher v2.6.3 or above (the latest stable version is recommended).
And in the upcoming v1.2, we plan to unleash the embedded Rancher that comes with Harvester (in the form of addons) so ppl can run bare-metal workloads directly without the hassle of setting up guest clusters.
🙌 1
👍 1
s
@red-king-19196 Thank you for your response. Being able to use the embedded Rancher will be great for people who have small clusters and would rather not have three additional servers just to run another Rancher. I was actually just testing out 1.2 rc4 and noticed a link to the embedded Rancher was there under Support, but that it says “only to be used for debugging and validation”. Is the plan for that disclaimer to go away by 1.2 release and using it for more than just debugging and validation will be a supported setup?
If I only have three bare metal servers, would that potentially be a problem for performance? Maybe ok as long as data and installation are on separate disks? Just thinking about the typical advice of having 3 control plane nodes separate from your worker nodes. My original plan has been to use Proxmox on the three bare metal nodes with each node running a control plane vm and worker node vm, but considering if I should replace Proxmox with Harvester in this situation.
r
The link under Support might not be the one of interest. For more details, please check this out. About your second question, I’m not pretty sure what you intended to do. From Harvester’s perspective, if you deploy it on three nodes, then all the three nodes will be management node. They’re all control nodes and worker nodes. You can run VMs on them and the cluster is highly-available.
👍 1
🤤 1