This message was deleted.
# harvester
a
This message was deleted.
w
For production I think the recommendation is probably to not do it - keep the Rancher managing Harvester separate from Harvester.
k
That is pretty inconvenient. Would that mean I should dedicate at least 3 servers to run Harvester on and install Rancher on those servers to manage the other servers... then install Rancher on those servers to manage the 3 which do nothing other than host the Rancher hosting Harvester servers?
I only have 4 servers in the group right now. I don't have enough servers to run two clusters of three servers each.
And even then, I would need to install rancher on one group to install rancher on the other group to install rancher on the first group ... etc
So, I guess maybe the better way of thinking about this is. If I have 4 servers I can use to evaluate Harvester and Rancher and the core purpose of the evaluation is high availability and manageability, what is the best way to deploy Harvester and Rancher on those servers? And adding more servers is not a solution.
m
We’ve heard about this feedback and planning to address it in the v1.2.0 release. Currently we recommend to start a Rancher server in 3 VMs (with anti-affinity rule) on top of Harvester.