This message was deleted.
# general
a
This message was deleted.
f
Any help would be appreciated, please ping me if you reply
It appears you can only change this when there are not any nodes inside your cluster or in the creation stages
If there are nodes deployed, it doesn't reconfigured, you must rebuild a new cluster
a
Might be!
f
Looks like i am about to rebuild 10 nodes!!
a
I think that server is more of an issue (master nodes to be synced)
agent nodes are not important
f
All 10 nodes have all 3 roles
a
NO
why
f
this is just a test for now
a
3 nodes server, others agents
f
do i just do 3 controlplane + etcd and the rest workers?
a
yes
You also save on a CPU 😄
and server count should be ODD
f
all roles is not good i assume from the all caps
🫣
a
1,3,5,...
this is because of quorum
f
I have a few super big clusters that are laid out that way, (thankfully)
I will follow that layout
a
production 3 is minimum, for dev/test you don't need it realistically more than 1
f
This other cluster has the following
32 workers
9 control plane
9 etcd
totalling 50 nodes
is this a decent layout?
a
I have management cluster with Fleet CD, and other clusters automatically spin up services for that cluster
you have separated etcd?
f
Yeah we have dedicated servers just for those 2 roles (controlplane + etcd)
the rest are considered workers
a
Well based on the numbers it's maybe an overkill based on the recommendation
f
Eventually I need to migrate from docker -> HA backend
a
what are you using?
rke2?
f
RKE1
a
Ah, okay
f
2.6.4 rancher
a
I'm more of rke2 😄
f
Someone else used the Docker Single Node for the Rancher interface, so I need to move the docker install for rancher into an HA kubernetes cluster soon
Hopefully thats not too difficult
a
Yeah I put the rancher into dedicated cluster
and then I connect all clusters to that rancher interface
so I have single point of view
well you just need to write all the manifests 😄
f
Yeah i have clusters that connect into a single rancher machine at the momenmt