Hi,. Yes you can disable ha-cluster.
When you install MicroK8s, the first thing to do is microk8s disable ha-cluster. You have to do this on every node.
After that you can join them together.
Thank you, the “do it on every node” was the missing piece 
Dear all,
I will build my “productive” home K8S based on microK8S and 4 Raspi4 4GB RAM and boot from USB3.0 SSD drives.
SInce I am currently preparing a clean install, did anyone try out microK8S HA with a similar setup? With 1 Raspi using SSD and 3 other ones using microSD I had poor performance and frequently got “database locked” error messages.
I think if this is working nicely, it would be an outstanding homelab running 4 nodes HA…
Thx mK8S community for the great support so far.
JF
@evilnick Can you give me access to this please?
2 Likes
Maybe this is enough.
1 Like
let’s hope so
Hi, I’m a bit lost about the failure domain part. Should this be set only on the “primary” master node, all of the masters or even the worker nodes? Also, if not only the primary one, should all the nodes have the same integer or separate integers?
The default setting of kube-apisever is as follow.
–service-cluster-ip-range=10.152.183.0/24
I want to change from now to 10.152.184.0/24 I couldn’t find the way to resolve. Is there the way to change what I want for avoiding network overlaping with private network ?
Best Regard canerbis
In the file /var/snap/microk8s/current/args/ha-conf what is the significance of the 42 in “failure-domain=42” ? The microk8s join operation writes “failure-domain=1” (and nothing else) to ha-conf. Are these arbitrary values or does 42 have some specific meaning?
2 Likes
If I read this article correctly, the HA only covers the container and orchestration of k8s services? How about storage layer such as PV?
Yes, that’s right - for storage it will depend on your storage solution and what HA strategies are available for it
I was wondering the same thing.