khuedoan-homelab/docs/how-to-guides/single-node-cluster-adjustments.md
2024-03-13 10:11:25 +07:00

56 lines
1.3 KiB
Markdown

# Single node cluster adjustments
!!! danger
This is not officially supported and I don't regularly test it,
I highly recommend using multiple nodes.
Using a single node could lead to data loss unless your backup strategy is rock solid,
make sure you are **ABSOLUTELY CERTAIN** this is what you want.
Update the following changes, then commit and push.
## Remove storage redundancy
Set pod counts and number of data copies to `1`:
```yaml title="system/rook-ceph/values.yaml" hl_lines="4 6 11 12 18 22 25"
rook-ceph-cluster:
cephClusterSpec:
mon:
count: 1
mgr:
count: 1
cephBlockPools:
- name: standard-rwo
spec:
replicated:
size: 1
requireSafeReplicaSize: false
cephFileSystems:
- name: standard-rwx
spec:
metadataPool:
replicated:
size: 1
dataPools:
- name: data0
replicated:
size: 1
metadataServer:
activeCount: 1
activeStandby: false
```
## Disable automatic upgrade
Because they will try to drain the only node, the pods will have no place to go.
Remove them entirely:
```sh
rm -rf system/kured
```
Commit and push the change.
You can revert it later when you add more nodes.