2024-11-03 21:24:17 +00:00
|
|
|
# Longhorn
|
|
|
|
|
|
|
|
|
|
In cluster storage management. Good for small amounts of data that needs to be
|
|
|
|
|
physically close to the host system, or data that needs to be available regardless
|
|
|
|
|
of the state of the NFS server.
|
|
|
|
|
|
|
|
|
|
> NOTE
|
|
|
|
|
>
|
|
|
|
|
> Longhorn is really cool, but I hate using it. It causes so many headaches trying
|
|
|
|
|
> to set it up. Really thankful it exists, but just use an NFS share if you can.
|
2024-11-09 20:33:12 +00:00
|
|
|
|
|
|
|
|
## Configuration
|
|
|
|
|
|
|
|
|
|
So once you have the helm release installed and you have run the ansible
|
|
|
|
|
script to bootstrap the nodes with a drive that is formatted and mounted
|
|
|
|
|
(probably to `/media/longhorn`) you need to configure each node to use that.
|
|
|
|
|
|
|
|
|
|
First disable all scheduling on all nodes. Delete all volumes (if there are any).
|
|
|
|
|
Set each default disk to scheduling false and eviction request true. Then delete it.
|
|
|
|
|
|
|
|
|
|
Next add a drive with root path at whatever you configured (probably `/media/longhorn`).
|
|
|
|
|
Just wait, it will take it some time to show the full space. Your goal is
|
|
|
|
|
to have the same amount of space on all nodes (its like RAID 0 in that way).
|
|
|
|
|
|
|
|
|
|
After that turn back on scheduling and you are good to go.
|
|
|
|
|
|
