This message was deleted.
# harvester
a
This message was deleted.
a
harvester-installer does not support configure raid; if your server BMC supports that, you need to do it there.
b
it would be possible to setup the partitions using gparted and then install harvester?
a
I did no try it before you may try this way to check if it works: in the installer UI, in the first page,
Alt+F2
to switch to tty2, use rancher/rancher to login,
sudo -i
, then check if
gparted
can be used; after that,
Alt+F1
switch back to install UI
b
got it, I’ll try and post the results here. thank you for your input
👍 1
I’ve managed to setup a LVM RAID 1 partition, but Harvester installation does not find it, I can chose only attached disks
I will try to install Harvester and after reboot to create a RAID 1 partition by updating GPT
I’ve created here an issue here about installing Harvester on a RAID1 partition: https://github.com/harvester/harvester/issues/3909 If anyone has any suggestions on how can we do this it would be so useful for us. Thank you!
w
So basically you'd like Harvester to do/handle software RAID (as opposed to BIOS or hardware RAID)?
b
yes, this is correct. most low-end dedicated servers do not have hardware RAID so it’s dangerous to deploy Harvester on them..
for example, on Hetzner BIOS RAID is disabled
@witty-jelly-95845 do you know if it’s ok to update harvester-installer to display also /dev/md0 raid disks? (https://github.com/harvester/harvester-installer/blob/76ee43a83eb97d1d7bab5803b2067626860b3a4d/pkg/console/install_panels.go#L338) • I was thinking to configure /dev/md0 as raid1 manually and then install harvester on it, but I see a filter there and I was thinking maybe there is a good reason for it..
w
Pass. You could try changing it and testing but it'll be set for a good reason and won't be supported plus you could cause yourself problems down the road.
b
does harvester support OS backup? (or at least to backup the kubernetes instance that comes with harvester)
w
b
We have two scenarios: 1. In the first scenario, the Harvester is deployed on a single node, and the VM data is backed up to S3. However, if we lose the OS disk, we will also lose the VM. Restoring the VMs on another Harvester cluster might be challenging in this case (or impossible). -> in this scenario we might lose data 2. In the second scenario, we have a high-availability (HA) Harvester cluster. If we lose one OS disk, the entire node will be lost, including the Longhorn replicas. -> in this scenario, we will re-create the node
304 Views