My Proxmox VE cluster consists of four servers, each equipped with a RAID 10 setup that has a total capacity of 88TB. I prefer not to use Ceph, as it seems a bit complex for me, and even if I manage to set it up, I wouldn't know how to troubleshoot any issues that arise. My intention is to directly utilize the 88TB disk created from this RAID 10 within the virtual machines. However, whether I attempt to mount it via LXC or create a disk for the virtual machine, I am informed that the capacity is too large to create. Are there any methods or solutions you could recommend?
How to Utilize RAID 10 for Large Storage in a Virtual Machine
- Thread starter zhousp666
- Start date
Hi @zhousp666 ,
You have local storage on your servers that you want to aggregate. That means you need a distributed storage of some sort.
Ceph is an example of distributed storage, there are others. You will have to destroy your RAID10 as none of the distributed storage types need an underlying disk protection.
Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
You have local storage on your servers that you want to aggregate. That means you need a distributed storage of some sort.
Ceph is an example of distributed storage, there are others. You will have to destroy your RAID10 as none of the distributed storage types need an underlying disk protection.
Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
The most basic would be to setup the 88TB on the RAID10 as lvm-thin. Without using distributed (ie: ceph) or shared storage you can't do automatic failover to another node (although I think you can with zfs and replication). That said, you can do live migration of nodes.
How big of a disk did you attempt to create? Assuming you have 64 bit kernel, you should not be running into any sizing issues.
How big of a disk did you attempt to create? Assuming you have 64 bit kernel, you should not be running into any sizing issues.
Dude I would NEVER trust 88TB to LVM under any circumstances. You want ZFS for that or you risk silent data corruption
Depends on the raid controller, drives, etc... I assume enterprise grade 520+ byte sectors with metadata correction bits and the raid controller does background scrub, etc. ZFS is fine, but plenty of other methods to protect data from silent data corruption.
My reading of the above is that the total capacity across all the servers is 88TB, so this capacity cant be achieved by creating thin LVM pool.
Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
Since you are using qcow2 I suppose you formatted the RAID as a file storage. If you used Ext4, by default it has a maximum file size of 16 TB. This is one reason to use lvm-thin instead of a filesystem.