You can use all 3 if you want to, 10TB each and see which one you like best.
For ISO you need file based storage. LVM/Thin is block based in its default state. You need to perform some extra steps (carve out some space) for the ISOs.
In most cases you wouldnt want to use LVM as that wont give you thin provisioning or snapshots.
If you want an advice - use LVM Thin with a slice for Directory storage usage on a side.
Good luck
Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
Thank you for this answer!
One question I still have out of curiosity: Is there anything against using the entire 30TB RAID as a directory for both?
can you rephrase this ^? What is both? And what exactly do you mean by "directory"?
My advice is to follow this document https://pve.proxmox.com/wiki/Storage:_LVM_Thin to create Thin LVM layer on top of your 30TB device.
Then using "lvcreate" manually allocate a slice, format it with your favorite filesystem and setup a mountpoint to which you can point Proxmox for ISO,snippet,etc storage.
Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
Last edited:
Personally i have lost quite a bit of data using lvm-thin. Unexpected power cuts and unexpected reboots completely destroy lvm-thin pools.
So far lvm has not let me down. I replicated the process and it seems lvm can handle unexpected power cuts and still be able to function as if nothin actually happened.
How did i notice this? I have a test proxmox server which i had this lvm-thin failure on
So far lvm has not let me down. I replicated the process and it seems lvm can handle unexpected power cuts and still be able to function as if nothin actually happened.
How did i notice this? I have a test proxmox server which i had this lvm-thin failure on
What is the use case for the 30TB Raid Drive ?
This all comes down to what is the system going to be used for.
I would like to use these 30TB for just about everything. Since the 64GB SSD and the 30TB raid remain pretty much the only disks on the server, I would use them for ISO images, VMs, containers and maybe even more.
Understandable but what i actually mean is what is going to use most of the storage. For example. This is how i would set it up for me. You situation might be different.
15TB - VMs, CT's
15TB - Backup Drive for VM's & CT's
What I mean; Is this a home server ? Are you a developer ? Is this going to store legally acquired movies and series?
Since you have 30TB you have over enough to split it in half and use 15 TB for VM's and CT's and leave the other 15TB for backups so for whatever reason something happens you have a backup of everything. This would also depend on how you have the raid setup.If this is all in 2 drives the backup would fall away. If one drive dies all the data is gone.
How is your raid setup?
Don't store your backups on the same disks you store the data you want to backup.
Thats always problematic. But for that reason you got redundant PSU, HW raid with BBU or in case of a homeserver with consumer hardware at least a UPS.
Directory storage adds overhead because of the additional filesystem you don't need to store virtual disks when a block storage like LVM-Thin is an option. A directory storage also won't allow you to use snapshots unless you switch from raw to qcow2 and with that you get additional overhead again because of the Copy-on-Write of that qcow2.
I would do it like bbgeek17 suggested:
Use the whole 30TB for LVM-Thin. You can then directly store your virtual disks as block devices (thin volumes), you get thin-provisioning and snapshot support for those. And if you also want to store files on that thin pool you could manually create an additional thin volume via CLI on top of that thin pool, format it with the filesystem of your choice, mount it via fstab and then add an directory storage pointing to that mountpoint.
That way you get both, a 30TB block storage for storing your VMs/LXCs as well as a 30TB directory storage for your files.
Or alternatively, if you want it thick, you could create a thin pool next to a thick LV using the same VG. But then you would need to create both manually via CLI and you would have to decide how much space you want for files and for virtual disks. For example 10TB for files and 20TB for virtual disks and this couldn't be easily changed later, in case you realize you need more space for files, as you can't shrink a thin pool.
Last edited:
Thank you for this detailed response! I think I'll go with that. Although it's a home server, I have a redundant power supply system, a hardware RAID with a BBU, and a UPS, which is why LVM-Thin would be suitable for me.
Wild alternative- remove the raid volume and map the disks direct to the os. then use zfs- one striped mirror pool for vm disk use, and the rest as a raidz2 for filestore.
Baby steps
Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
you wouldnt. lvm thin is not supported for multi initiator environments. If you want to use lvm thin you can only have ONE host connected to the storage.
You mean "can only have ONE host connected to - one volume - of the storage, so still couple of connections same time."
code_language.shell:
#lvs
LV VG Attr LSize Pool Origin Data% Meta% Move Log Cpy%Sync Convert
data pve twi-aotz-- <1.27t 20.75 1.80
root pve -wi-ao---- 542.51g
swap pve -wi-ao---- 8.00g
vm-100-disk-0 pve Vwi-aotz-- 96.00g data 4.10
vm-105-disk-0 pve Vwi-a-tz-- 4.00m data 14.06
vm-105-disk-1 pve Vwi-a-tz-- 100.00g data 31.98
vm-105-disk-2 pve Vwi-a-tz-- 4.00m data 1.56This is close to what pve installer created for me (I resized the volumes but the structure is the same). You can also do this manually if you have space available, create a new volume and convert it to a thin pool. Then pve can create thin volumes in that pool.
https://pve.proxmox.com/wiki/Storage:_LVM_Thin