[SOLVED] Best FS for Snapshots / Migration / Cloning

[Razva]

Hey,

What FS allows me to do snapshots + migrations + cloning? Please note that I'm trying to avoid Ceph.

Thank you,
Razvan

 

[tom]

Check https://pve.proxmox.com/pve-docs/pve-admin-guide.html#chapter_storage

 

[Razva]

Check https://pve.proxmox.com/pve-docs/pve-admin-guide.html#chapter_storage

Right now I have LVM-thin. It allows for backup and snapshots. But I cannot clone or use replication.

Basically I need to know what allows me to do backup, snapshots and replication, except Ceph.

 

[Maher Khalil]

Right now I have LVM-thin. It allows for backup and snapshots. But I cannot clone or use replication.

Basically I need to know what allows me to do backup, snapshots and replication, except Ceph.

To use replication I think you have to use ZFS or BTRFS

 

[tom]

Right now I have LVM-thin. It allows for backup and snapshots. But I cannot clone or use replication.

Basically I need to know what allows me to do backup, snapshots and replication, except Ceph.

You did not mention replication in your first post.

=> If you need storage replication, ZFS is the way to go (and the only option)

 

[Razva]

You did not mention replication in your first post.

=> If you need storage replication, ZFS is the way to go (and the only option)

Any way to have disk overcommitment on ZFS? I've migrated a VM that has a 1 TB disk but only 20GB are used. On the LVM-thin node 20 GB are "blocked". On the ZFS node the `raw` image appears as 1 TB and it "blocks" 1 TB of space in the pool.

Any way to workaround this?

 

[Dunuin]

ZFS is thin-provisioned. A 1TB Zvol that only uses 20GB should only consume 20GB of the pool by default. And LZ4 compression should be enabled by default which will also strip down all zero data. If it really uses the full 1TB you should check if there are snapshots present that are blocking the freeup process or if discard/trim isn't working. ZFS will only freeup storage if the guest is telling it to do it by sending trim commands.

 

[Razva]

ZFS is thin-provisioned. A 1TB Zvol that only uses 20GB should only consume 20GB of the pool by default. And LZ4 compression should be enabled by default which will also strip down all zero data. If it really uses the full 1TB you should check if there are snapshots present that are blocking the freeup process or if discard/trim isn't working.

I've restored the VM on the "local-zfs" pool, created by default. The size is correct, 20 GB.

Do I need to create the ZFS pool via CLI? When creating it via UI the only options are RAID Level, Compression (LZ4) and ashift. I think that we're missing sparse somewhere?

 

[LnxBil]

the configuration is in the datacenter -> storage -> local-zfs:

 

[Maher Khalil]

Thank you