Configuration Questions

A

adresner

Member
Mar 22, 2023
114
18
23
I have been using PBS for many years, but mostly in my home lab. The needs were basic... but at least I got familiar with it.

Started using it at the office a year ago, however I am building a new structure and can't seem to find a solid guide.

While I have used ZFS for pbs in the past, I have pretty good raid cards and plan to use them for the 3 PBS servers. Raid 10.

3 locations: HQ, Remote 1 and Remote 2. Main backups at HQ, push copies to remote locations.

Do any features within PBS require ZFS?

xfs vs ext4?

On my main PBS, I have 2 arrays, should I create a daily DS and weekly monthly DS?

Thank you for your thoughts, suggestions.
 
Last edited:
100% wasn't asking about running ZFS on hardware raid, thats a no go. I'm wondering if using ZFS opens features within PBS? Not interested in ZFS mirroring, i want to use features within PBS GUI. However, will push pull work better with ZFS storage?
 
Hi,
adresner said:
Do any features within PBS require ZFS?
Click to expand...
no, there is nothing which would not work on another of the supported filesystems. The PBS is mostly storage agnostic, garbage collection relies on atime support however, see https://pbs.proxmox.com/docs/maintenance.html#garbage-collection.
ZFS brings however some nice to have features regarding data integrity and safety, but nothing in PBS is tailored for the filesystem in use. But note, as mentioned already, do not use ZFS on top of a hardware raid.

adresner said:
On my main PBS, I have 2 arrays, should I create a daily DS and weekly monthly DS?
Click to expand...
This depends on your needs, but note that chunks are not deduplicated across datastores, so if you want to have the same data on different media that is fine. Note however that you should perform backups always to the same target datastore and use sync jobs (and its filtering capabilities) to transfer the snapshots to the other datastore, as e.g. the dirty bitmap for VM backups gets invalidated when changing backup target, so you will loose out on fast incremental backup mode. The second datastore might also have different data retention settings for prune jobs. But offsite syncs are better for redundancy as your backups are on a totally different host then.
 
  • Like
Reactions: adresner and UdoB
Hi Chris,

Thank you for your kind reply. I really appreciate all those points.

Any further thoughts re: xfs vs ext4? As will use hardware RAID, hoping that I get more performance than possible with ZFS. Obviously, there are a lot of variables there... so thats a loaded statement, ill do my own testing.

I will use a single datastore at our HQ, and then create remote connections to my remote locations. Thanks for that info.
 
You must log in or register to reply here.
Top Bottom