One disk for host data - best practices.

[vitaminman98]

Please forgive me, as I'm a total beginner. I'm wondering best practices/feedback on using one disk as the storage for multiple VMs and CTs. Here's what I have right now:


/dev/sdc is used for backups. /dev/sdd is used for security cam storage.

Things get a bit funky with ZFS formatted disk /dev/sda because that's what I'm using as subvols for a few containers and VMs:



I know this is kind of bad for redundancy purposes, as it's a single point of failure for those containers, so I'm wondering what the best practice is for this. Is it okay to use one single disk for those containers? One of them is a homeassistant OS VM, which stores logs and whatnot. I back that up elsewhere for safekeeping.

Here's what I'm running for reference:

Where 101 is a turnkey fileserver, 102-108 are tteck's (RIP) scripts for various services, and 100/104/105 are VMs.

 

[l.leahu-vladucu]

Just to check: do you have a home server? Does it have any additional redundancy and fault tolerance, like RAID? Of course, like always with storage, splitting your data over multiple drives ensures that the probability of losing a lot of data is lower, but it's always better to have redundancy and backups, especially if the server is mission-critical. In the end, it always depends on your use case.

What I'm personally also worried about is the wearout of the SSD. Note that it might still work for some time, but you should think about replacing it soon, preferably by an enterprise-grade SSD.

 

[vitaminman98]

Yep, have a home server. Do not have any additional redundancy or fault tolerance though - I actually need to create a mirror ZFS of the two 22TB drives.

The SSD's are weird - one measures down and the other measures up. so I have 99% life left with 1% wearout, and 3% wearout on the other drive with 97% life left.

 

[l.leahu-vladucu]

Yep, have a home server. Do not have any additional redundancy or fault tolerance though - I actually need to create a mirror ZFS of the two 22TB drives.

Ah, alright. If it is not mission-critical, it's of course your choice how much redundancy is enough for your use case (and peace of mind) However, keep in mind that - as always, when it comes to storage - the more you have on one single disk, the more you can lose in case something goes wrong (which is why redundancy and fault-tolerance is a good tip). In the end, whether it's ok to use a single disk redundancy and fault-tolerance depends on how important the data is to you

The SSD's are weird - one measures down and the other measures up. so I have 99% life left with 1% wearout, and 3% wearout on the other drive with 97% life left.

Ah, it seems that Proxmox VE simply reads the SMART data from the disks, but the format itself is not standardized, so it might happen that it reads the wrong values. As long as you are sure that the 99% are actually 1%, it is fine. Maybe an update of the SMART database will be updated and fix the issue, but due to the non-standardized info, it might also stay as it is. Either way, even after reaching 100%, it doesn't mean that the disk will fail, but just that the manufacturer no longer guarantees that the disk will work without issues.