So how would this have ended with ZFS?

[tawnytim]

I've got my clients on Synology boxes, because I know and understand them. However, recently I had sort of a soft falling out with Synology and decided I'd take my business elsewhere for both me and my clients. A day ago, one of the Synology 2-bay systems reported a crash on disk 2. I went there, swapped out the disk, hit 'repair' on the volume manager and within 6 hours the problem was non-existent. Given this same scenario, a 2-drive machine running TrueNAS core, how would this scenario have played out?

 

[winnielinnie]

Given this same scenario, a 2-drive machine running TrueNAS core, how would this scenario have played out?

The same?

You'd offline/remove the failed drive, pop in the replacement, and then resilver the mirror vdev.

 

[tawnytim]

The same?

You'd offline/remove the failed drive, pop in the replacement, and then resilver the mirror vdev.

I thought in a zpool you couldn't just replace drives as they fail, instead having to create a new pool, transfer, etc. Am I confusing this with something else?

 

[winnielinnie]

I thought in a zpool you couldn't just replace drives as they fail, instead having to create a new pool, transfer, etc.

That's not true at all. It's actually how I described in my reply.

I'd highly recommend reading some of the guides and primers on here, not just specific to TrueNAS, but also ZFS in general. ArsTechnica has a great write-up about ZFS: https://arstechnica.com/information...01-understanding-zfs-storage-and-performance/

Keep in mind that you're still able to use the Shell / command-line / SSH to do things manually, but to get the best out of TrueNAS, try as much to stick to using it strictly as an appliance with the web GUI.

(I do use the command-line quite often, but it's usually limited to "zfs" and "iocage" commands that do not deviate from what the GUI does anyways. For example, I prefer to delete snapshots in bulk using the terminal, rather than the GUI snapshot menu.)

EDIT: There is a caveat about TrueNAS Core's offering of "Apps" (marketed as "Plugins".) Avoid them. You're better of creating and managing your own jails. This will require you to understand some fundamental FreeBSD commands and maintenance.

 

[Samuel Tai]

I thought in a zpool you couldn't just replace drives as they fail, instead having to create a new pool, transfer, etc. Am I confusing this with something else?

That's not true at all. It's actually how I described in my reply.

I'd highly recommend reading some of the guides and primers on here, not just specific to TrueNAS, but also ZFS in general. ArsTechnica has a great write-up about ZFS: https://arstechnica.com/information...01-understanding-zfs-storage-and-performance/

Keep in mind that you're still able to use the Shell / command-line / SSH to do things manually, but to get the best out of TrueNAS, try as much to stick to using it strictly as an appliance with the web GUI.

(I do use the command-line quite often, but it's usually limited to "zfs" and "iocage" commands that do not deviate from what the GUI does anyways. For example, I prefer to delete snapshots in bulk using the terminal, rather than the GUI snapshot menu.)

Actually, @tawnytim, you'd only be forced into that situation if your zpool was a stripe VDEV of 2 single disks. If your zpool was a mirror VDEV of 2 single disks, you'd do as @winnielinnie says.

 

[winnielinnie]

you'd only be forced into that situation if your zpool was a stripe VDEV of 2 single disks.

In such a situation, if you lose a single drive, you lose the entire pool, so there's nothing to "copy over" to the new pool.

 

[danb35]

Pretty much the same. Replace the failed disk and the array is rebuilt.

 

[Whattteva]

I thought in a zpool you couldn't just replace drives as they fail, instead having to create a new pool, transfer, etc. Am I confusing this with something else?

I think you're confusing resilvering with some people trying to grow/shrink their RAIDZ vdev, which indeed requires you to migrate the pool as the number of disks in a RAIDZ vdev is fixed once created.

 

[Davvo]

You also might have seen it coming thanks to the automation given by the multi_report script.
I don't know if synology has any maintenance tool built in or community created.

 

[tawnytim]

I think you're confusing resilvering with some people trying to grow/shrink their RAIDZ vdev, which indeed requires you to migrate the pool as the number of disks in a RAIDZ vdev is fixed once created.

This is exactly correct. So, if I set up a client with a 4TB pool, and they outgrow it, I will need to add larger drives, create a new pool with those drives, then migrate the data from one pool to the other - correct?

 

[tawnytim]

That's not true at all. It's actually how I described in my reply.

I'd highly recommend reading some of the guides and primers on here, not just specific to TrueNAS, but also ZFS in general. ArsTechnica has a great write-up about ZFS: https://arstechnica.com/information...01-understanding-zfs-storage-and-performance/

Keep in mind that you're still able to use the Shell / command-line / SSH to do things manually, but to get the best out of TrueNAS, try as much to stick to using it strictly as an appliance with the web GUI.

(I do use the command-line quite often, but it's usually limited to "zfs" and "iocage" commands that do not deviate from what the GUI does anyways. For example, I prefer to delete snapshots in bulk using the terminal, rather than the GUI snapshot menu.)

EDIT: There is a caveat about TrueNAS Core's offering of "Apps" (marketed as "Plugins".) Avoid them. You're better of creating and managing your own jails. This will require you to understand some fundamental FreeBSD commands and maintenance.

I'll get into that writeup asap. Thanks!

 

[winnielinnie]

So, if I set up a client with a 4TB pool, and they outgrow it, I will need to add larger drives, create a new pool with those drives, then migrate the data from one pool to the other - correct?

If you have available (physical) space and ports, and sufficient power and cabling, you can create a new vdev to expand the pool's total capacity.

Essentially you'd go from something like this:

• myPool, 4TB total
  
  • vdev1 (mirror) 4TB

To this:

• myPool, 8TB total
  
  • vdev1 (mirror) 4TB
  • vdev2 (mirror) 4TB

 

[Whattteva]

This is exactly correct. So, if I set up a client with a 4TB pool, and they outgrow it, I will need to add larger drives, create a new pool with those drives, then migrate the data from one pool to the other - correct?

Well. You have a couple options for expansion. In general, you can either:

• Replace your drives and resilver them one by one. Once the last drive resilvers fully, your pool should automatically expand.

Or:

• Add another vdev and stripe the pool across the new vdev.

The first bullet point is the reason that I go with striped mirrors instead of RAIDZ. The upgrade path is just much more flexible cause you'd only need 2 disks instead of 3+. Bit easier on my wallet. You also gain more performance out of it. Of course, the one major downside is only 50% space efficiency.

 

[danb35]

This is exactly correct. So, if I set up a client with a 4TB pool, and they outgrow it, I will need to add larger drives, create a new pool with those drives, then migrate the data from one pool to the other - correct?

Not at all; you have (at least) two other options:

• Replace the disks in the pool, one by one, with larger ones--the pool will expand once the last disk is replaced; or
• Add a new vdev to the pool.

 

[John Doe]

you might want to check out zfs hidden cost. just for beeing transparent

The 'hidden' cost of using ZFS for your home NAS

 

[Davvo]

you might want to check out zfs hidden cost. just for beeing transparent

The 'hidden' cost of using ZFS for your home NAS

I mean, it's not really hidden? It's basic understanding of how ZFS works.

 

[winnielinnie]

I mean, it's not really hidden?

Can confirm it is hidden. I had to call my bank's customer service number to challenge a fee on my account titled "additional processing". The person on the phone told me it's because I use ZFS.

This might explain why my monthly costs were lower on Synology.

 

[Whattteva]

Is that really hidden? I knew that from the start. I feel like if you don't know that, that means you didn't do your due diligence of research and just replicated whatever some YouTube video tells you blindly without actually knowing what it is the step-by-step instructions are actually telling you what to do.

 

[danb35]

Is that really hidden? I knew that from the start. I feel like if you don't know that, that means you didn't do your due diligence of research and just replicated whatever some YouTube video tells you blindly without actually knowing what it is the step-by-step instructions are actually telling you what to do.

No, but calling it a "hidden cost" is great click-bait.

 

[Redcoat]

Is that really hidden? I knew that from the start. I feel like if you don't know that, that means you didn't do your due diligence of research and just replicated whatever some YouTube video tells you blindly without actually knowing what it is the step-by-step instructions are actually telling you what to do.

Agreed!
It was crystal clear from my early 2014 reading here, allowing me, too, to know it from the start.
Actually, that level of forum clarity and "brutal frankness" was what tipped me over the edge on choosing FreeNAS as my NAS platform, a decision I have not questioned until the arrival of SCALE and the attendant doubts about the future support of FreeBSD-based Core.