Unraid OS version 6.12.0-rc1 available

Just now, Hellomynameisleo said:

To make a ZFS pool would I have to create a pool instead of an array just like making a cache pool?

yep! 

Just now, Hellomynameisleo said:

To make a ZFS pool would I have to create a pool instead of an array just like making a cache pool?

Yes, so what I did was use a spare USB stick as my 1 array drive, then I created a new pool Im calling "zfs" to test with and adding drives to it. Once you add the drives, click the first drive and change the file system to "zfs" and it will show you your options for mirror/raidz/etc. It takes care of the rest. Id reccomend 2 plugins, "ZFS Master" for snapshot management and datasets, and ZnapZend for snapshot automation per the thread linked above. 

Is Docker changing?

 

https://www.reseller.co.nz/article/706245/docker-sunsets-free-team-subscriptions-roiling-open-source-projects/#:~:text=This tier included many of,April 14%2C 2023 ...

So I noted in the post above that I successfully stopped the array, pulled a sled and replaced a drive and unraid imported it just fine, is that best practice? I know with zfs you’re supposed to offline the failed drive, will unraid have that in the gui when a drive starts to fail? So you would offline the drive, stop the array, replace drive and start? Unraid imports and resilvers the new drive. 

1 hour ago, Paul_Ber said:

Is Docker changing?

 

https://www.reseller.co.nz/article/706245/docker-sunsets-free-team-subscriptions-roiling-open-source-projects/#:~:text=This tier included many of,April 14%2C 2023 ...

 

Fortunately 6.12.x versions support alternate Repositories.

I imported the pool that was created with a plugin. I am having couple issues.

1. I cant delete a dataset.

root@UnRAID:/mnt/citadel# zfs destroy -f citadel/vsphere
cannot unmount '/mnt/citadel/vsphere': unmount failed
root@UnRAID:/mnt/citadel# zfs unmount -f citadel/vsphere
cannot unmount '/mnt/citadel/vsphere': unmount failed

2. As soon as I export my zfs SMB share, it shows unprotected in the shares menu. also shows that the share lives on the zfs pool and "Disk1" which is a ssd and in a btrfs pool (image attached).

 

edit: now I am getting "no such pool or dataset"

 

root@UnRAID:/mnt/citadel# zfs destroy -f citadel/vsphere
cannot unmount '/mnt/citadel/vsphere': no such pool or dataset
root@UnRAID:/mnt/citadel# zfs list
NAME                        USED  AVAIL     REFER  MOUNTPOINT
citadel                    5.27T  5.17T      440G  /mnt/citadel
citadel/Documents          1.97T  5.17T     1.97T  /mnt/citadel/Documents
citadel/Media              1.75T  5.17T     1.75T  /mnt/citadel/Media
citadel/Torrent_Downloads  16.5G  5.17T     16.5G  /mnt/citadel/Torrent_Downloads
citadel/nextcloud           531G  5.17T      531G  /mnt/citadel/nextcloud
citadel/software            128G  5.17T      128G  /mnt/citadel/software
citadel/veeam-backup        402G  5.17T      402G  /mnt/citadel/veeam-backup
citadel/vms                41.4G  5.17T     41.4G  /mnt/vms
citadel/vsphere            31.7G  5.17T     31.7G  /mnt/citadel/vsphere

 

Edited March 17, 2023 by Xxharry
add more info

On 3/15/2023 at 7:01 PM, B_Sinn3d said:

Great work guys. looks like lots of good improvements.  Gonna love the customizable dashboard.

 

I will have to read up on advantages of ZFS but can anyone point out some use cases that would make it more valuable to the normal unraid user that just has the disks formated in xfs/btrfs with single/dual parity?  Is read/write performance better with ZFS?

 

I for one would also appreciate a "101" summary write-up on the significance of ZFS - particularly with regard to this comment from the 6.12.0-rc1 blog:

Quote

Additionally, you may format any data device in the unRAID array with a single-device ZFS file system.

 

Does this new feature mean that Unraid will achieve some sort of "silent corruption" resilience through the deployment/adoption of ZFS? (And how does such a facility impact capacity - presumably it's a tunable thing??) I don't know enough practically anything about ZFS to be able to get a feel for this, and it's hard to find this level of info in here.

 

Thanks in advance to any kind souls willing to expand on this.

Edited March 17, 2023 by magic144

I have a couple of child datasets in my zfs pool from the zfs plugin in 6.11. E.g. pool/usr/usrname. These datasets do not seem to be showing up under Shares. I only see pool/usr and not the datasets under that one. Would just adding a share named usr/usrname for that pool add that dataset to shares?

3 hours ago, magic144 said:

Does this new feature mean that Unraid will achieve some sort of "silent corruption" resilience through the deployment/adoption of ZFS?

Using zfs for array drives will allow you to detect data corruption, but if found zfs cannot repair it since there's no redundancy in the filesystem, it's works the same as it already did with btrfs.

1 minute ago, Ter said:

I only see pool/usr

Only top level folders are Unraid shares.

 

 

Just now, JorgeB said:

Only top level folder are Unraid shares.

Ah okay, thank you. I'm going to make a pool/usrname share and just move the files from pool/usr/usrname then.

9 hours ago, Paul_Ber said:

Is Docker changing?

7 hours ago, BRiT said:

Fortunately 6.12.x versions support alternate Repositories.

Yes, but only for organization accounts.

Community accounts will stay untouched.

 

BTW, this would be better suited in the General sub Forums.

44 minutes ago, JorgeB said:

Using zfs for array drives will allow you to detect data corruption, but if found zfs cannot repair it since there's no redundancy in the filesystem, it's works the same as it already did with btrfs.

 

I tried to "study" what benefits moving to zfs would bring compared to the traditional Unraid array with xfs formatted drives. I failed to get a clear picture of this. I'm sure quite many will wonder this same when a stable 6.12.x will be released. Would it be possible to generate a short FAQ for this? That is,

• What benefits/disadvantages there are for formatting individual Unraid array disks to zfs instead of xfs?
• What benefits/disadvantages there are for changing from an Unraid array (xfs) to a zfs pool? For example, does one loose the advantage that in case of disk failure the files in other disks are still intact as the files are written to a single disk instead of distributed to several array disks?
• What are the requirements of zfs? That is, I understood that it heavily utilizes RAM and therefore one should have plenty of it to maintain system performance.
• ECC-memory is also mentioned when googling about zfs. Is it more important than with xfs for example? Of course, ECC should always been used but I guess most of Unraiders do not have ECC in their systems.
• Etc.

 

Anyway, great progress on Unraid! One happy user here.

7 hours ago, Jclendineng said:

will unraid have that in the gui when a drive starts to fail?

Stop array, unassign a device, start array, device will be offline and the pool will remain working in a degraded state (for redundant pools obviously), you can then replace the device, with the same or a new one.

 

Note that for now you can only offline one device at a time even if the pool has more redundancy, like a raidz2 pool or stripped mirrors.

1 hour ago, Ruato said:

ECC-memory is also mentioned when googling about zfs. Is it more important than with xfs for example? Of course, ECC should always been used but I guess most of Unraiders do not have ECC in their systems.

Please see this post from the ZFS Guru Allan Jude itself from Twitter, @steini84 sent me this a long time ago:

 

Sure thing ECC is "better" then non-ECC but most users are basically home users, for businesses I really get the point of ECC, at least that's my perspective.

 

I don't use ECC in my system either and use ZFS for now about 3 years and have had zero issues.

4 hours ago, Ruato said:

 

• ECC-memory is also mentioned when googling about zfs. Is it more important than with xfs for example? Of course, ECC should always been used but I guess most of Unraiders do not have ECC in their systems. 

 

I can comment on this, zfs uses available ram as a cache/buffer of sorts and so if you have a bit flip it is really bad. ECC is super important for any server housing important data but ESPECIALLY zfs. An example, I have 1 stick of ram in my test server going bad, and so when I copied a bunch of test data over, I got about 3 hardware checks telling me I had 3 corrections on one stick of ram. If that weren’t ECC that would have been 3 corruptions in my copied data. 
 

Edit. As stated above zfs still has a ton of nice features that make it better even without ecc but imo if you can, use ecc. Also, ddr5 has a lot of cool features, 1 of which is built in single and multi bit ecc on the consumer chips. It’s in the spec. It’s very expensive compared to ddr4 but someday it will be the same price and then most will have ecc without even realizing it. 

Edited March 17, 2023 by Jclendineng

4 hours ago, JorgeB said:

Stop array, unassign a device, start array, device will be offline and the pool will remain working in a degraded state (for redundant pools obviously), you can then replace the device, with the same or a new one.

 

Note that for now you can only offline one device at a time even if the pool has more redundancy, like a raidz2 pool or stripped mirrors.

Awesome! I am running raidz2 since imo that’s the most efficient for my setup. 

When using unraid now, do I still need to keep at least one hard disk in the array to start the array? Can't I start the array directly with zfs raidz pool?

6 minutes ago, ncceylan said:

When using unraid now, do I still need to keep at least one hard disk in the array to start the array? Can't I start the array directly with zfs raidz pool?

Yes that is still the case for 6.12

28 minutes ago, ncceylan said:

When using unraid now, do I still need to keep at least one hard disk in the array to start the array? Can't I start the array directly with zfs raidz pool?

Yes I just used a usb stick 

Wow so far so good, seemles change to zfs.

 

i importet my 8x2TB ssd  raidz1 pool with clone without any hicup, even renamed my pool. Awsome next step backup my "old" and "slow" HDD array and set ZFS instead

10 minutes ago, domrockt said:

next step backup my "old" and "slow" HDD array and set ZFS instead

 

If you are sure you need it, go for it! But the Unraid array is in no way deprecated, I'd say it still makes sense as the main data storage option for most of our users. We need to put together some guidance around this. 

7 minutes ago, ljm42 said:

 

If you are sure you need it, go for it! But the Unraid array is in no way deprecated, I'd say it still makes sense as the main data storage option for most of our users. We need to put together some guidance around this. 

It seems to me (could be wrong) a zfs pool is better than the array/parity anyways unless you have multiple drives of differing sizes. Unraid draw is being able to use mutt drives and just work, but if you have all the same sizes and can do zfs I would think it would be a definite upgrade? It’s complicated and can get just about as complicated as you like which is an issue, but I have a server running 12 now and setup was really user friendly, the average user would have no issues imo. Guidance definitely is a good thing though, but I was pretty surprised at how well it’s working so far. 

17 minutes ago, ljm42 said:

 

If you are sure you need it, go for it! But the Unraid array is in no way deprecated, I'd say it still makes sense as the main data storage option for most of our users. We need to put together some guidance around this. 

 

sure you are indeed right. i should say for my usecase it is better. 

I have my "sensitive" Data on any sort of array Unraid provides AND on my Synology. 

The rest of my Data is just Games/Movies/Arrs 

 

for me its a pita to copy 20+ TB with Parity over anything 100mb/s 

 

i did not want to bash the Unraid Array and want all Devs to konwo now its more accesable than ever and versataile than ever  Unraid OS Pro totally worth it!!

18 minutes ago, Jclendineng said:

but if you have all the same sizes and can do zfs I would think it would be a definite upgrade?

Until you want to add a drive and can't just do that