NOOB OVM questions

    • New

    I am New to OMV and have a few questions.


    First some context. I am looking to move from truenas-scale to OMV. My Truenas looks like this:


    Supermicro X11SCL-F Micro ATX LGA1151

    Intel(R) Core(TM) i3-9100 CPU

    32 GB ECC ram

    boot disk 112 GB ssd (mirrored)

    storage 4 x 4TB cmr 7200 drives 1 pool raidz2


    I am assuming my HW will work with OMV.

    Usecase : I basically have 5 SMB shares and a couple NFS shares. The SMB are for shared documents (small files like pdf, doc, ect) and for windows full disk backup. The NFS are shares used by my Proxmox cluster.

    Question 1 : is there a way to install OMV on mirrored boot drives at install time or or would I need to follow something like this (Installing OMV on RAID1 ?) to get mirrored drives.


    Q1 part B Generally does OMV best practice recommend mirrored boot devices?


    Question 2 : truenas uses zfs, but on OMV I plan to use raid 6 + BTRFS


    Q2 part A: Why btrfs instead of zfs on OMV : I don’t want to deal with installing (and keeping up with ) a different kernel. (I’m kinda lazy ) is there is a technical reason I should use zfs I’d like to know.

    I have also read that btrfs is better integrated to OMV because of some licensing issue with zfs (I don't really understand that because truenas scale includes it )


      Q2 part B: I know BTRFS has issues with raid 5/6 but if I use linux md(?) to create the raid and then place btrfs on top of that, I would avoid this issue. Is that correct or am I misinformed?


      Q2 part C: I have read that because btrfs uses cow, this causes disk fragmentation and the btrfs page (https://btrfs.readthedocs.io/en/latest/Status.html) says the fragment operation is not stable. What is the best way to deal with disk fragmentation since the built in tool is not “stable”


    Question 3: I want to back up OMV to backblaze b2. They suggest using rclone (https://www.backblaze.com/docs…-rclone-with-backblaze-b2 ). Is there a better approach?


    Question 4: What is the best way to backup/restore the boot drive ?


    Question 5: is there an SMB turning guide for OMV 7.x ?


    Question 6: is there a best practices check list of stuff you should do/configure/think about (like change the password, turn on the firewall, etc ) for OMV 7.x ?

    I realize this post has a lot of questions. I have done the research I know to do with my limited knowledge of OMV and would really appreciate feedback from people with more knowledge than I.

    Thanks so much for your time,

    GF

  • chente

    Approved the thread.
    • New

    I switched to OMV about a year ago (coming from proprietary Qnap and Synology NAS), but I already had quite some knowledge in Linux. So I don't consider myself a newbie (not at all), but this still means, that some of my knowledge was (and maybe still is) outdated and I had no knowledge about OMV.

    So I hope, my answers will be helpful for you.

    Quote from genfoch01

    Question 1 : is there a way to install OMV on mirrored boot drives at install time

    I did exactly this on one of my NAS.

    I have a NAS with 2 NVME SSDs. Both SSDs are equally partitioned (/boot (mdadm Raid-1); / (btrfs Raid-1); cache (mdadm Raid-1)).

    This is not completely straight forward, but I took notes and could provide you those notes / post them here in the forum.


    Quote from genfoch01

    is there is a technical reason I should use zfs

    I asked myself the same question and I've put a fair amount of time investigating it. But this is still only my answer for my needs.

    That having said, I decided to use btrfs for the following reasons:

    • it can handle raids of different sized disks (yes, I'm using Raid-5, but not on all of my drives)
    • it is built into the kernel. If I run into a problem, I can read my disks with any other Linux machine.
    • btrfs can even run on older HW (32 bit), ZFS can't

    ZFS is said to be more stable and this is especially true for Raid-5/6, where the btrfs-devs still consider it as unstable. I've read (and understood) all the implications this has and can deal with it.


    Quote from genfoch01

    if I use linux md(?) to create the raid and then place btrfs on top of that, I would avoid this issue.

    This is true. You must be aware though, that you will loose one (important) feature and that is auto correction. Detection will still work.

    That means on a true raid (Raid-1/5/6) when btrfs detects a bitflip, it will have a second set of data where it can get the original data from. On a btrfs (with single profile) on top of an mdadm raid, it will only detect the bitflip and prevent you from accessing the data (read error). You must then get the original data from a backup.

    Yes I have all that seen myself in the last year since I have both an mdadm Raid-5 with brtfs (single profile) on top, and a btrfs raid-5 and also a btrfs Raid-1.



    Quote

    What is the best way to deal with disk fragmentation

    Not to deal with it. You cant have both.

    When you want to have the benefits of COW (especially deduplication), then you have to accept fragmentation.

    If you don't want to have fragmentation and its important to you, deactivate COW (wich implies to not use snapshots, if you take this assumption seriously).


    Quote from genfoch01

    [Backup] Is there a better approach?

    I don't know about rclone, but I use btrfs snapshots with btrfs send / receive (using the tool btrbk).


    I used to rely on rsnapshot before (that was, when I was still using Ext4; rsnapshot is based on rsync). First I wanted to simply go on with rsnapshot, but when I learned about the abilities of btrfs snapshots and btrfs send / receive, I changed my mind.


    Quote from genfoch01

    What is the best way to backup/restore the boot drive ?

    That is a good question. I don't have the perfect solution so far.

    (At least not when compared to my own standards of a perfect solution)

    So, I use omw-regen to save all the settings.

    But I also have my boot drive on a Raid-1 (as seen above) to gain myself some time if (one of) the system drive(s) fails.

    And of course I do btrfs snapshots before each major update, to at least have the chance for a roll-back.


    Quote from genfoch01

    Question 6: is there a best practices check list of stuff you should do

    There are some really good how-to's and tutorials on OMV-Extras and there also some guides and how-to's here in the forum in the "guides" section.

    TerraMaster T6-423

    Qnap TS-853A

    Qnap TS-451+

    Qnap TS-259 Pro+

    • New
    Quote from Quacksalber

    This is true. You must be aware though, that you will loose one (important) feature and that is auto correction. Detection will still work.

    That means on a true raid (Raid-1/5/6) when btrfs detects a bitflip, it will have a second set of data where it can get the original data from. On a btrfs (with single profile) on top of an mdadm raid, it will only detect the bitflip and prevent you from accessing the data (read error). You must then get the original data from a backup. "guides" section.

    First,

    thanks for taking the time to respond I know I have a lot of questions in this post.


    This (quote above } was unexpected (thank you for that, I was unaware! ) as self healing and resiliency are my most important concerns. So btrfs on top of md raid does not sound like a good solution. Looking at btrfs ( presented from the OMV gui) only it appears to only support raid0, 1, and 10, so a fault tolerance of 0, 1, or 1(disk failures) if I want auto bitflip correction (I'm assuming this is what a "scrub" is.).


    you said you had raid5: did you implement this in a way that allows btrfs to correct detected bitflips?


    -GF

    • New

    Yes, the OMV Web-UI is restrictive with regards to btrfs.

    I did most of the stuff on the command line.

    And you need to be prepared to use the command line in case you want to do anything more advanced than creating / formatting a device with btrfs (e..g. adding another device and thus growing the FS, or replacing a failed disk).

    So, if you are afraid of the command line, this might not be the way for you to go...

    Quote from genfoch01

    I'm assuming this is what a "scrub" is.

    Not exactly. You get bitflip correction even without a scrub.

    Let's assume you want to read a file (on raid-1) and the mirrored file on the "left" is corrupt and the file on the "right" is still valid.

    When accessing that file, btrfs will only access one of both. If it tries to read the "left" file, it will notice the bitflip and correct it on the fly (without a scrub, simply by accessing the "right" file then). But if it happens to access the "right" file right from the beginning, the corruption will remain undetected. This is where scrub comes in. Scrub tests all of the files for bitflips (and corrects them, if possible).

    Quote from genfoch01

    you said you had raid5: did you implement this in a way that allows btrfs to correct detected bitflips?

    Yes. I used a native btrfs raid-5. This implements auto correction.

    TerraMaster T6-423

    Qnap TS-853A

    Qnap TS-451+

    Qnap TS-259 Pro+

Participate now!

Don’t have an account yet? Register yourself now and be a part of our community!

Register Yourself Login