Looking for Storage Drives Recommendations / Advice on Configuration and File System to use

  • I am currently putting together an OMV Server for a friend to be primarily used for his own cloud storage using Nextcloud AIO Docker and some SMB Shares... Mostly storing Photos, Documents, etc from their phones & computers and maybe backups of their computers... I am repurposing an old Desktop Computer I custom built for him sever years ago that he retired recently in favor of a laptop as his primary computer... The only thing I need to add to the system is Storage... I have a budge of $500 or less... I only have room and available SATA Ports for 4 Storage Drives... I am currently looking at Seagate Exos Factory Re-certified Drives from serverpartdeals.com ...


    My main question I have and would like to get some advice and recommendations on is the setup and configuration of the Data Storage Drives... I want to setup at least 2-3 drives for the primary storage pool with (1) larger drive for a backup of the primary storage pool... At some point in the future an offsite backup plan may be implemented or even another Network Attached Device for backups located in another building on the same network...


    One option I am considering is 3x 8TB drives to use in either a EXT4 MergerFS/Snapraid Single Parity setup, or a ZFS Raidz1, both giving me 16TB Storage, and then 1x 18TB EXT4 or XFS Drive for the Backup Drive with regular scheduled backups of the 16TB Data Pool using something like rsync, rsnapshot, or borgbackup...


    A 2nd option I am looking at is 2x 12TB drives to use in a Raid1 Mirrored Array using either EXT4/mdadm, BTRFS, or ZFS, and 1x 18TB drive for the Backup of the mirrored array like above...


    I don't anticipate their storage needs ever exceeding either of these two options anytime in the near future... He is just wanting to get away from the likes of Microsoft, Google, and Apple having control of his data... His family has never came close to using all the 1TB of Onedrive storage per person with their Microsoft 365 subscription and I don't think any of them has ever added more than 50GB of iCloud storage for their photos and such on their phones...


    Current Prices of the drives I am looking at:

    8TB Drives = $99 / each * 3

    12TB Drives = $115 / each * 2

    18TB Drive = $185


    Any Thoughts, Suggestions, or Better Alternatives to anything I have described above would be appreciated...


    Thanks...

    OMV System Information:

    Version: 7.4.11-1 (Sandworm)

    Processor: Intel(R) Core(TM) i5-8500 CPU @ 3.00GHz

    Installed Memory: 16GB

    Kernel: Linux 6.5.13-6-pve

    Edited 2 times, last by bryanc1968 ().

  • bryanc1968

    Changed the title of the thread from “Looking for Storage Drives Configuration / Recommendations / Advice” to “Looking for Storage Drives Recommendations / Advice on Configuration and File System to use”.
  • I am currently looking at Segate Exos Factory Re-certified Drives...

    Realize that there is no way to know the actual condition or age of these drives. Despite popular belief, their SMART stat's and age can be reset. Realize that going in.

    Another option I am looking at is (2) 12TB drives to use in a Raid1 Mirrored Array using either mdadm, btrfs, or ZFS and the 18TB drive for the Backup of the mirrored array like above...

    Full disclosure, after using it for years,I have a preference for ZFS:


    The above is the route I would go. ZFS in the equivalent of RAID1 (a zmirror) provides guaranteed data integrity. That means bitrot, file corruption, mag media bit flips and similar data degrading events are automatically fixed during a scrub. Further (with an eye on used drives), replacing a failed or failing drive in a zmirror is straight forward. It is a CLI operation, but it's not complicated.

    When automated snapshots are added, things become a bit more bullet proof. Purging ransomware attacks and similar incidents becomes a matter of rolling back to a previous snapshot. Having full backup (I would use EXT4 on the 18TB) will help to mitigate the risk of using refurb drives.

    With this setup, there will be a lot of redundancy and restoration options available, in the event of a drive issue.

  • Full disclosure, after using it for years,I have a preference for ZFS:


    The above is the route I would go. ZFS in the equivalent of RAID1 (a zmirror) provides guaranteed data integrity. That means bitrot, file corruption, mag media bit flips and similar data degrading events are automatically fixed during a scrub. Further (with an eye on used drives), replacing a failed or failing drive in a zmirror is straight forward. It is a CLI operation, but it's not complicated.

    When automated snapshots are added, things become a bit more bullet proof. Purging ransomware attacks and similar incidents becomes a matter of rolling back to a previous snapshot. Having full backup (I would use EXT4 on the 18TB) will help to mitigate the risk of using refurb drives.

    With this setup, there will be a lot of redundancy and restoration options available, in the event of a drive issue.

    Thanks for your input... I have been leaning toward the 1st option with mergerfs/snapraid since I am currently using that in my own setup and am already familiar with it... But I have a set of mixed size drives that i had laying around that I saved from old retired systems, with the largest being my 2TB Snapraid Parity Drive... So Mergerfs/snapraid was a good fit for that... I hope to one day soon be able to buy some larger, same size drives when funds become available...


    But since I am building this system for someone else and using same size drives, except for the larger drive for a redundant backup, maybe I need to consider the Mirrored Array approach... I will be the one maintaining and supporting this system, and doing most of it remotely after getting it all up and running... So I wonder which option would be the easiest and safest to maintain and troubleshoot should something go wrong..?

    OMV System Information:

    Version: 7.4.11-1 (Sandworm)

    Processor: Intel(R) Core(TM) i5-8500 CPU @ 3.00GHz

    Installed Memory: 16GB

    Kernel: Linux 6.5.13-6-pve

    • Official Post

    I have been leaning toward the 1st option with mergerfs/snapraid since I am currently using that in my own setup and am already familiar with it... But I have a set of mixed size drives that i had laying around that I saved from old retired systems, with the largest being my 2TB Snapraid Parity Drive... So Mergerfs/snapraid was a good fit for that... I hope to one day soon be able to buy some larger, same size drives when funds become available...


    But since I am building this system for someone else and using same size drives, except for the larger drive for a redundant backup, maybe I need to consider the Mirrored Array approach..

    I have both setups:
    - ZFS in mirrors (also I have one server in RAIDZ1 - but that's not relevant to this thread)
    - mergerfs and SnapRAID with dissimilar sized drives.

    While mergerfs/SnapRAID works well over USB connections (with SBC's and other hardware platforms), SnapRAID has some restoration limitations that some users may not be aware of.

    Chiefly, they are:
    - There is one (a single) level of restoration available. Files, folders or a drive may be restored as of their state after the last sync operation. Any file or drive state prior to the last sync is not restoreable.
    - If there's a large delete, where files and folders are deleted across more than one data drive, restoration of all of the deleted files and folders may not be possible. The same applies to rebuilding a failed drive. While the drive will be restored, unrecoverable file level errors may result. This is because existing parity calculations require file data on member drives to be in place, to restore a specific missing file. To prevent issues with restoring a drive, a large delete should be followed immediately by a sync.

    ZFS doesn't have those limitations. With snapshots, preexisting files states can be restored for up to a year later (or even longer if custom configured). Drive restorations are more straight forward. When a drive is lost, the mirrored drive allows access to data until the failed drive can be replaced. (Even BTRFS may have issues with data being available, in a failed RAID1 / mirror setup.) Information on setting up simple fully automated and self purging snapshots is available -> here.

    On the other side of the coin, just as it was with mergerfs/SnapRAID, there are things to learn regarding ZFS. Sometimes going with what one is comfortable with may rule the day. This is especially true if you have already done restoration operations with mergerfs/SnapRAID. (If not restoration information is available here ->SnapRAID -> mergerfs.)

    I will be the one maintaining and supporting this system, and doing most of it remotely after getting it all up and running... So I wonder which option would be the easiest and safest to maintain and troubleshoot should something go wrong..?

    Well, it depends on "what went wrong". Obviously, if you have to swap a data drive, you're going to have to visit. With both packages, file restorations can be done remotely, on the CLI, "but" that would take specific information (which files) from someone who is on site. Given the number of potential variables, the only clear answer here is, "it depends".

    Personally, as already mentioned, my preference would be ZFS. It's very stable, well documented and relatively free of unexpected surprises.

    In the bottom line, remote admin requires some thought and planning. It's easy to remote into a server and cause a problem that wouldn't, otherwise, exist.
    ____________________________________

    Lastly, if you're not going to set up dockers or other server add-on's, that would reside on the boot drive, I'd recommend booting from a thumbdrive. The reason is, cloning thumbdrives is dirt easy. With a cloned thumbdrive on top of the server, if the boot drive fails, all someone on site has to do is:

    - Shutdown.
    - Swap out thumbdrives.
    - Boot Up.

    And,, they're back in business.
    That's as easy as it gets and it wouldn't require a site visit.

    More info on that is -> here.

  • I will be installing some Docker Containers... Nextcloud AIO Docker and Nginx Proxy manager... Maybe Home Assistant later on... The OS is installed on a small SSD with a 2nd 500GB SSD for the Docker Stuff... the Nextcloud Data and any SMB Shares will be stored on the HDDs I am looking to buy...

    OMV System Information:

    Version: 7.4.11-1 (Sandworm)

    Processor: Intel(R) Core(TM) i5-8500 CPU @ 3.00GHz

    Installed Memory: 16GB

    Kernel: Linux 6.5.13-6-pve

    • Official Post

    If you off load Docker containers to an SSD, (easy enough to do) you'd be able to boot from a thumbdrive. I've been using thumbdrives on all my servers, with the exception of one, for several years. I've found that a good quality drive will last 5 or more years.


    For the shear simplicity of maintaining OS backup and especially for being able to do guide a remote user through a quick server restore, It's something to think about.

  • Well as I mentioned in my original post I am building this for a close friend and this server will be located at his farm only about 35 min drive from me... Plus he splits his time between his farm house and his beach house which is currently his primary residence located in another state... So I would most likely be making an onsite visit for any hardware issue anyway... He is not very tech savvy at all... I have been his IT guy ever since we became friends years ago from doing several IT related jobs for him and his family... During that time we became close friends and even go to his beach house on vacation at least once or twice a year... He is setting up his farm to eventually become his primary residence and as a place to go if things ever take to crap in the world or country so he can raise his own live stock and crops for food and that sort of thing... He has had me set up WiFi all over his farm, security cameras, some automation in his greenhouse, etc... He currently has 1GB se-metrical fiber internet service and is talking about adding Star-link satellite internet as a backup... He has a large backup power generator and is talking about adding Solar Panels to the roof of his barn as well... He has partitioned off a room in his barn that we are setting up his network infrastructure and server in... I pretty much handle all his IT stuff for him...

    OMV System Information:

    Version: 7.4.11-1 (Sandworm)

    Processor: Intel(R) Core(TM) i5-8500 CPU @ 3.00GHz

    Installed Memory: 16GB

    Kernel: Linux 6.5.13-6-pve

    • Official Post

    I've found that a good quality drive will last 5 or more years.

    Note: To get the good life from a thumbdrive (5 years'ish), you'd need the -> flashmemory plugin.

    He is setting up his farm to eventually become his primary residence and as a place to go if things ever take to crap in the world or country so he can raise his own live stock and crops for food and that sort of thing...

    A lot of people buy a farm for a bug out location. While it's always a good idea to get out of high population density areas (they're death traps if "it" hits the fan), he'd better have about 2 to 3 years of shelf life food stores to get started. With sufficient food stores, he'd have enough time to figure out what he needs to do, to feed himself and his family without staving in the interim.

    Farming ("subsistence gardening" is probably a better term for the scale I'm thinking of) and animal husbandry is nowhere near as easy as people think it is. I grew up in rural environs and I currently live at the edge of the grid, so I know a few things about this. There's a LOT to learn and it's, arguably, far more important than technology (solar panels, batteries, inverters, generators, etc.). Having power is a good thing but eating is far more important.

    When it comes to animals, chickens are easy but they have to be fed which means growing corn and peas. A Dexter cow, in milk, (preferably with a calf) is a good investment. Like the Texas Long Horn, but far better suited to colder weather, the Dexter can get by on poor pastures. They do well on grass and hay alone. But, depending on how much land he has, rotational grazing might be required.

    In any case growing vegetables, and varieties of dried beans, without pesticides and externally purchased soil inputs (things like chemical fertilizers, lime, pesticides etc.) takes some dedication.

    Again, I applaud his efforts, there's a LOT involved in becoming more self sufficient. If he really wants to achieve something close to self sufficiency, he should move to the farm (100% of the time) and get to it.

Participate now!

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