OMV 3 for ODROID-XU4/HC1/HC2/MC1

    • Offizieller Beitrag

    It seems that mv doesn't rewrite the inode, but instead performs copy/remove which as you understand is very slow.

    It looks like you are copying from one nfs export folder to another?? If so, those would be two different bind mounts and Linux will treat them as two separate filesystems and won't rewrite inodes.

    omv 7.0-32 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.9 | compose 7.0.9 | cputemp 7.0 | mergerfs 7.0.3


    omv-extras.org plugins source code and issue tracker - github


    Please try ctrl-shift-R and read this before posting a question.

    Please put your OMV system details in your signature.
    Please don't PM for support... Too many PMs!

  • tkaiser: armbianmonitor -u generates no useful info, pls see pastebin-link.
    armbianmonitor -u output


    It looks like you are copying from one nfs export folder to another?? If so, those would be two different bind mounts and Linux will treat them as two separate filesystems and won't rewrite inodes.

    yepp, the disc was mounted on export as default after installning and I haven't given it any thought. And if it is as you say, remounting the drive to /media should remedy my problem?

    • Offizieller Beitrag

    If you move files from directories within the same mountpoint, the move should be very fast.

    omv 7.0-32 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.9 | compose 7.0.9 | cputemp 7.0 | mergerfs 7.0.3


    omv-extras.org plugins source code and issue tracker - github


    Please try ctrl-shift-R and read this before posting a question.

    Please put your OMV system details in your signature.
    Please don't PM for support... Too many PMs!

  • Hi again and sorry for a late reply.
    Sure enough moves within nested directories work as intended.
    I'm not sure why the directories are mounted directly under "/export" instead of "/export/UUID/dir" or "/export/disc-label/dir".


    None the less thank you both for your input and hard work!

  • I'm a noob when it comes to OMV, so forgive me.


    I'm looking to build my own NAS. I may go with the XU-4. Just a simple question regarding OMV on the XU4.


    From what I'm read, I just need to burn the image onto an SD card, plug it into the computer, and power it on, wait a few minutes and I'm able to use it right away? Or do I need to plug the computer into a TV and do some config while installing it before I can use it?


    If I can just plug in the SD card, power on the computer and wait a bit before using and no other steps needed, then the XU4 is exactly what I'm looking for.

    • Offizieller Beitrag

    I just need to burn the image onto an SD card, plug it into the computer, and power it on, wait a few minutes and I'm able to use it right away?

    That's it. You need to check your router which IP adress has been assigned to your OMV. Then you can access the GUI using this IP adress. Next is to configure the system.

  • I may go with the XU-4.

    I started with the XU4.
    I want my NAS to be as low power as possible.


    So I hooked an USB3 external 2,5" HD (tried several) to it -> the XU- gave me many grave problems due to power issues from the USB port. I tried the "normal" and the 6A Hardkernel power supply.


    After being available (I bought the XU4 one week too early) I switched to the HC1. This workstm.


    I only use the XU4 now for experiments and backup. In backup mode I have a 6TB 3,5" HD hooked via USB3 with its own power supply. No problems with this setup except for the power requirements and using two power supplies.


    My recommendation: either go with the HC1 or HC2 for your NAS.

  • So I hooked an USB3 external 2,5" HD (tried several) to it -> the XU- gave me many grave problems due to power issues from the USB port.

    Yeah, for whatever reasons on the XU4 the power rails to USB3 ports are responsible for some severe voltage drops under load and while there are a couple of HDD/SSD out there that can cope pretty good with voltages that get as low as 4.5V or even 4.4V there are other HDD brands that start to fail already at 4.75V (but when looking into their specs the term "5V with 5% tolerance" clearly states that!). The HC1 PCB and especially the 12V variant used in HC2 fixed this issue (at the cost of a little higher consumption on the latter since the DC-DC circuitry that generates always stable 5V from the 12V input has only a +90% efficiency and not 100%)

  • As soon as there is a 2 Slot SATA 3 board I will buy it.

    These things exist even without the unnecessary USB3 layer in between and provide much higher (especially random IO) performance. But then you need to look at devices that use 'NAS SoCs' instead of TV box, tablet or smartphone SoCs or you need latest ARM SoC generation that features PCIe (see below).


    "Full SATA performance" (no USB3-SATA!) sorted by port count:

    • Clearfog Base with u-boot tweaks to turn mPCIe slot into mSATA: 2 SATA ports
    • EspressoBin + ASM106x mPCIe card: 3 SATA ports
    • Clearfog Pro with u-boot tweaks to turn mPCIe slots into mSATA: 3 SATA ports
    • Clearfog Base + ASM106x mPCIe card: 3 SATA ports
    • Helios4: 4 SATA ports
    • EspressoBin or Clearfog Base + Marvell 88SE9215 mPCIe card: 5 Ports
    • Clearfog Pro + 2 x Marvell 88SE9215 mPCIe cards: 9 Ports



    All the variants shown above have an own USB3 port in addition to one or multiple SATA ports too! And to be able to use normal SATA disks with M.2 or mSATA ports inexpensive mechanical adapters for 1 or 2 bucks are needed (this picture should give the idea and the thread additional information)


    (Not so) surprisingly all these boards are based on Marvell Armada SoCs (that are made for this use case and show high internal IO bandwidth even if their CPU specs look somewhat limited. But even the dual core Armada 3700 at less than 1 GHz clockspeed in the EspressoBin easily outperforms the octa-core SoC in XU4, HC1 or HC2 since the latter is an old smartphone SoC and lacks internal IO capabilties)


    In 2018 we'll see a lot new inexpensive and low power ARM boards that feature one to four PCIe 2.x lanes so by combining them with appropriate SATA controllers you can get a bunch of performant SATA ports (but each of these controllers need energy on their own so you could end up in consumption regions where you could also simply have a look at low power x86 boards with a bunch of SATA ports on it)

  • Thank you very much tkaiser - the EspressoBin looks like a suiting board for my needs. Just have to look where I order it as I have not seen any retailer, yet. (Just looked for a few minutes though :)


    Or I´ll just wait for the boards to come in 2018. As I still have my XU4 up and running pretty well!

  • EspressoBin looks like a suiting board for my needs. Just have to look where I order it

    Kinda off-topic in this thread but anyway. Beware of ASM106x mPCIe cards where the connectors look like this since they won't work with EspressoBin since the Ethernet jacks are in the way. You need a card with upright SATA connectors. And the ASM1062 card I bought is something that goes to the bin soon (or I find cables with better fitting contacts) since I constantly get SATA transmission errors with this card just after a few cable matings (well, 'buy cheap, buy twice' as expected. Bought the cheapest card I could find in Germany at notebooksbilliger.de since I could pick it up there without shipping costs)

  • Yeah, for whatever reasons on the XU4 the power rails to USB3 ports are responsible for some severe voltage drops under load and while there are a couple of HDD/SSD out there that can cope pretty good with voltages that get as low as 4.5V or even 4.4V there are other HDD brands that start to fail already at 4.75V (but when looking into their specs the term "5V with 5% tolerance" clearly states that!)

    Would having an a USB 3 hard drive with it's own power supply solve this issue? That seems to be the main issue I see with external drives. USB powered ones often fail, where one with it's own power supply will not.

  • Would having an a USB 3 hard drive with it's own power supply solve this issue?

    This specific issue yes. But with XU4 (or any other boards that provides USB3 SuperSpeed via the shitty USB3-A receptacle) a lot of other problem sources still exist (though with my single XU4 I had the most cable/contact related problems -- reported that to TL Lim (Pine64/ROCK64 founder) pretty early and they ensured that the USB3-A receptacle on ROCK64 released later fits very tightly so contact problems are less of an issue).


    Then there's the chance to attach a crappy USB3 disk enclosure to an USB3 port (eg. all Seagate or WD disks or those with Norelsys chipsets or those with older Genesys Logic chipsets) and you still run into problems. All these issues do not exist on HC1, HC2 or soon to be released Transformer since cable/contact issues can't happen there and the used chipset is a good one.


    You might want to read through this whole post (all links included): https://forum.armbian.com/topi…findComment&comment=32340


    And it should be noted that all official OMV images for ARM devices try to automatically UAS blacklist all Norelsys, Seagate and WD disk enclosures that are attached (might require at least two reboots with disk enclosure connected for changes to take effect). This does also apply to those for ROCK64 and Transformer from ayufan.

  • And it should be noted that all official OMV images for ARM devices try to automatically UAS blacklist all Norelsys, Seagate and WD disk enclosures that are attached (might require at least two reboots with disk enclosure connected for changes to take effect). This does also apply to those for ROCK64 and Transformer from ayufan

    Good thing I don't have any WD, Seagate or Norelsys enclosures. I have WD and Seagate disks, and Vantec enclosures.

    • Offizieller Beitrag

    I installed OMV3 on Odroid HC2 and made the upgrade to OMV4 with omv-release-upgrade. Now I noticed that armbian.list is still on jessie.


    deb http://apt.armbian.com jessie main jessie-utils jessie-desktop


    Should this manually be changed to stretch?

  • Now I noticed that armbian.list is still on jessie... Should this manually be changed to stretch?

    Short answer: doesn't really matter but it also doesn't hurt to do a


    Code
    sudo sed -i 's/jessie/stretch/' /etc/apt/sources.list.d/armbian.list && sudo apt update

    Long answer: Armbian is not an own distribution but a build system that debootstraps either Debian or Ubuntu based OS images. Almost all packages are from the respective upstream repositories (Debian/Ubuntu and OMV in our case) and Armbian itself only adds a few utility packages and most importantly packages also kernel and u-boot upgrades in a Debian compliant way (that's the main entry above).


    When looking at the origin of these packages http://apt.armbian.com/pool/ one can see that currently all those few additional jessie-utils or stretch-utils packages share same sources and version and it's very unlikely that they're ever used by an OMV user (only exception are Armbian's own hostapd packages in case someone wants to do a wireless hotspot with his ARM board which IMO is rather pointless anyway without a powerful external USB dongle since onboard Wi-Fi on all currently supported devices is slow and crappy)

Jetzt mitmachen!

Sie haben noch kein Benutzerkonto auf unserer Seite? Registrieren Sie sich kostenlos und nehmen Sie an unserer Community teil!