Building OMV automatically for a bunch of different ARM dev boards

    • Offizieller Beitrag

    omv 7.0.5-1 sandworm | 64 bit | 6.8 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.13 | compose 7.1.4 | k8s 7.1.0-3 | cputemp 7.0.1 | mergerfs 7.0.4


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


    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!

    Zitat von ryecoaaron

    stress -i 1 -t 10 && sleep 1 && smartctl -d sat -a /dev/sdb

    Thank you but the first occurence of '&&' renders the whole test useless (a single '&' is needed to send the 'stress' task to the background and then doing the SMART query in parallel after one second).


    And the interesting output would be the one from 'armbianmonitor -u' directly after since I've heard from one user 'UAS' would cause issues when running SMART queries. I just wanted to test for potential issues with JMS561 with UAS disabled being busy doing IO while a smartctl call is running in parallel. On the other hand this is stuff Hardkernel guys should try out...

    • Offizieller Beitrag
    Zitat von tkaiser

    Thank you but the first occurence of '&&' renders the whole test useless

    I just cut & paste what you had. You must have edited your post after I ran the test. New run below and armbianmonitor (which was run directly after) link is in the code window.


    omv 7.0.5-1 sandworm | 64 bit | 6.8 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.13 | compose 7.1.4 | k8s 7.1.0-3 | cputemp 7.0.1 | mergerfs 7.0.4


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


    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!

    Thank you. The issue I tried to look into with Cloudshell 2 seems to be confirmed in the meantime by user 'Kosmatik' over at Hardkernel forums. When OMV does SMART queries with UAS blacklisted he gets


    Code
    usb 4-1.1: reset SuperSpeed USB device number 3 using xhci-hcd

    messages in dmesg output, with UAS enabled he gets additional messages (as expected) that read

    Code
    sd 0:0:0:0: [sda] tag#0 uas_eh_abort_handler 0 uas-tag 1 inflight: CMD IN
scsi host0: uas_eh_bus_reset_handler start


    (BTW: No idea why Hardkernel folks discuss their quality control problems -- sending out defective USB cables with Cloudshell 2 -- and general Cloudshell 2 SMART issues in a thread dedicated to software issues with 'USB Attached SCSI')

    @ryecoaaron: In preparation of Helios4 arrival I rebuilt OMV images for Clearfog Base and Pro using the same MicroSOM model:


    OMV_3_0_79_Clearfogbase_4.11.4.7z
    OMV_3_0_79_Clearfogpro_4.11.4.7z


    Both can IMO be added to download directory on SF since now based on 4.11.x and next major Armbian update just around the corner. NAS performance is excellent.

    • Offizieller Beitrag
    Zitat von tkaiser

    Both can IMO be added to download directory on SF since now based on 4.11.x and next major Armbian update just around the corner. NAS performance is excellent.

    Uploading now.

    What is the difference between Base and Pro? Can we any of them in a XU4?


    Edit: I found out now. It just looked like a codename for a new release.

    Riddle me this, riddle me that
    Who is afraid of the big, black bat?
    I write on a blog (Romanian mostly)
    Testing (latest) OMV 6.x (HURRAY) on an Intel i5820K NAS (currently with proxmox kernel 6.2)

    Zitat von tmihai20

    What is the difference between Base and Pro?

    Just for others coming accross the same question: Both devices base on the same high-end ARM SoC and MicroSOM as Helios4 and those three are the most powerful ARM based DIY NAS platforms for now (for more performance you need to switch to different CPU architectures or at least more powerful ARMv8 designs like Armada 8040 on MacchiatoBin)


    Background information, performance numbers and pictures: Clearfog Base and Clearfog Pro.


    The Marvell Armada 38x allows for some really cool features. With some u-boot/SERDES configuration voodoo 2 SATA connections can be turned into PCIe or vice versa. As you can see in the picture below I use a dual SATA mPCIe adapter in one slot (configured as PCIe) and a mechanical mSATA to SATA adapter in the other slot (configured as SATA).



    On the PCB back in the M.2 slot (SATA only) another mechanical adapter is used to connect a SSD directly. So my Clearfog Pro features 4 SATA ports just like the Helios4 but since I use an ASM1062 PCIe adapter performance is somewhat affected (you need some quirks and disable NCQ when attaching more than one disk to an ASM1061/ASM1062).


    With Helios4 this does not apply and 4 real SATA 3.0 ports that most probably allow for a total storage bandwidth of more than 2GB/s are accessible.

    Hallo tkaiser,
    i use OMV_3_0_76_Odroidxu4_4.9.28.7z and want to overclock my DRAM (http://odroid.com/dokuwiki/doku.php?id=en:xu4_ram_overclock)
    i update manually from http://beta.armbian.com


    .) kernel to version 4.9.30(image and headers)
    .) uboot
    .) firmware dtb


    then i changed my boot.ini.


    Is this the correct way, because it seems that changes in my boot.ini has no affect


    PS:
    there is an buid error
    http://beta.armbian.com/buildlogs/log.html

    Zitat von proper88

    want to overclock my DRAM

    With focus on OMV (since we're here in OMV forum) this is

    • off-topic here (you might ask in Armbian forum but should be prepared that questions aiming at getting devices more unreliable get ignored)
    • not a good idea at all (XU4 suffers already from a variety of problems both the vendor and ODROID micro community ignore -- I'm referring to the various USB problems and the insanely stupid 'UAS is broken on Linux everywhere' campaign they run over at their forum)
    • useless anyway since what do you want to improve? NAS throughput numbers won't get any better without exchanging the device and technologies (eg. switching to 2.5GbE Ethernet or exchanging unreliable/slow USB with SATA)

    .) sorry for offtopic
    .) i thought USB3(UAS) problems only for ASMEDIA and JMICRON bridges, i dont's use them.
    .) NAS throughput is fantastic(110MB/s Samba)


    but i use snapraid and snapraid suffers from low memory bandwirh so this is the reaseon why i want to overclock my DRAM.

    Zitat von proper88

    i use snapraid and snapraid suffers from low memory bandwirh so this is the reaseon why i want to overclock my DRAM.

    Seriously? You think overclocking DRAM by not even 15 percent should improve $anything? Well, I would believe you're a victim of Hardkernel folks doing evil marketing? In your link odroid.com/dokuwiki/doku.php?id=en:xu4_ram_overclock they only show manipulated and delusive graphs to fool users. Overall AnTuTu numbers 'improve' by laughable 2 percent but they show this impressive bar chart:



    Using 50,000 as base is really a brilliant idea if you want almost no difference showing up as something important. And in reality the graph looks like this (every difference in numbers less than 5 percent can be considered equal or measuring fault, that's one of the most basic rules when doing benchmarks!):



    Please don't get me wrong. Memory settings/performance do affect some NAS/server related tasks a lot but 'overclocking' DRAM is a really stupid idea unless you know the memory operates reliable at this clockspeed. We're talking about 'toy grade' hardware here (the XU4 relies on smartphone technology) and (silent) data corruption is something really horrible since almost always users realize it way too late (when all backups are already corrupted too).


    I would not even think about this unless you invest a huge amount of time, efforts and btrfs scrub tests to get an idea whether you suffer from data corruption or not. But why anyway? Choosing a faster and more reliable different platform is also cheaper if you take the testing efforts into account.


    BTW: Can you do me a favour and provide tinymembench numbers for your XU4 with default settings and in case you succeed with overclocked too (please don't unless... see above ;) ) Takes just 5 minutes and requires just 4 lines copy&paste: https://forum.armbian.com/inde…rts/&page=2#comment-33444

    Zitat von tkaiser

    Oops, it just came to my attention that on most if not all OMV images for A20 boards currently thermal readouts for the CPU are missing. A fix has been posted to Armbian forums (please read carefully and ensure that you do NOT reboot in between!)

    Today Armbian 5.30 has been released (Changelog) and as such updated packages for kernel+headers, u-boot and board support package are available through apt.armbian.com (apt update && apt upgrade).


    So the above workaround isn't necessary any more and a simple 'apt upgrade' should fix the thermal readouts on A20 boards. All mainline kernels are now at 4.11.4 (except ODROID-XU4 since Hardkernel only maintains a 4.9 kernel fork which is now at 4.9.30).

    Zitat von tkaiser

    It seems some caution is necessary wrt 5.30 update (on the other hand we've such reports with every release and many times it's just an SD card dying)

    5.31 fixed the networking issues some board were affected, doing an upgrade is now safe again: https://forum.armbian.com/inde…findComment&comment=33628

    i know but there are some reasons..


    1.) quick install
    2.) sdcard write cycles minimized
    3.) internet connection required
    4.) takes some time to make all "up to date" (especially if you have a sdcard with slow write performance)


    anyway thanks for supporting this odroid UX4 platform.
    your last build is stable and high performance out of the box

Jetzt mitmachen!

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

Benutzerkonto erstellen Anmelden