XCP-ng 8.0.0 Beta now available!



  • I've taken a loot at the USB-stick and found grub config files. What I found was a maxmem of 8192 MB already configured.
    I changed it to 2048M (I don't see any reason why the setup should need more than a few hundret MBs) aaaand it worked.
    As it affected 2 different systems (opsi bootimage for PXE install and CH/XCP) I'm very sure it's a kernel bug!
    Though I can't say which kernels are affected, except that old kernels work.
    When I have time, I may test it on our Epyc server, but I guess that it
    s more about the embedded Vega GPU and the shared memory.

    https://forum.opsi.org/viewtopic.php?f=8&t=10611 Et al.
    I'm not the first one, having trouble with Ryzen/Vega APU and the Linux kernel.

    tl;dr:
    As a workaround, I recommend setting the mem-limit for install stick generally to 2048M, as there shouldn't be any benefit for more anyways and it pushes compatibility.

    Yet have to upgrade the system and see if XCP-ng 8 Beta boots properly (had it still on CH 8.0 for test).


  • XCP-ng Team

    Indeed, I suppose it's related to Ryzen+Vega APU, because I can't reproduce on EPYC nor Ryzen without APUs.

    Regarding making the chance upstream, I'm not entirely confident to make the modification, because it's hard to evaluate the impact (our RC is pretty close now). But it's up to @stormi to decide 🙂


  • XCP-ng Team

    At this stage, I'd rather document the issue to help people workaround it than change the default values inherited from Citrix (who tested the installer on a lot of hardware). I also suggest to open an issue at https://github.com/xcp-ng/xcp/issues so that other users that would meet the same issue could share their experience and let us try to estimate whether it's a widely spread issue or something very specific to some hardware.


  • XCP-ng Team

    Looks like the best way to me 🙂



  • @cg said in XCP-ng 8.0.0 Beta now available!:

    https://forum.opsi.org/viewtopic.php?f=8&t=10611 Et al.

    I created an issue, IMHO it would be good if the problem itself makes it to upstream to either bump someone to fix it or get feedback that problem is known/fixed in version X.Y.Z.

    For reference:
    https://github.com/xcp-ng/xcp/issues/206


  • XCP-ng Team

    @cg said in XCP-ng 8.0.0 Beta now available!:

    I created an issue, IMHO it would be good if the problem itself makes it to upstream to either bump someone to fix it or get feedback that problem is known/fixed in version X.Y.Z.

    Thanks for creating the issue.

    You could test with Citrix Hypervisor and then report it at https://bugs.xenserver.org
    Maybe they'll want to investigate even if the hardware is not officially supported. If someone can reproduce the issue on supported hardware, that would make things easier.



  • @stormi said in XCP-ng 8.0.0 Beta now available!:

    You could test with Citrix Hypervisor and then report it at https://bugs.xenserver.org
    Maybe they'll want to investigate even if the hardware is not officially supported. If someone can reproduce the issue on supported hardware, that would make things easier.

    Though I doubt it was woth the effort, as the support over there is well known (especially for not supported configurations etc.) I did it - for the reason of trying to improve stuff (and giving back to OSS community):
    https://bugs.xenserver.org/projects/XSO/issues/XSO-955



  • I just did the upgrade from 7.6 to 8.0.0 beta via USB-stick (with workaround): I successfully upgraded the existing installation including the softwareraid on the first 2 SSDs.

    # cat /proc/mdstat 
    Personalities : [raid1] 
    md127 : active raid1 sda[0] sdb[1]
          117220736 blocks super 1.0 [2/2] [UU]
          bitmap: 1/1 pages [4KB], 65536KB chunk
    

    Everythings fine, even RAID survived. ZFS stuff will come later. Would be good if that would find a way into XCP-ng Center (is it in XOA yet?) for easier management. Likewise for Ext4.


  • XCP-ng Team

    Find a way to what?



  • New SR -> Type -> ZFS... creatig a new SR via GUI/XCP-ng Center.


  • XCP-ng Team

    You should really keep an eye on XO dev 😉

    https://github.com/vatesfr/xen-orchestra/pull/4266

    It's coming this week 🙂



  • My time is (sadly) limited, I can't wacht/monitor everything.
    XCP-ng Center? 😉


  • XCP-ng Team

    XCP-ng Center is only community maintained, so feel free to contribute if you want that inside 🙂



  • XCP-ng already ate a bunch of time, and I'm sure it'll continue for testing etc
    I'm Sysadmin with very limited coding ability - I'm definately out for that.

    I guess it would need someone being good at C#.NET. Also XAPI would probably need to know about it (what may already be, as you're implementing it into XOA)...


  • XCP-ng Team

    Clearly, adding feature in XCP-ng Center would require a lot of contributions, that's why I don't think it's the future of XCP-ng "client". Having one great client allows to focus all effort instead of "spreading" the thin capabilities on various similar "ways" to administrate it.


  • XCP-ng Center Team

    Yes, this seems to be the case 😕 But XCP-ng Center will remain as it is, at least.


  • XCP-ng Team

    I think it's a good thing to focus on one "best" tier-one solution, especially on a "small" community like us 🙂



  • HP DL360 v7 with (2) Intel E5649 working good so far. Just finished installation and yum updates and everything came right up, as expected.



  • As a sidenote for everyone interested into ZFS: Look at the release notes:
    https://github.com/zfsonlinux/zfs/releases

    Especially for (my recommended to consider) dedup:

    Allocation classes #5182 - Allows a pool to include a small number of high-performance SSD devices that are dedicated to storing specific types of frequently accessed blocks (e.g. metadata, DDT data, or small file blocks). A pool can opt-in to this feature by adding a special or dedup top-level device.
    

    That means RAM is no more critical (for people who didn't have enough): These dater can now 'oflload' to SSDs, whereas special storage types like Intel Optane should be perfect for that, as they can demonstrate their advantages (high IOPS at small queues, durability, access-time).
    Optane m.2 cards became pretty affordable - but that feature requires mirroring, so you need 2 of them (I see them priced about 26 € for 16 GB)!

    As the satet It's not only for dedup, it can also cover small file blocks and metadata, lowering IOPS on your storage.

    ALSO: TRIM/discard support!



  • I've installed on a HP ML10 G9 server and all works fine on this server using UEFI based BIOS configuration etc.

    The only issue I have found so far is when imported vmdk images using the wizard, even when I choose to set the BIOS to UEFI after the import and conversation has completed the VM won't boot. After looking at the settings for the newly created VM the BIOS is set to BIOS not UEFI.

    Simple to resolve either be detaching the storage and removing/recreating the virtual machine and re-attaching the original imported storage. Or I am guessing (although haven't tried yet), using the console to set the machine type.

    So possibly an issue with the Import Wizard?

    UPDATED

    I tried the shell commands to change the imported VM to UEFI, and although initially it looks fine when examining the console and seeing that the BIOS setting has updated. When trying to start the VM it won't boot due to an error. So after detaching the disk and deleting the VM I recreated a new VM using UEFI and no disk attached. Then once created I attached the original imported disk and it works fine.

    Also I using VirtualBox to export a VM to OCI Format 1.0 which was setup running as UEFI, after the import into XCP it had changed the BIOS to BIOS not UEFI. So again I followed what I did above and the VM works fine.

    So it does look like an issue with the Import Wizard, although it allows you the option to choose BIOS type it doesn't honour the setting and reverts to standard BIOS.


Log in to reply
 

XCP-ng Pro Support