XCP-ng 8.3 betas and RCs feedback 🚀

stormi

@brezlord If you haven't reinstalled it yet, yes, we could use more information about the host, the PCI passthrough setup on it, how the error is triggered exactly, and various logs. And/or a support tunnel to have a look by ourselves.

BenjiReis

@brezlord Can you sahre the value you put in the xen-cmdline on 8.2?

I think the XAPI awaits this format: xen-pciback.hide=(pci-id1)(pci-id2)... and will fail the value doesn't match this format.

brezlord

@BenjiReis Sorry I have destroyed the host and loaded a fresh install of 8.3. If you'd like I can install 8.2 on a host and pass through a pci device via the cmd then upgrade to 8.3 and see if the error is reproducible.

BenjiReis

@brezlord If it's not to much a bother that would be great yeah.
Comparing the xen-cmdline when doing the passthrough manually on 8.2 VS how it looks on 8.3 and when done via the XAPI.

stormi

I'm pushed a new set of updates, hopefully the last one before the release of XCP-ng 8.3.0 RC2, which itself should be followed shortly by the release of XCP-ng 8.3.0 itself.

Main packages

intel-microcode-20240717-1.xcpng8.3: updated microcode for Intel vulnerabilities
sm-3.2.3-1.4.xcpng8.3: fix the cause of a warning displayed during update, and restore changes that we had removed because they were suspected to cause issues in some cases with iSCSI, but revealed themselves necessary to support another kind of setup.
vim-7.4.629-8.el7_9 (which provides vim-minimal, installed by default): bugfixes and security fixes
xapi-24.19.2-1.3.xcpng8.3: Fixes an issue where new fields in XAPI DB for certificate fingerprints were not populated, which under some circumstances caused joining new hosts to a pool fail.
xcp-ng-release-8.3.0-28:
- Update repository files for CentOS and EPEL.
- Point at repo.vates.tech for CentOS since mirrorlist.centos.org was cut
- Add "(EOL)" to repo descriptions for EOL repos
- Drop unused repos

Optional package

kernel-alt-4.19.316+1-2.xcpng8.3: Enable CONFIG_X86_AMD_PLATFORM_DEVICE in kernel config
ldns-1.7.0-21.xcpng8.3 + libreswan-4.12-2.3.1.xcpng8.3: security updates

Ajmind 0

On my 8.3 test pool I am unable to create SR ISO libary (SMB/cifs).

On my production pools with XCP-NG 8.2.x it does work as expected.

olivierlambert

Can you provide a bit more details? Errors and such.

Ajmind 0

@olivierlambert
yes I can

on dom0:

mount -t cifs --verbose -o username=admin,password=******** //192.168.1.202/iso /mnt/test
mount.cifs kernel mount options: ip=192.168.1.202,unc=\\192.168.1.202\iso,user=admin,pass=********
mount error(112): Host is down

XCP-NG Center:

Creating ISO SR 'SMB ISO library' on 'IT1HALIZARD-TEST1'
Unable to mount the directory specified in device configuration request	it1xcp-ng-test-slave1	Sep 5, 2024 12:59 PM	Dismiss

XO from Sources:

sr.createIso
{
  "host": "c1f34b07-c4dc-4584-8bc0-a01bcec81c5b",
  "nameLabel": "test",
  "nameDescription": "test",
  "path": "\\\\192.168.1.202\\public\\iso",
  "type": "smb",
  "user": "admin",
  "password": "* obfuscated *"
}
{
  "code": "SR_BACKEND_FAILURE_222",
  "params": [
    "",
    "Could not mount the directory specified in Device Configuration [opterr=mount error(112): Host is down
Refer to the mount.cifs(8) manual page (e.g. man mount.cifs)]",
    ""
  ],
  "call": {
    "method": "SR.create",
    "params": [
      "OpaqueRef:cee521f7-dc00-5d91-1499-6143d2fd0040",
      {
        "type": "cifs",
        "username": "admin",
        "cifspassword": "* obfuscated *",
        "location": "//192.168.1.202/public/iso"
      },
      0,
      "test",
      "test",
      "iso",
      "iso",
      true,
      {}
    ]
  },
  "message": "SR_BACKEND_FAILURE_222(, Could not mount the directory specified in Device Configuration [opterr=mount error(112): Host is down
Refer to the mount.cifs(8) manual page (e.g. man mount.cifs)], )",
  "name": "XapiError",
  "stack": "XapiError: SR_BACKEND_FAILURE_222(, Could not mount the directory specified in Device Configuration [opterr=mount error(112): Host is down
Refer to the mount.cifs(8) manual page (e.g. man mount.cifs)], )
    at Function.wrap (file:///srv/xen-orchestra/packages/xen-api/_XapiError.mjs:16:12)
    at file:///srv/xen-orchestra/packages/xen-api/transports/json-rpc.mjs:38:21
    at runNextTicks (node:internal/process/task_queues:60:5)
    at processImmediate (node:internal/timers:447:9)
    at process.callbackTrampoline (node:internal/async_hooks:128:17)"
}

The SR is mounted on our production pools.

olivierlambert

Well, clearly, as it comes from 2 different systems (XOA and XCP-ng) you have an issue to reach this IP address. It's simply not up and running or not connected to this IP (or blocked) from XCP-ng/XO perspective.

steff22

Don't know if this is the right place for this forum post.

But is it possible to pass through Usb keyboard, mouse or bluetooth adapter to a vm? None of these things appear in Xen Orchestra.

Smart house z-wave usb stick appears right away in Xen Orchestra

Or must I buy a usb to pcie card and passthrough this to vm to make it work

olivierlambert

I think yes, but probably need some tinkering in the USB script filtering some devices for security reasons. It's documented: https://docs.xcp-ng.org/compute/#passing-through-keyboards-and-mice

steff22

@olivierlambert ok thanks

Ajmind 0

@olivierlambert

you are right

The standard gateway has blocked access of NAS.

Could be closed.

CJ

Is there a way to track the progress of applying patches to a host? I'm updating the master of one of my pools and it dropped out of XO and hasn't come back yet. I don't recall applying patches to take this long normally.

olivierlambert

Yes, you can check the yum history for example, eg yum history list

CJ

@olivierlambert I'm seeing 27 EE which seems to indicate that it worked but there was an error?

I have had XO display errors when performing updates but there's currently nothing in the logs and the pool still isn't showing up.

Using yum history info it appears to output EE whenever the scriptlet has an output. In this case it returned ok and I also have lines regarding disabling the repos.

Now I'm getting UND_ERR_SOCKET as an error on the Servers page. If I disable and then enable the server, I get write ECONNRESET.

redakula

@stormi said in XCP-ng 8.3 betas and RCs feedback :

Optional package

kernel-alt-4.19.316+1-2.xcpng8.3: Enable CONFIG_X86_AMD_PLATFORM_DEVICE in kernel config

I see some mentions of this option in relation to fan speeds.

But what is the status on AMD support in general?
I have a recent AMD platform, and from what i can see a lot has been added for AMD support in the past few kernel versions.
Particularly in relation to power management.

I did a purely unscientific test and booted a ubuntu 24.04 live cd on my XCP-NG server, and the power consumption was 10-15% lower at similar load level, compared to XCP-NG.

Is the hardware support for e.g. power management handled by the kernel in dom0 or by Xen?

What are the current plans for a more recent kernel for dom0 - i have not seen anything on kernel version plans for a long time?

Keep up the good work

olivierlambert

@redakula said in XCP-ng 8.3 betas and RCs feedback :

Is the hardware support for e.g. power management handled by the kernel in dom0 or by Xen?

It's a very complex question where there's no definite answer. But to answer your question, yes, XCP-ng 9 will have a recent kernel, providing many benefits for I/O and potentially power usage.

stormi

@CJ it may be that XAPI or stunnel failed to fully start.

Check:

systemctl status xapi-wait-init-complete.service
systemctl status stunnel@xapi.service

CJ

@stormi wait init shows exited with a success status. stunnel shows active and running.

I disabled the server and enabled it again and this time it connected without an error. However, it lists as unknown pool and doesn't show in any of the host/pool/vm pages.

Also, when I go on the console, it states that the host is already in maintenance mode. Selecting Enter/Exit Maint Mode results in the following.

(''NoneType' object has no attribute 'opaqueRef'')