XCP-ng 8.2 updates announcements and testing

stormi

Was it the only update applied? Is the stunnel service running?

Oh, I also read that you can't connect as root.

brezlord

stormi Yes only update.

stormi

Were you using sudo on the host before?

brezlord

stormi No just default install

stormi

I hardly see a cause-effect link between the update and the issues (both SSH and XAPI not responding anymore?), but computers are full of surprises.

Did you change the firewall configuration? Could the IP address have changed or the same be attributed to another device?

brezlord

stormi Done nothing but apply the update through XO web console. I have yanked the plug and making sure it actually reboots.

brezlord

That fixed it I can login via ssh with root and XO sees the host.

stormi

Maybe it was still rebooting, stuck on the shutdown phase, waiting for some kind of I/O or something. This would explain why it didn't respond.

olivierlambert

I concur. If the shutdown process is stuck somewhere (eg an NFS share), you can't connect at all (connection refused in SSH, no XAPI connection) and it can stays like this for a while.

brezlord

stormi It was not responding after the update from XO. I could long in via ssh and restarted the tool stack but this did not help XO still could not login. I issued a reboot command via ssh which dropped the ssh session and the host did not reboot most likely due to running VMs. I then yanked the power and the host rebooted and everything is working as it should. The update definitely caused the issue.

olivierlambert

I did update my home lab without any issue, before, during and after the update (I did test just after the update without any reboot).

brezlord

olivierlambert This is my home lab as well running on a small form factor PC with an Intel(R) Core(TM) i7-10700 CPU @ 2.90GHz

stormi

XO restarts the toolstack after installing updates. So what first went wrong is this: it couldn't restart. You can't say for sure it's caused by the update, because there are many other reasons that can make this fail. The logs would tell. For example a known bug, being fixed, in xenospd-xc, which makes it unable to restart when something specific happened to VM metadata.

So we'll keep an eye and ears open for any other occurrence of this issue in relation with the update, but I still think there's little chance an update of sudo would itself cause this.

We'll do a few additional tests to see if we can reproduce.

brezlord

stormi If you direct me to where the log you need are I can provide them.

stormi

/var/log/xensource.log and /var/log/daemon.log would be the first ones to check.

brezlord

stormi you can download the logs here.

stormi

I see this in daemon.log, a message from systemd attempting to shut the system down:

Feb  1 22:11:35 xcp-ng-01 systemd[1]: Unmounted /run/sr-mount/5f5a9343-b95a-9bfa-bd3a-bc30d7368058.
Feb  1 22:11:35 xcp-ng-01 systemd[1]: Failed to propagate agent release message: Transport endpoint is not connected
Feb  1 22:11:35 xcp-ng-01 systemd[1]: Failed to propagate agent release message: Transport endpoint is not connected
Feb  1 22:11:35 xcp-ng-01 systemd[1]: Failed to propagate agent release message: Transport endpoint is not connected
Feb  1 22:11:35 xcp-ng-01 systemd[1]: Failed to propagate agent release message: Transport endpoint is not connected
Feb  1 22:11:35 xcp-ng-01 systemd[1]: Failed to propagate agent release message: Transport endpoint is not connected

There definitely was a network mountpoint (a NFS SR) which was not connected anymore. This explains the long reboot time.

Going up the logs, I see this:

Feb  1 22:01:43 xcp-ng-01 systemd[1]: xenopsd-xc.service: main process exited, code=exited, status=2/INVALIDARGUMENT
Feb  1 22:01:43 xcp-ng-01 systemd[1]: Unit xenopsd-xc.service entered failed state.
Feb  1 22:01:43 xcp-ng-01 systemd[1]: xenopsd-xc.service failed.

This explains the failed XAPI restart and is likely the known issue with xenopsd I mentioned above.

So, if I'm not wrong, it's good news:

The xenospd issue is known and a fix is on its way and usually disappears after a reboot.
The update itself probably didn't cause your issues.

olivierlambert

haha my "gut feeling" approved

NielsH

gduperrey said in Updates announcements and testing:

New Update Candidates (xen, xapi, templates)

Xen: Enable AVX-512 by default for EPYC Zen4 (Genoa)

Xapi: Redirect http requests on the host webpage to https by default.

Guest templates:

Add the following templates: RHEL 9, AlmaLinux 9, Rocky Linux 9, CentOS Stream 8 & 9, Oracle Linux 9

Test on XCP-ng 8.2

From an up to date host:

For Xen, Xapi and Guest templates:
yum clean metadata --enablerepo=xcp-ng-testing
yum update xen-dom0-libs xen-dom0-tools xen-hypervisor xen-libs xen-tools xapi-core xapi-tests xapi-xe guest-templates-json guest-templates-json-data-linux guest-templates-json-data-other guest-templates-json-data-windows --enablerepo=xcp-ng-testing
reboot
Versions:

xen-*: 4.13.4-9.29.1.xcpng8.2

xapi-*: 1.249.26-2.2.xcpng8.2

guest-templates-json-*: 1.9.6-1.2.xcpng8.2

What to test

Normal use and anything else you want to test. The closer to your actual use of XCP-ng, the better.

Test window before official release of the updates

No precise ETA, but the sooner the feedback the better.

Hello,

Is there any update on the ETA for this? Since it has been almost a month. We'll do the xcp-ng updates again soon and if these patches are close to release we will wait for them to prevent double work.

Cheers,
Niels

stormi

NielsH We'll wait for the next security update, to ship them together. When exactly security updates are released can't always be predicted or disclosed.