XCP-ng 8.3 updates announcements and testing

stormi

@gduperrey said in XCP-ng 8.3 updates announcements and testing:

Since it doesn't indicate that an updated 8.2 is actually 8.2.1, only the major version is displayed.

Not a very good example actually (but what's true is that for 8.2 we didn't add a "LTS" mention next to the version number in the system either).

But yes, we decided not to increment the version number for XCP-ng 8.3.0. That breaks compatibility with some third party software because they wouldn't recognize a "8.3.1".

However we're working on a secondary version identifier that would allow you to know your precise patch level.

gduperrey

New security and maintenance update candidate

A new XSA (Xen Security Advisory) was published on the 1st of July, and an update to Xen addresses it. We also publish other non-urgent updates which we had in the pipe for the next release.

Security updates

xen-*:
- Fix XSA-470 - An unprivileged guest can cause a hypervisor crash, causing a Denial of Service (DoS) of the entire host.

Maintenance updates

http-nbd-transfer: moved some logs into debug to reduce log spam
sm:
- XOSTOR: avoid a rare migration error when the GC would run on our migration snapshot
- Use GC daemon code for LINSTOR like other drivers (no changes for users)
  This updated package is already included in the refreshed 8.3 installation ISOs.
xapi:
- Fix remote syslog configuration being broken on updates
- Fix several RRD (stats collection) issues and make the plugins more robust:
  - Cap Derive values within a certain range without making them NaN
  - Use a computed delay time for RRD loop to prevent gaps in metrics collection
  - Avoid duplicating datasources on plugin restore
  - Protect against a resource leak in the plugins
  - Avoid running out of mmap-ed pages in xcp-rrdd-cpu for large numbers of domains
  - Prevent exceptions from escaping and introducing gaps into metrics collection
  - Avoid missing metrics from new and destroyed domains
- Prevent xapi concurrent calls during migration from indirectly make each other fail (already fixed in the refreshed ISOs)
- Fix a deadlock in xenopsd due to atom nesting (already fixed in the refreshed ISOs)
xo-lite: update to 0.12.0
- [VM/system] Display system information in vm/system tab
- [Host/system] Display system view information in host/system tab
- [Host/Dashboard] Fix color of tag list (PR #8731)
- [Table] add pagination on table (PR #8573)
- [Pool/system] Display pool information in pool/system tab (PR #8660)
- [VM/Dashboard] Display VM information in dashboard tab (PR #8529)
- [Tab/Network] Updated side panel in tab network behavior for mobile view (PR #8688)
- [Stats] Fix graphs that were sometimes not displayed or displayed incorrectly (PR #8722)

Optional packages:

Alternate Driver: Updated to newer version.
- broadcom-bnxt-en-alt: Update to version 1.10.3_232.0.155.5

Test on XCP-ng 8.3

yum clean metadata --enablerepo=xcp-ng-testing
yum update --enablerepo=xcp-ng-testing
reboot

The usual update rules apply: pool coordinator first, etc.

Versions:

http-nbd-transfer: 1.6.0-1.xcpng8.3
sm: 3.2.12-3.2.xcpng8.3
xapi: 25.6.0-1.9.xcpng8.3
xen: 4.17.5-13.2.xcpng8.3
xo-lite: 0.12.0-1.xcpng8.3

Optional packages:

Alternate drivers:
- broadcom-bnxt-en-alt: 1.10.3_232.0.155.5-1.xcpng8.3

What to test

Normal use and anything else you want to test.

Test window before official release of the updates

~2 days.

flakpyro

@gduperrey Updated both of my test hosts.

Machine 1:
Intel Xeon E-2336
SuperMicro board.

Machine 2:
Minisforum MS-01
i9-13900H (e-cores disabled)
32 GB Ram
Using Intel X710 onboard NIC

Everything rebooted and came up fine. Ran a backup job which also completed without issue. Using a mix of ext, XFS and NFS SRs

ph7

@gduperrey
Ran some test on my old i7 and it seems to work fine

bufanda

Did install on my lab pool

2 HP EliteDesk 800 G3 mini

One with i7 6700T, one with i5 6500T
Both 32GB RAM
iSCSI SR, NFS SR and local NVMe SR. VMs from all SR run. Migration during update worked. Migration of VMs after update worked. Backups (exporting VMs) is working
Didn't see any issue so far.

One thing the pool had the qcow2 patches installed since XAPI and sm packages gto updated I guess it won't support qcow2 at the moment right?

dthenot

@bufanda Hello,

There is equivalent sm packages in the qcow2 repo for testing, XAPI will be coming soon.
You can update while enabling the QCOW2 repo to get the sm and blktap QCOW2 version and get the XAPI version letter if you want.

bufanda

@dthenot No worries, just wanted to make sure I understand the expected behavior when I had the qcow2 patches installed.

dthenot

@bufanda You just need to make sure to have a sm and blktap qcow2 version.
Otherwise, having a normal sm version would drop the QCOW2 VDI from the XAPI database and you would lose VBD to VM aswell as name of VDI.
So it could be painful depending on how much you have
But in the case, you would install a non QCOW2 sm version, you would only lose the QCOW2 from DB, those would not be deleted or anything. Reinstalling a QCOW2 version then rescanning the SR would make them re-appear. But then you would have to identify them again (lost name-label) and relink them to VM.
We try to keep our QCOW2 version on top of the testing branch of XCP-ng but we could miss it

bufanda

@dthenot Only have 2 Test VMs currently not running with qcow2. Nothing production and nothing crucial. It's a lab pool. As I said just to understand the behavior when I test the new updates. But thanks for the details. Really appreciated

JeffBerntsen

Installed and working on my lab test machine, HP ProLiant DL165 G7 with AMD Opteron 6136 CPU and 32 GB of RAM.

Andrew

@gduperrey Installed and running on a few hosts. Good so far.

gduperrey

Updates published: https://xcp-ng.org/blog/2025/07/03/july-2025-security-and-maintenance-update-for-xcp-ng-8-3-lts/

Thank you for the tests!

gb.123

Sorry to join the bandwagon late this time, but i got these errors while booting:

[    0.826901] ACPI BIOS Error (bug): Could not resolve [\_SB.PCI0.GPP2.WWAN], AE_NOT_FOUND (20180810/dswload2-160)
[    0.826908] ACPI Error: AE_NOT_FOUND, During name lookup/catalog (20180810/psobject-221)
[    0.826910] ACPI Error: Ignore error and continue table load (20180810/psobject-604)
[    0.826912] ACPI Error: Skip parsing opcode OpcodeName unavailable (20180810/psloop-543)
[    0.827236] ACPI BIOS Error (bug): Could not resolve [\_SB.PCI0.GPP2.WWAN], AE_NOT_FOUND (20180810/dswload2-160)
[    0.827239] ACPI Error: AE_NOT_FOUND, During name lookup/catalog (20180810/psobject-221)
[    0.827241] ACPI Error: Ignore error and continue table load (20180810/psobject-604)
[    0.827242] ACPI Error: Skip parsing opcode OpcodeName unavailable (20180810/psloop-543)
[    0.827245] ACPI BIOS Error (bug): Failure creating [\_SB.PCI0.GPP5.EWPM], AE_ALREADY_EXISTS (20180810/dswload2-316)
[    0.827248] ACPI Error: AE_ALREADY_EXISTS, During name lookup/catalog (20180810/psobject-221)
[    0.827250] ACPI BIOS Error (bug): Failure creating [\_SB.PCI0.GPP5._PRW], AE_ALREADY_EXISTS (20180810/dswload2-316)
[    0.827252] ACPI Error: AE_ALREADY_EXISTS, During name lookup/catalog (20180810/psobject-221)
[    0.827254] ACPI Error: Skip parsing opcode OpcodeName unavailable (20180810/psloop-543)
[    0.827257] ACPI BIOS Error (bug): Failure creating [\_SB.PCI0.GPP5.RTL8._S0W], AE_ALREADY_EXISTS (20180810/dswload2-316)
[    0.827259] ACPI Error: AE_ALREADY_EXISTS, During name lookup/catalog (20180810/psobject-221)

I think this is after the new update

Processor : AMD Ryzen 7 7840HS

olivierlambert

Hi,

It's just an ACPI error, it shouldn't be a problem.

gb.123

@olivierlambert

I agree. But just thought I should report nevertheless.

Update: This seems to be related to 7840HS motherboard. I tried it on another AMD Ryzen 7945HX but did not get this error.

olivierlambert

It's clearly due to the motherboard, yes. Mostly buggy BIOS/UEFI and ACPI tables. This world is like the far west.

gduperrey

New security and maintenance update candidate

A new XSA (Xen Security Advisory) was published on the 8th of July, related to hardware vulnerabilities in several AMD CPUS. Updated microcode mitigate it, and Xen is updated to adapt to the changes in the CPU features. We also publish other non-urgent updates which we had in the pipe for the next update release.

Security updates

amd-microcode:
- Update to 20250626-1 as redistributed by XenServer.
xen-*:
- Fix XSA-471 - New speculative side-channel attacks have been discovered, affecting systems running all versions of Xen and AMD Fam19h CPUs (Zen3/4 microarchitectures). An attacker could infer data from other contexts. There are no current mitigations, but AMD is producing microcode to address the issue, and patches for Xen are available. These attacks, named Transitive Scheduler Attacks (TSA) by AMD, include CVE-2024-36350 (TSA-SQ) and CVE-2024-36357 (TSA-L1).

Maintenance updates

http-nbd-transfer:
- Fix missing import exceptions in log files.
- Fix a potential HA startup failure with LINSTOR.
xo-lite: update to 0.12.1
- [Charts] Fix tooltip overflow when too close to the edge
- [Host/VM/Dashboard] Fix timestamp on some charts

Test on XCP-ng 8.3

yum clean metadata --enablerepo=xcp-ng-testing
yum update --enablerepo=xcp-ng-testing
reboot

The usual update rules apply: pool coordinator first, etc.

Versions:

amd-microcode: 20250626-1.1.xcpng8.3
http-nbd-transfer: 1.7.0-1.xcpng8.3
xen: 4.17.5-15.1.xcpng8.3
xo-lite: 0.12.1-1.xcpng8.3

What to test

Normal use and anything else you want to test.

Test window before official release of the updates

~2 days.

stormi

And fixes for XCP-ng 8.2 are coming but they require more work to backport the patches.

flakpyro

@stormi

Tested on my usual Intel test servers you've seen me post in this thread in the past as well as a 2 host AMD Epyc pool,(HP DL325 Gen10) rolling pool reboot worked as expected, hosts rebooted without issue, backups and replication appear to function as usual.

Andrew

@gduperrey Installed and running on Intel systems, and Zen3 system that sees the microcode update.