Proxmox VE 8.0 released!

fiona · Sep 8, 2023

nomizfs said:
Yes, the cryptsetup complaints are normal.

here's my complete term.log

The full log relating to the error is here

Code:

Setting up linux-headers-6.1.0-11-amd64 (6.1.38-4) ...^M$
/etc/kernel/header_postinst.d/dkms:^M$
dkms: running auto installation service for kernel 6.1.0-11-amd64.^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
Deprecated feature: REMAKE_INITRD (/etc/dkms/framework.conf)^M$
Sign command: /usr/lib/linux-kbuild-6.1/scripts/sign-file^M$
Signing key: /var/lib/dkms/mok.key^M$
Public certificate (MOK): /var/lib/dkms/mok.pub^M$
Certificate or key are missing, generating self signed certificate for MOK...^M$
Deprecated feature: REMAKE_INITRD (/var/lib/dkms/sfc/4.15.14.1001/source/dkms.conf)^M$
^M$
Building module:^M$
Cleaning build area...^M$
'make' -C /var/lib/dkms/sfc/4.15.14.1001/build/linux_net KPATH=/lib/modules/6.1.0-11-amd64/build NDEBUG=1.....................(bad exit status: 2)^M$
Error! Bad return status for module build on kernel: 6.1.0-11-amd64 (x86_64)^M$
Consult /var/lib/dkms/sfc/4.15.14.1001/build/make.log for more information.^M$
Error! One or more modules failed to install during autoinstall.^M$
Refer to previous errors for more information.^M$
dkms: autoinstall for kernel: 6.1.0-11-amd64 failed!^M$
run-parts: /etc/kernel/header_postinst.d/dkms exited with return code 11^M$
Failed to process /etc/kernel/header_postinst.d at /var/lib/dpkg/info/linux-headers-6.1.0-11-amd64.postinst line 11.^M$
^[[1mdpkg:^[[0m error processing package linux-headers-6.1.0-11-amd64 (--configure):^M$
 installed linux-headers-6.1.0-11-amd64 package post-installation script subprocess returned error exit status 1^M$

The kernel 6.1.0-11-amd64 is not one provided by Proxmox. Did you install it manually at some point? Or was Proxmox VE installed on top of Debian? Then you should also remove the linux-image-amd64 package: https://pve.proxmox.com/wiki/Install_Proxmox_VE_on_Debian_11_Bullseye#Remove_the_Debian_Kernel

nomizfs · Sep 8, 2023

ok, seems i have manually installed sfc-dkms_4.15.14.1001-1_all.deb

previously, which is solarflare driver from their homepage. apt remove --purge sfc-dkms
seems to have resolved the issue i think..

EDIT: issue solved, it was manually install solarflare kernel module dkms, i had forgot about that. Thanks for your help @fiona

i'll delete some posts above to keep thread clean

Andrii.B · Sep 8, 2023

I installed one more server with pve8. There is no issue with migration on it. I'll reinstall the first server and try again.
I'll create a new threads if the issue will be repeated again.

adamb · Sep 8, 2023

Anyone else having issues with the 6.2.x kernel and new Intel 4th gen servers?

https://forum.proxmox.com/threads/6-2-x-kernel-issues-on-new-hardware.132890/

Andrii.B · Sep 9, 2023

My issue was on kernel pve-kernel-6.2.16-3-pve and was gone on kernel proxmox-kernel-6.2.16-6-pve.

vincentedison · Sep 23, 2023

Dunuin said:
What I have read so far OVH isn't deploying a a default PVE when setting this up using their tools. Default PVE8 and Debian 12 come with crontab.

Thankyou! I just run

Code:

apt-get install cron

and it's working.

lijan55 · Oct 4, 2023

Pve kernel 6.x is the only option in Proxmox 8 and it has a bug with nested ESXi on AMD platform.
https://forum.proxmox.com/threads/nested-esxi-vms-dont-start.134291/
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2008583
https://git.kernel.org/pub/scm/linu.../?id=174a921b6975ef959dd82ee9e8844067a62e3ec1

@ Proxmox team Is it possible to omit the patch mentioned in the link above in next kernel 6.5 which is to be released in Q4 ?

t.lamprecht · Oct 5, 2023

lijan55 said:
Pve kernel 6.x is the only option in Proxmox 8 and it has a bug with nested ESXi on AMD platform.

This is not a bug in the kernel and reverting that patch currently would be simply wrong and possibly introduce actual bugs (and also still break some newer VMWare tools like Workstation 17). After a discussion with another kernel developer, one of our devs will take a stab into adding (conveying) FLUSHBYASID support ourselves – but no promises yet.

Whatever · Oct 5, 2023

@t.lamprecht

Is there any progress on sorting out the degradation in KSM and huge performance (CPU 100% spikes) loss in Windows guests with 6.2 kernels?

dMopp · Oct 5, 2023

Iam really waiting for a 6.3+ kernel for the much better AMD power management. This might be a big difference in my homelab

Neobin · Oct 6, 2023

dMopp said:
Iam really waiting for a 6.3+ kernel

t.lamprecht said:
Exactly we currently base off Ubuntu Lunar's kernel as base for our own adaptions and will switch over to the 6.5 based kernel from Ubuntu 23.10 Mantic in Q4

https://forum.proxmox.com/threads/proxmox-ve-8-0-released.129320/post-573961

guruevi · Oct 6, 2023

There is currently a bug in ZFS on Sapphire Rapids with the AMX features in the processor. It appears to be locking up the kernel frequently (even just ZFS boot). Add clearcpuid=600 to /etc/kernel/cmdline, do proxmox-boot-tool refresh, then reboot. A patch is in ZFS 2.2 which is currently in RC4.

Neobin · Oct 6, 2023

guruevi said:
There is currently a bug in ZFS on Sapphire Rapids with the AMX features in the processor. It appears to be locking up the kernel frequently (even just ZFS boot). Add clearcpuid=600 to /etc/kernel/cmdline, do proxmox-boot-tool refresh, then reboot. A patch is in ZFS 2.2 which is currently in RC4.

For reference:

dale · Oct 14, 2023

t.lamprecht said:
We have a prototype here for a pure serial console install, but it was to late in the dev cycle for Proxmox VE 8.0 and the TUI installer is already helpful for many without that, but we will ship supporting plain serial consoles with a future release.

Hi, Thomas.

Ok, is there any possibility to use the "pure serial console" installation today without waiting for the next release?

adam23450 · Oct 15, 2023

Will proxmox finally support a cluster of 2 nodes and if one of them goes out, it will no longer be read-only? such behavior is absurd

ness1602 · Oct 15, 2023

How is 2-node cluster anything but the liability?

t.lamprecht · Oct 15, 2023

adam23450 said:
Will proxmox finally support a cluster of 2 nodes and if one of them goes out, it will no longer be read-only? such behavior is absurd

No, we still actually care about providing data integrity and resilience about issues stemming from unsynchronized resource access due to a split-brain cluster, we certainly do not see that as being absurd, but rather a must-have standard feature for enterprise class hyper-visors.

You can use an external QDevice if your setup/budget/whatever doesn't allow you for using more than one node:
https://pve.proxmox.com/pve-docs/chapter-pvecm.html#_corosync_external_vote_support

t.lamprecht · Oct 15, 2023

dale said:
Ok, is there any possibility to use the "pure serial console" installation today without waiting for the next release?

Not easily via our official ISO, as the ISO's bootloader grub would need to be configured for that to be able to do all input/output via serial connection from the very start.

While you might be able to add the required changes and repack the ISO, it'd be way easier to install Proxmox VE on top of a plain Debian installation, see:
https://pve.proxmox.com/wiki/Install_Proxmox_VE_on_Debian_12_Bookworm

adam23450 · Oct 15, 2023

t.lamprecht said:
No, we still actually care about providing data integrity and resilience about issues stemming from unsynchronized resource access due to a split-brain cluster, we certainly do not see that as being absurd, but rather a must-have standard feature for enterprise class hyper-visors.

You can use an external QDevice if your setup/budget/whatever doesn't allow you for using more than one node:
https://pve.proxmox.com/pve-docs/chapter-pvecm.html#_corosync_external_vote_support

why are you unable to ensure data integrity on 2 nodes and e.g. vmware can?

ness1602 · Oct 15, 2023

But vmware(as in vSAN) uses witness (aka third node) as a quorum ,same as proxmox/

Proxmox VE 8.0 released!

Proxmox Staff Member

Renowned Member

Well-Known Member

Famous Member

Well-Known Member

New Member

New Member

Proxmox Staff Member

Renowned Member

Member

Distinguished Member

Active Member

Distinguished Member

Renowned Member

Member

Renowned Member

Proxmox Staff Member

Proxmox Staff Member

Member

Renowned Member