PVE 5.1: KVM broken on old CPUs

[pacow]

Got kernel 4.13.8-3-pve #1 SMP PVE 4.13.8-30 today. So happy, VMs work again. Many thanks for the effort in supporting older CPUs!

Have a nice day.

 

[profpolymath]

Thanks, Proxmox team, for resolving this. Much appreciated.

 

[Vairo]

Same problem here, two old servers (HP DL380 G5 and HP DL310G4)
Fabian post solved my problem as well
Thanks!

 

[dlasher]

Broken in "Linux pmx2 4.13.13-2-pve #1 SMP PVE 4.13.13-32 (Thu, 21 Dec 2017 09:02:14 +0100) x86_64 GNU/Linux" (Default 5.1 ISO install + updates)


root@pmx2:~# cat /proc/cpuinfo | grep nmi
root@pmx2:~#

model name : Quad-Core AMD Opteron(tm) Processor 8356
pveversion : pve-manager/5.1-41/0b958203 (running kernel: 4.13.13-2-pve)

newly created container, fails:

May 07 10:15:18 pmx2 systemd[1]: Starting PVE LXC Container: 505...
May 07 10:15:18 pmx2 lxc-start[14492]: lxc-start: 505: lxccontainer.c: wait_on_daemonized_start: 751 No such file or directory - Failed to receive the container state
May 07 10:15:18 pmx2 lxc-start[14492]: lxc-start: 505: tools/lxc_start.c: main: 371 The container failed to start.
May 07 10:15:18 pmx2 lxc-start[14492]: lxc-start: 505: tools/lxc_start.c: main: 373 To get more details, run the container in foreground mode.
May 07 10:15:18 pmx2 lxc-start[14492]: lxc-start: 505: tools/lxc_start.c: main: 375 Additional information can be obtained by setting the --logfile and --logpriority options.

Have Tested : http://download.proxmox.com/temp/pve-kernel-4.13.13-6-pve_4.13.13-42~test1_amd64.deb

^^^ no change

-- Unit pve-container@505.service has begun starting up.
May 07 10:31:43 pmx2 audit[5004]: AVC apparmor="DENIED" operation="mount" info="failed mntpnt match" error=-13 profile="/usr/bin/lxc-start" name="/" pid=5004 comm="lxc-start" flags="rw,
May 07 10:31:43 pmx2 kernel: audit: type=1400 audit(1525714303.665:12): apparmor="DENIED" operation="mount" info="failed mntpnt match" error=-13 profile="/usr/bin/lxc-start" name="/" pi
May 07 10:31:43 pmx2 lxc-start[4997]: lxc-start: 505: lxccontainer.c: wait_on_daemonized_start: 751 No such file or directory - Failed to receive the container state
May 07 10:31:43 pmx2 lxc-start[4997]: lxc-start: 505: tools/lxc_start.c: main: 371 The container failed to start.
May 07 10:31:43 pmx2 lxc-start[4997]: lxc-start: 505: tools/lxc_start.c: main: 373 To get more details, run the container in foreground mode.
May 07 10:31:43 pmx2 lxc-start[4997]: lxc-start: 505: tools/lxc_start.c: main: 375 Additional information can be obtained by setting the --logfile and --logpriority options.
May 07 10:31:43 pmx2 systemd[1]: pve-container@505.service: Control process exited, code=exited status=1
May 07 10:31:43 pmx2 systemd[1]: pve-container@505.service: Killing process 5004 (lxc-start) with signal SIGKILL.
May 07 10:31:43 pmx2 systemd[1]: Failed to start PVE LXC Container: 505.
-- Subject: Unit pve-container@505.service has failed
-- Defined-By: systemd
-- Support: https://www.debian.org/support
--
-- Unit pve-container@505.service has failed.
--
-- The result is failed.
May 07 10:31:43 pmx2 systemd[1]: pve-container@505.service: Unit entered failed state.
May 07 10:31:43 pmx2 systemd[1]: pve-container@505.service: Failed with result 'exit-code'.
May 07 10:31:43 pmx2 pct[4994]: command 'systemctl start pve-container@505' failed: exit code 1
May 07 10:31:43 pmx2 pct[4993]: <root@pam> end task UPID:pmx2:00001382:00002916:5AF08D7F:vzstart:505:root@pam: command 'systemctl start pve-container@505' failed: exit code 1

 

[dlasher]

^^^^^

FIXED IN: Linux pmx2 4.15.17-1-pve #1 SMP PVE 4.15.17-8 (Thu, 03 May 2018 08:43:38 +0200) x86_64 GNU/Linux

added pve-no-subscription, updated, resolved.

Hopefully there will be a new 5.1 ISO rolled shortly, since this is broken from install.

 

[fabian]

^^^^^

FIXED IN: Linux pmx2 4.15.17-1-pve #1 SMP PVE 4.15.17-8 (Thu, 03 May 2018 08:43:38 +0200) x86_64 GNU/Linux

added pve-no-subscription, updated, resolved.

Hopefully there will be a new 5.1 ISO rolled shortly, since this is broken from install.

it's not. it's only broken if you don't setup the PVE repositories/subscription correctly after installation, ignore the warnings/errors that apt gives you because of that misconfiguration, and then install updates for Debian only but not PVE. also, your problem is not related to this thread in any way.

 

[dzr22]

Hola Fabián,

Tengo el mismo problema que dlasher, podría decirme donde puedo encontrar el hilo correspondiente para solucionarlo?

Saludos.

-------------------------

Hello Fabian,

I have the same problem as dlasher, could you tell me where can I find the corresponding thread to solve it?

Greetings.

 

[fabian]

you need to configure valid PVE repositories, and upgrade the PVE packages (using "apt update; apt dist-upgrade"). check the documentation: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#sysadmin_package_repositories

 

[dzr22]

Thank you very much

 

[dlasher]

it's not. it's only broken if you don't setup the PVE repositories/subscription correctly after installation, ignore the warnings/errors that apt gives you because of that misconfiguration, and then install updates for Debian only but not PVE. also, your problem is not related to this thread in any way.

Not trying to be difficult here, perhaps we're facing a language/translation barrier.

Statement "Hopefully there will be a new 5.1 ISO rolled shortly, since this is broken from install."

That statement is true.

If you:

1. Download 5.1 ISO
2. Install
3. Create VM's
4. Try to start them. They fail.

This is repeatable across half a dozen opteron-based machines I have in the lab.

They won't start UNTIL you (1) add the repos, and (2) update, and (3) reboot

That's broken from installation. Hence the hope that a new ISO will be rolled shortly.

 

[wolfgang]

You can test PVE 5.2 iso

 

[lilCDNnrg]

I recently purchased 5 DL380 G5 servers and was able to get everything running smoothly using the 4.13 kernel but 2 of the 5 servers would not boot under 4.15 kernel and gave an ACPI error which i could correct by turning MPS tables off.

I appears there is some incompatibility with the new kernel but I am not sure exactly where as they are all the same hardware except for some CPU differences but only set of CPUs is different than the others that are working.

I do have all the packages up to date as of a few days ago with the exception of the kernel. I did try the 5.2.1 ISO but was unable to get the installer to load with MPS tables on, with the 5.1.3 ISO everything went fine.

 

[wolfgang]

I appears there is some incompatibility with the new kernel but I am not sure exactly where as they are all the same hardware except for some CPU differences but only set of CPUs is different than the others that are working.

I would guess there are different Bios version on this server installed.

 

[lilCDNnrg]

That was the first thing i thought and updated the BIOS on all of them to the latest version.

 

[fabian]

Not trying to be difficult here, perhaps we're facing a language/translation barrier.

Statement "Hopefully there will be a new 5.1 ISO rolled shortly, since this is broken from install."

That statement is true.

If you:

1. Download 5.1 ISO
2. Install
3. Create VM's
4. Try to start them. They fail.

This is repeatable across half a dozen opteron-based machines I have in the lab.

They won't start UNTIL you (1) add the repos, and (2) update, and (3) reboot

That's broken from installation. Hence the hope that a new ISO will be rolled shortly.

my post was refering to a different issue, because the user I responded to hijacked this thread. sorry if that caused confusion..