Linux Kernel 5.4 for Proxmox VE

t.lamprecht · Apr 10, 2020

Joogser said:
This single saint tuning /etc/modprobe.d/zfs.conf doesn't work in Proxmox-ve_6.1-2 with kernel 5.3.18-2-pve. For this issue resolving need to do kernel rollback\downgrading to kernel 5.0.15-1-pve.

It works here:

Code:

root@prod2:~# cat /sys/module/zfs/parameters/zfs_arc_max
0
root@prod2:~# uname -a
Linux prod2 5.3.18-3-pve #1 SMP PVE 5.3.18-3 (Tue, 17 Mar 2020 16:33:19 +0100) x86_64 GNU/Linux
root@prod2:~# grep c_max /proc/spl/kstat/zfs/arcstats
c_max                           4    16780830720
root@prod2:~# echo 'options zfs zfs_arc_max=8589934592' > /etc/modprobe.d/zfs.conf
root@prod2:~# update-initramfs -u
update-initramfs: Generating /boot/initrd.img-5.3.18-3-pve
Running hook script 'zz-pve-efiboot'..
Re-executing '/etc/kernel/postinst.d/zz-pve-efiboot' in new private mount namespace..
No /etc/kernel/pve-efiboot-uuids found, skipping ESP sync.
root@prod2:~# reboot 

...

# ssh root@192.168.30.65
Linux prod2 5.3.18-3-pve #1 SMP PVE 5.3.18-3 (Tue, 17 Mar 2020 16:33:19 +0100) x86_64
    [ --8< -- snip --8<-- ]
Last login: Thu Apr  9 18:08:38 2020 from 192.168.16.38
root@prod2:~# grep c_max /proc/spl/kstat/zfs/arcstats
c_max                           4    8589934592
root@prod2:~# cat /sys/module/zfs/parameters/zfs_arc_max
8589934592
root@prod2:~#

As screenshots seem to be more proof worthy..:

Seems to work fine..

So please open a new thread and specify what you changed in your setup that makes it act like this, as it def. works on a default setup from our installer here, more than one re-checked, so if you really do not forgot a step then it has to be something out of the ordinary to not work.. You can use cat /sys/module/zfs/parameters/zfs_arc_max to check if the module parameter is actually in effect.

harvie · Apr 10, 2020

Maybe ZFS version changed between these kernels? You can check it eg. using modinfo zfs | head
And report problems to https://github.com/openzfs/zfs/issues

Didier Misson · Apr 17, 2020

t.lamprecht said:
No, the 5.4 Linux Kernel has no native Wireguard support.

I think native Wireguad support will be in Kernel 5.6 ... not before

cyp · Apr 17, 2020

Installed on a few servers and on a cluster of 5 nodes (5 servers like this https://www.scaleway.com/en/dedibox/pro/pro-5-l/ ) with Ceph.

Everything run fine and stable since two weeks.

t.lamprecht · Apr 17, 2020

Didier Misson said:
I think native Wireguad support will be in Kernel 5.6 ... not before

Yes, as said 5.4 won't have it natively, but using wireguard-dkms package from buster-backports works great; and as that is more or less the same code from the same author(s), so same experience.

t.lamprecht · Apr 17, 2020

cyp said:
Installed on a few servers and on a cluster of 5 nodes (5 servers like this https://www.scaleway.com/en/dedibox/pro/pro-5-l/ ) with Ceph.

Everything run fine and stable since two weeks.

thanks for the feedback!

moxfan · Apr 19, 2020

t.lamprecht said:
This works just fine here on both, 5.4 and 5.3 based kernels, actually I just successfully retested here.. Maybe you just forgot to update the initramfs:
https://pve.proxmox.com/pve-docs/chapter-sysadmin.html#_limit_zfs_memory_usage

Quick feedback: Your method to limit ZFS's memory usage works fine here with the 5.3 kernel.
Just 2 things I am not sure about.
The amount of ARC memory being used for a container, is that included in that container's and the host's (h)top command?
Also, is there a command to reliably find out the ARC mem used for each container?

morph027 · Apr 20, 2020

Also running cluster of 3 mixed nodes (ZFS) without any problems since the first release.

- Intel(R) Xeon(R) CPU E3-1230 v3 w/ 32GB RAM
- Intel(R) Xeon(R) CPU E3-1240 v5 w/ 64GB RAM
- AMD EPYC 7351P 16-Core Processor w/ 128GB RAM

morph027 · Apr 20, 2020

Also on my home lab (Intel(R) Xeon(R) CPU E3-1220L V2, 16GB RAM)

UniverseX · Apr 21, 2020

Is there any plans to add support for VirtIO-FS in Proxmox VE 6.2? It's been added to Linux 5.4 and now also to QEMU 5.0.

t.lamprecht · Apr 21, 2020

UniverseX said:
Is there any plans to add support for VirtIO-FS in Proxmox VE 6.2? It's been added to Linux 5.4 and now also to QEMU 5.0.

No, we will look at that at a later point, but not for Proxmox VE 6.2. Note here also that QEMU 5.0 is still not final released (it will maybe today, though ^^) and that the guest kernel needs to support this too, IIRC.

Bengt Nolin · Apr 23, 2020

Has anyone used the 5.4 kernel for in-tree support for the *new* Intel x710 NICs released 2019 using the i40e module? I had to build the module from the intel source package for pve-kernel-5.3 as it does not support these new cards (https://ark.intel.com/content/www/u.../intel-ethernet-network-adapter-x710-t2l.html for example).

There are things that seem to indicate drivers are included and working but some confirmed uses would be nice.

Mr.Goodcat · Apr 23, 2020

What's the general policy/roadmap on kernel updates? It appears as if Ubuntu kernels are introduced until Proxmox x.2. The next update after that will become available with the upcoming mayor release (i.e. PVE 7.0)?

t.lamprecht · Apr 23, 2020

Mr.Goodcat said:
What's the general policy/roadmap on kernel updates? It appears as if Ubuntu kernels are introduced until Proxmox x.2. The next update after that will become available with the upcoming mayor release (i.e. PVE 7.0)?

Yes, it doesn't have to be .2 but it normally falls together. We introduce a major version with a recent kernel, continue to upgrade to always stay on a supported (upstream wise) kernel until we get to the next LTS release, ideally both from the real upstream (kernel.org) and from Ubuntu, as then we know that the version will get the best support for older issues but also newer hardware. (4.15 was an exception due to the intel meltdown/specter mess then happening).

kkasberg · Apr 23, 2020

I get

Code:

Unable to locate package pve-kernel-5.4

when trying to install kernel 5.4 using the command in the first post.

t.lamprecht · Apr 23, 2020

Then you either

did not recently run a apt update
do not have your repositories access correctly configured, see https://pve.proxmox.com/wiki/Package_Repositories

The second one is more likely, I'd guess.

BobMccapherey · Apr 28, 2020

I'm hoping this gets rid of problems I've been having with the most recent 5.3 kernel and random page faults arising from the use of Intel GVT-g on a Coffee Lake ER Xeon processor.

NessageHostsINC · Apr 29, 2020

Dear,

Upon using the 5.4 kernel, we have these unexpected kernel logs in syslog. I have created a separate thread to reduce clutter here. https://forum.proxmox.com/threads/cpu-lockup-on-5-4-kernel.69099/

Yours truely

CRCinAU · May 6, 2020

NessageHostsINC said:
Dear,

Upon using the 5.4 kernel, we have these unexpected kernel logs in syslog. I have created a separate thread to reduce clutter here. https://forum.proxmox.com/threads/cpu-lockup-on-5-4-kernel.69099/

Yours truely

I see the same, CPU soft lockups leading to an unresponsive machine... Added more to the above thread...

onepamopa · May 11, 2020

Well, looks like mine are "hard" lockups ... everything freezes, and not a single line @ any of the logs..,,
I've enabled crash dumps.. hopefully at least they will provide something..

Linux Kernel 5.4 for Proxmox VE

Proxmox Staff Member

Well-Known Member

Member

Renowned Member

Proxmox Staff Member

Proxmox Staff Member

Active Member

Renowned Member

Renowned Member

Member

Proxmox Staff Member

Active Member

Member

Proxmox Staff Member

Renowned Member

Proxmox Staff Member

Member

Member

Well-Known Member

Active Member