A data point - problems with 6.14.11-1-pve kernel

[sobaka]

One of the proxmox servers in my home lab exhibited kernel faults, high load average and extreme sluggishness after booting up with the 6.14.11-1-pve kernel.

After rebooting with the 6.8.12-13-pve kernel, all was well.

Seems to be a corner case, since my other nodes seem fine on the latest kernel.

Machine specs: Dell XPS 8960
Intel(R) Core(TM) i7-14700 w/ 28 cores
64 GB RAM
1 TB hard disk - OS
4 TB NVME - ceph volumes
Main network - Realtek Semiconductor Co., Ltd. Killer E3000 2.5GbE Controller
DMZ network - Intel Corporation 82575EB Gigabit Network Connection
Ceph heartbeat network - Intel Corporation 82575EB Gigabit Network Connection

 

[sobaka]

To verify, and narrow things down a bit more, I booted back up into the 6.14.11-1 kernel, and within a few minutes, the load average was up over 20.

I've attached a screenshot of top showing the load average, and of the kernel panic.

I also tried kernel 6.14.8-2-pve and quickly wound up in the same state.

Then, I booted into kernel 6.8.12-3-pve, currently running 7 VMs after an hour, with load average at 0.32

There's definitely a problem with the 6.14 kernel series.

 

[helio]

To verify, and narrow things down a bit more, I booted back up into the 6.14.11-1 kernel, and within a few minutes, the load average was up over 20.

I've attached a screenshot of top showing the load average, and of the kernel panic.

I also tried kernel 6.14.8-2-pve and quickly wound up in the same state.

Then, I booted into kernel 6.8.12-3-pve, currently running 7 VMs after an hour, with load average at 0.32

There's definitely a problem with the 6.14 kernel series.

I have a similar problem
HP EliteDesk 800 G5 Desktop Mini
Intel(R) Core(TM) i5-9500T CPU @ 2.20GHz
64 GB RAM
1 TB hard disk - OS
with 6.14.11-1-pve kernel panic.
revert to 6.8.12-14-pve working with no problem

 

[sobaka]

I have a similar problem
HP EliteDesk 800 G5 Desktop Mini
Intel(R) Core(TM) i5-9500T CPU @ 2.20GHz
64 GB RAM
1 TB hard disk - OS
with 6.14.11-1-pve kernel panic.
revert to 6.8.12-14-pve working with no problem

Someone suggested that it was a problem with the intel igb drivers, and they are different from 6.8 to 6.14 -

strings 6.14.8-2-pve/kernel/drivers/net/ethernet/intel/igb/igb.ko|grep srcversion
srcversion=4694DAF7C43B810B11AF58C
__UNIQUE_ID_srcversion505
strings 6.8.12-13-pve/kernel/drivers/net/ethernet/intel/igb/igb.ko|grep srcversion
srcversion=A624BAF2888F20DF0AC93F4
__UNIQUE_ID_srcversion290

However, I have 2 other proxmox nodes with the intel ethernet cards, running on the igb driver, and they don't have a problem with 6.14

Hopefully this will come to the attention of the right people.

 

[helio]

Someone suggested that it was a problem with the intel igb drivers, and they are different from 6.8 to 6.14 -

strings 6.14.8-2-pve/kernel/drivers/net/ethernet/intel/igb/igb.ko|grep srcversion
srcversion=4694DAF7C43B810B11AF58C
__UNIQUE_ID_srcversion505
strings 6.8.12-13-pve/kernel/drivers/net/ethernet/intel/igb/igb.ko|grep srcversion
srcversion=A624BAF2888F20DF0AC93F4
__UNIQUE_ID_srcversion290

However, I have 2 other proxmox nodes with the intel ethernet cards, running on the igb driver, and they don't have a problem with 6.14

Hopefully this will come to the attention of the right people.

I have a backup system in a
LENOVO ThinkCentre M93p Intel(R) Core(TM) i5-4590S CPU @ 3.00GHz
32 GB RAM
512 GB hard disk - OS
This system is runnig 6.14.11-1-pve kernel
But it just runs when I make same work on the other system.

Tonight I have a schedule backup to a PBS on a nfs share.
on the HP EliteDesk 800 G5 Desktop Mini that at that time running 6.14.11-1-pve
I could not access the web interface neither ssh to the machine.
Connected to a monitor and just revert back to 6.8.12-14-pve reboot
And it was back.
I dont know if it help to find a solution.
Thanks