Proxmox random reboots on HP Elitedesk 800g4 - fixed with proxmox install on top of Debian 12 - now issues with hardware transcoding in plex

[dustojnikhummer]

Have you read the thread? The suggestions here fixed my issue.

For some reason yesterday most of this thread didn't load for me. I will need to try the GPU power management workaround to see power consumption figures. I really don't want to run HyperV

Other threads https://forum.proxmox.com/threads/proxmox-restarts-randomly.116080/page-2 recommend disabling TPM, which might be the best solution so far

 

[rophan]

Disabling TPM didn't make any difference on my machine (Elitedesk 800 mini G4 i7-8700t which was rebooting randomly).

 

[dustojnikhummer]

We will see. Installing PVE right now, see if it survives tomorrow without a reboot. I don't want to disable the iGPU power management.

 

[rophan]

I don't want to disable it either but rathre that than disable all CPU power management. `enable_dc=0` doesn't disable all GPU power management - powertop shows for my headless machine the GPU spends all its time in "RC6" sleep state. It just disables the more fancy deeper sleep states, others have suggested maybe 2-3W difference.

 

[Spodster]

Fairly new user to Proxmox here (less than 1 week) and I have what appear to be 'sleep' issues out of the gate.

Scenario is If left for a few hours I can come back to it and there's no HDMI output to the screen.
NUC Power light is lit permanently like it's on.
Wiggle mouse give the keyboard a poke and a few seconds later you see the NUC power light go off, hear the fans spin up, you see the BIOS screen, at this point it looks like Proxmox crashed.

Ran a memtest for 4-5 hours with no issue but then got pointed to this issue indirectly.
I don't have the ability to leave a HDMI screen plugged in so have deployed the `enable_dc=0` GRUB change so far.
Last night it didn't crash. sleep, whatever it does in my case but early days as only tended to be overnight when left alone.

FYI this is on a NUC7i7DNKE NUC with an Intel® Core™ i7-8650U Processor.

 

[dustojnikhummer]

Fairly new user to Proxmox here (less than 1 week) and I have what appear to be 'sleep' issues out of the gate.

Scenario is If left for a few hours I can come back to it and there's no HDMI output to the screen.
NUC Power light is lit permanently like it's on.
Wiggle mouse give the keyboard a poke and a few seconds later you see the NUC power light go off, hear the fans spin up, you see the BIOS screen, at this point it looks like Proxmox crashed.

Ran a memtest for 4-5 hours with no issue but then got pointed to this issue indirectly.
I don't have the ability to leave a HDMI screen plugged in so have deployed the `enable_dc=0` GRUB change so far.
Last night it didn't crash. sleep, whatever it does in my case but early days as only tended to be overnight when left alone.

FYI this is on a NUC7i7DNKE NUC with an Intel® Core™ i7-8650U Processor.

Proxmox 8.0.4 I assume? If this https://gitlab.freedesktop.org/drm/intel/-/issues/9244 issue affects Kernel 6 on Intel 8000 CPUs (now also mobile assuming your issue is the same thing), it might be worth trying Proxmox 7

 

[Spodster]

Yeah I'm on 8.0.4 and that link is the same one from earlier in this thread.
Well so far I have 1d 9h uptime with no random reboot, crash, suspend whatever it was doing.
Testing ongoing as I try not to use the HDMI port to screen in case that's keeping it 'alive'.

One thing I did notice though and I'm not sure if it did this before the GRUB change is I'm seeing errors around i915
Is this problem in general a Proxmox, Intel or other areas issue as in will and is someone working on a permanent fix?

 

[dustojnikhummer]

So far I'm at 60 hours myself. (totally stock install, no disabling iGPU power management) Maybe there was an upstream fix that is alread out?

 

[Spodster]

Just checked for updates and nothing came up, I have the PVE-no-subscription defined.
proxmox-ve: 8.0.2 (running kernel: 6.2.16-19-pve)
pve-manager: 8.0.4 (running version: 8.0.4/d258a813cfa6b390)

 

[Spodster]

So my uptime is now 3.5 days so safe to say the `enable_dc=0` GRUB change so far has held.
More a workaround than a fix but it works for now.

 

[Spodster]

Also found this archlinux wiki article on the subject

 

[Apicedda]

I found another weird behaviour, this might make my unit issue different then the rest of you (or not?). It was working fine for 10+ days without any of the workarounds applied yet, and with everything disconnected except power and lan. I decided to test the sleep function, so i put it to sleep overnight with

sudo rtcwake -m mem --date 10:00

on wakeup, it started rebooting every 2 to 20 minutes. As soon as i plugged in the usb keyboard, it stopped rebooting. After some time i unplugged the keyboard and it didn't reboot.
I tried the sleep again the next night, this time with the keyboard plugged in. Woke up and didn't reboot once.
I tried again, with the keyboard unplugged, it woke up and froze (so no reboot but completely frozen, no response from pings, had to hard shutdown), after restarting, the rebooting began, and stopped as soon as i plugged the keyboard in.
I'll finally try the workarounds, and also ask the seller for another unit to see if it has the same issues

 

[mad2moons]

I have a HP Elitedesk G5 8500 and its been randomly rebooting since i got it. Tried swapping memory, SSD, NVME, new thermal paste and still the same problem. Glad i found this post as i renamed the bin file mentioned and its now been up for 18 hrs!

Thanks again for the solution.

 

[d3sp3rados13]

Very interesting stuff here.
I'm sad to say that this doesn't work for my HP Elitedesk 800 G4 SFF.
I renamed the bin file and also tried the kernel parameter. Still got random reboots every x hours...
From my experience it doesn't make any difference to my system if there is a monitor attached/turned on or not. It crashes independently of that.

 

[minimo]

Hello, that's sounds like my problem.
But the device keep active and blows, nevertheless, it was getting warmer.
The case then had quite a high temperature.

Yesterday, I set the kernel parameter and restarted the device without a monitor, and it was still running this morning.

In case someone else hasn't done something like this yet:

nano /etc/default/grub

Search for the line with

GRUB_CMDLINE_LINUX_DEFAULT

and add it here in the inverted commas.
e.g.:

GRUB_CMDLINE_LINUX_DEFAULT="quiet i915.enable_dc=0"

update-grub

after restart check config with

cat /proc/cmdline

Source: https://askubuntu.com/questions/19486/how-do-i-add-a-kernel-boot-parameter

Thanks for the providing the commands. I applied this and now have uptime over 1day.

 

[skandal]

GRUB_CMDLINE_LINUX_DEFAULT="quiet i915.enable_dc=0" works for me perfectly. I had issue with HP Prodesk 600 G4 mini. It randomly reboots, sometimes once at 24 hours, sometimes more often. After apply changes in GRUB i have 3 day + without reboot.

 

[cristianmc]

So re-loaded proxmox with the bare-metal installer and renamed the firmware file, haven't had a crash yet. Monitor is fully disconnected, been running since last night. I was also able to get hardware transcoding working in plex (lxc container), i never was able to get it to work with the debian install. Probably some driver or something.

I had the same problem and I solved it by renaming the file that my colleague mentions here, now the question is, does anyone know what function that file performs?

 

[victorc26]

I recently upgraded to Proxmox 8 from 7 a few weeks ago and immediately saw the issue mentioned in this thread. I found this thread and wanted to report that adding i915.enable_dc=0 at the end of the GRUB_CMDLINE_LINUX_DEFAULT line in grub resolved the issue for me (Complete line now looks like this: GRUB_CMDLINE_LINUX_DEFAULT="quiet i915.enable_dc=0" )

 

[granrup]

I apologize for necro-posting but this was still happening on Proxmox 8.2 for my HP ProDesk 600 G4 mini with the Coffee Lake gen i5-8500T. Same problem with crashes like 3-11 minutes (or every hour, depends) when disconnecting DP from TV. TV just has to be present, not on for it to be recognized. This issue was happening with stock Debian 12 on a previous install (w/o Proxmox) for this machine.

Edit: i915.enable_dc=0 works for me.

 

[Winterd85]

Hello, I'm new to the forum and don't use Proxmoxx. But I had the same problems on an HP ProDesk 400 G5 DM Mini PC with Ubuntu 20.04 to 24.10 from kernel 4.8.x >.

The main problem with this problem is that the hardware reaches C-State C10 without a monitor from kernel 4.8.x and activated ASPM (pcie_aspm=force).

You can check the whole thing with powertop
"sudo powertop" and "Tab to the C-States"

My solution here is to add the entry i915.enable_dc=0 intel_idle.max_cstate=7" to the entry in /etc/default/grub.

sudo nano /etc/default/grub
find the line looks a like GRUB_CMDLINE_LINUX_DEFAULT=" and add like this exable.

GRUB_CMDLINE_LINUX_DEFAULT="i915.enable_dc=0 intel_idle.max_cstate=7"

save with Ctrl-C > enter.

Update your grub to use the new entries at the next boot with:
sudo update grub

Curiously, the CPU C-States are now only shown up to C9 in Powertop, but the computer now runs even more economically than ever before. 2.2-2.4 watts with the running services.

You can check the max_cstate before and after with this command.

cat /sys/module/intel_idle/parameters/max_cstate


the whole thing is actually not a bug. It is more the excellent driver connection where I can reach the IGPU C10 on non mobile Intel I-Series 8/9 CPUs in headless mode. However, the energy management in the CPU is not designed for this and it crashes. I think it is due to a lack of power.