I've been dealing with an issue with the HP Elitedesk systems for a few weeks now. I ordered 2 identical (but from different sellers) HP elitedesk 800G4's from ebay. The first on I received, i was able to load Proxmox from the official bare metal installer and everything has worked fine. No random reboots, no noticeable issues. I did add a nvme drive for ZFS local storage and I installed new RAM (32GB).
The second unit arrived, I loaded up proxmox the same way, installed an additional nvme drive for ZFS local storage and added the new 32GB of RAM. I created a cluster from node 1 (pve1 in this case, the working node that has no issues) and added node 2 (pve2) to the cluster. I also setup an external qdevice for the cluster so quarum is OK. Things seemed fine, until the second device randomly rebooted. I thought maybe it was just a blip of some sort and continued on.. then more random reboots and just no system stability at all on the second node. Searching the syslog and doing some searching on this forum didn't really find a working solution. I did see someone mention to try adding pci=assign-busses apicmaintimer idle=poll reboot=cold,hard to the default command line for grub, this did seem to help however it caused another issue, very high CPU temps even when nothing was going on. I suspect it's the idle=poll. if I remove that, the problems came back.
I tried swapping the power supply on both devices to see if that was the problem but pve2 continued to have the issue. I've reinstalled proxmox several times, trying BIOS and UEFI, i tried doing a factory reset on the BIOS, tried disabling the TPM. I finally decided there must be some issue with the CPU or motherboard since everything else was new (nvme drives, ram, tested different power supply, etc) so I sent the device back and paid the restock fee. I then ordered another elitedesk 800g4 from a different vendor and lo and behold the same exact issues are happening with this device also. Random reboots, no rhyme or reason. Its been a couple of weeks now of messing with settings, reloading proxmox etc etc.
So finally, I decided to load Debian 12 on the device and just see how it runs, that's when I found out I could install proxmox on a debian install. I followed this guide: https://pve.proxmox.com/wiki/Install_Proxmox_VE_on_Debian_12_Bookworm
Here's the thing, I've had this device running now for 24 hours and I've had 0 stability issues. Got it joined to the cluster, moved several VM's and containers over, things are working great. So first question, what could be the difference here? why would this work over the bare-metal proxmox install?
Now surfaces a new issue, I have plex installed on an ubuntu LXC container and I have GPU passthrough working (hardware transcoding works fine) on pve1. I was able to get it working on pve2 no problem before, but now after the debian -> proxmox install I cannot get hw transcoding to work. I can see that the iommu is enabled and I can see the render device so i'm not exactly sure what the issue is. Prerhap some driver is missing that is included in the base proxmox install? Maybe some module needs to be loaded, haven't been able to find much in my googling for this specific type of issue.
The second unit arrived, I loaded up proxmox the same way, installed an additional nvme drive for ZFS local storage and added the new 32GB of RAM. I created a cluster from node 1 (pve1 in this case, the working node that has no issues) and added node 2 (pve2) to the cluster. I also setup an external qdevice for the cluster so quarum is OK. Things seemed fine, until the second device randomly rebooted. I thought maybe it was just a blip of some sort and continued on.. then more random reboots and just no system stability at all on the second node. Searching the syslog and doing some searching on this forum didn't really find a working solution. I did see someone mention to try adding pci=assign-busses apicmaintimer idle=poll reboot=cold,hard to the default command line for grub, this did seem to help however it caused another issue, very high CPU temps even when nothing was going on. I suspect it's the idle=poll. if I remove that, the problems came back.
I tried swapping the power supply on both devices to see if that was the problem but pve2 continued to have the issue. I've reinstalled proxmox several times, trying BIOS and UEFI, i tried doing a factory reset on the BIOS, tried disabling the TPM. I finally decided there must be some issue with the CPU or motherboard since everything else was new (nvme drives, ram, tested different power supply, etc) so I sent the device back and paid the restock fee. I then ordered another elitedesk 800g4 from a different vendor and lo and behold the same exact issues are happening with this device also. Random reboots, no rhyme or reason. Its been a couple of weeks now of messing with settings, reloading proxmox etc etc.
So finally, I decided to load Debian 12 on the device and just see how it runs, that's when I found out I could install proxmox on a debian install. I followed this guide: https://pve.proxmox.com/wiki/Install_Proxmox_VE_on_Debian_12_Bookworm
Here's the thing, I've had this device running now for 24 hours and I've had 0 stability issues. Got it joined to the cluster, moved several VM's and containers over, things are working great. So first question, what could be the difference here? why would this work over the bare-metal proxmox install?
Now surfaces a new issue, I have plex installed on an ubuntu LXC container and I have GPU passthrough working (hardware transcoding works fine) on pve1. I was able to get it working on pve2 no problem before, but now after the debian -> proxmox install I cannot get hw transcoding to work. I can see that the iommu is enabled and I can see the render device so i'm not exactly sure what the issue is. Prerhap some driver is missing that is included in the base proxmox install? Maybe some module needs to be loaded, haven't been able to find much in my googling for this specific type of issue.
Last edited:
