[HELP] TR System Hangs When GPU Flagged for Passthrough

[colehoffman]

Hey all, been fighting with this the past few days.

 

My current specs are:

TR 1950X

ASRock X399M Taichi

2x8GB Corsair Vengeance DDR4

AMD Radeon HD 6450 (first slot)

Powercolor AMD 5700XT Red Devil (second slot)

EVGA G3 850W

2x500GB Samsung SSD

WD 1TB M.2 SSD

 

The problem:

So the common issue is that the system will randomly hang when I try to enable the 5700XT for passthrough, the only way to fix it is to shut down the system. However, it goes into a boot-loop. The only solution I have found is to unplug the computer, let it sit for a few minutes, and then plug it back in. After this, it boots however the BIOS are reset.

 

I have tried this on a multitude of different software. First unRAID, Proxmox, then ESXi, now back to unRAID. Always the same issue. I assumed it was a software issue so I made the switch, then it happened on proxmox so I figured it was a manual config error, and with ESXi, it happened again, so I switched back due to Unraid being more user-friendly. 

 

I left it on overnight (no passthrough or anything configured, base install) and did not have any hanging issue. However this morning I went to to enable the 5700XT for passthrough rebooted, and then the entire system hangs (sometimes it will survive for a few minutes, sometimes a few seconds).

 

Steps I have taken:

Updated BIOS

Disabled anything C-State related in BIOS

Reseated RAM, GPUs

Tried different software

 

The cooling situation:

Corsair iCUE H100i Elite Capellix. The MB consistently reports temps of ~60C. However, putting my hand on the exhaust fan, does not feel hot. To be safe - I have made sure all the fan headers are running at full speed from the BIOS (I set it every time the BIOS resets)

 

These are all known working parts from separate systems I have.

 

I am at a loss for what to do.

 

Thank for you for time.

 

EDIT: To anyone having this issue in the future. I just reflashed the same version of the BIOS and was able to passthrough using Unraid. 

Edited June 6, 2022 by colehoffman

[ghost82]

35 minutes ago, colehoffman said:

So the common issue is that the system will randomly hang when I try to enable the 5700XT for passthrough

Is this really hanging or unraid is reacheble from an external device through ssh or web gui?

If it really hangs you should enable syslog server, so it can log things even if it crashes; syslog inside diagnostics will log only the current boot errors (if any).

Once you have the syslog you should analyze it.

[colehoffman]

Just now, ghost82 said:

Is this really hanging or unraid is reacheble from an external device through ssh or web gui?

If it really hangs you should enable syslog server, so it can log things even if it crashes; syslog inside diagnostics will log only the current boot errors (if any).

Once you have the syslog you should analyze it.

Sounds good. Thanks for the input, will update once I have it

[colehoffman]

7 minutes ago, ghost82 said:

Is this really hanging or unraid is reacheble from an external device through ssh or web gui?

If it really hangs you should enable syslog server, so it can log things even if it crashes; syslog inside diagnostics will log only the current boot errors (if any).

Once you have the syslog you should analyze it.

It terms of the GUI - it completely fails. It will work for some x amount of time (before the "hang") then at some point I will click on something in the GUI and it fails to load (the "hang"). I do not know about ssh

Edited June 6, 2022 by colehoffman

[ghost82]

Start by attaching diagnostics so we can see if the vm and gpu are correctly configured.

If you have multiple vm remember to tell us what is the name of the vm we should look at.

[colehoffman]

30 minutes ago, ghost82 said:

Start by attaching diagnostics so we can see if the vm and gpu are correctly configured.

If you have multiple vm remember to tell us what is the name of the vm we should look at.

 

Before you had these suggestions I decided to reflash bios and reinstall unraid. So far so good. I have logging enabled just in case. Will update if issue reappears.