SOLVED: Transcoding Error?

[Jdo139]

I have jellyfin running in a container on proxmox with the helper script and I have passed through a GTX1080 for hardware encoding and it was working great. Nvidia-SMI showed usage and videos would transcode when it needed to perfectly.

I mainly use Infuse to watch most of my content but sometimes other family members use swiftfin or the native jellyfin app depending on the device they’re watching on. 

I just finished watching an episode of a show on my iOS device through infuse, and tried to play the next episode and got this error.

Code:

An error occurring loading this content.

I figured my WiFi was slow at the hotel and maybe it just needed to be transcoded to something smaller, but infuse is direct play only so I tried on other iOS apps like jellyfin and swiftfin. Nothing worked. I tested a few other quality files and none of the 4K files would transcode but the lower quality videos that could be direct play work fine, until I try to force transcoding.

Not sure if the hardware acceleration just suddenly broke or if it’s even the root of the problem. But attached are my jellyfin logs and my ffmpeg logs.

Seems like “FFmpeg exited with code 187” might be the problem?

https://pastebin.com/caXq5Tk9

https://pastebin.com/B4FQWW0s

Please let me know what else I can share to help troubleshoot.

[TheDreadPirate]

Reboot the host? I've seen users report LXCs losing access to the GPU. I can't remember if there was a root cause found or permanent solution.

[Jdo139]

Reboot the host?  I've seen users report LXCs losing access to the GPU.  I can't remember if there was a root cause found or permanent solution.

I rebooted the proxmox node and I don’t think some of the containers auto started and I can’t access proxmox right now until I get back home tomorrow night. Will update tomorrow and see if that fixed it. 

I do vaguely remember seeing somewhere that there was a known glitch with Nvidia passthrough that would require reinstall of drivers on host and on the container.

[Jdo139]

Reboot the host?  I've seen users report LXCs losing access to the GPU.  I can't remember if there was a root cause found or permanent solution.

Restarted everything and still same issue. Hardware transcoding no longer working, and certain 4K files won’t play even in direct play. I did notice that the files that won’t direct play also don’t have any media info attached to them either.

[TheDreadPirate]

Did a storage volume detach or something?

Regarding the GPU passthrough, I've never used LXCs nor Nvidia GPUs with Linux so my ability to provide guidance is limited.

[Jdo139]

Did a storage volume detach or something?

Regarding the GPU passthrough, I've never used LXCs nor Nvidia GPUs with Linux so my ability to provide guidance is limited.

Figured it out. Apparently when the proxmode node or container updates the mapping of the devices get screwed up. I used this guide that I found to remap the LXC's config files as the numbers had changed after an update. Frustrating if this will have to be done every time after an update.