Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[markwrc]

Hi,

I've been having ongoing issues with streaming video over Media Network to my LG WebOS TV and both my iOS Devices via JRemote (iPad and iPhone XS Max) It begins playing just fine then after 2-3mins my entire PC locks up and crashes MC24 or my entire PC with a WHEA_UNCORRECTABLE_ERROR.

Oddly enough streaming audio, no matter the quality is totally fine and has no crashing issue. I've already tried letting MC24 roam free from AV checks and also done the same with my audio/video files; however I've never had to do this previously and it always worked fine with no issue. I've also experienced no issues streaming directly from Media Library (Windows 10; main pool) to Media Library (Mac) for some reason this does not crash my main machine (W10), both are on MC24.

I should mention that this started to happen recently; now I'm a Windows Insider (Fast ring) and Apple Dev so I'm always on the beta OS's. I could say this started on 1809 but I'm not exactly sure. The build I'm currently on is 1809 - 18262.1000. Looking and the crash dump for my BSOD it seems to flag up ntoskrnl.exe and hal.dll. In terms of system stability its fine and has all the latest drivers available.

Since running Media Centre - Media Centre doesn't cause any issue for the host PC, which leads me to think its a DLNA issue. DLNA is currently set to default with it playing the full original files only (no transcoding).

Anyone have any ideas to whats crashing it?

~Mark

[Hendrik]

WHEA_UNCORRECTABLE_ERROR basically indicates a hardware failure, software really cannot cause it directly. So its either a hardware problem directly, or a driver issue.
I would recommend to monitor CPU temperatures while streaming just to make sure nothing unexpected is happening, and also verify that all drivers are up to date. Generally using Windows Insider on production systems is not advised however.

[markwrc]

Yup I know, but MC is currently the only thing causing the issue, and while its streaming is the only time it happens. Can playback anything all day all night on the machine and not have an issue, same with audio streaming; but video? it crashes after 2-3 mins of playing on an external device.

My PC works as a renderer using AVX2 loads, so if it was a CPU/mem issue then it would have shown up. Temperatures are non-issue since the system is custom with custom water cooling, its also an 18c so i doubt MC streaming would cause any issues for it. I've been an insider since W10 Pro launched and never had an issue like this.

I might try updating the ethernet drivers but if its streaming audio fine and streaming video fine to another MC player on the LAN then i dont see it being that.

[markwrc]

Just looked at my hardware logs; seems that its spiking the CPU usage to 100% upon selecting a video to stream then crashing the machine after a few seconds, odd... does MC24 normally do this or is it some sort of bug/memory leak?

[Hendrik]

Its quite normal to get to 100% load at least briefly when loading up a video for streaming.

[BradATIMA]

Are you doing any overclocking?

[markwrc]

System is OC'd yes but has been completely stability tested for hours under a near 500W continuous load, but ill take it off and try anyway and report back

[JimH]

Good guess, Brad.

[markwrc]

system is stock now, video playback crashed while my CPU has been chugging away at 60-100% load and my GPU at 35% load (GTX1080) System hasn't crashed however but that can't be right surely, if it is its incredibly inefficient - my system is eating up 600W in power consumption just trying to stream a 1080p mkv?

[markwrc]

I'm reporting back again as i have tried other things...

1) I tried re-installing MC24 on my server and restoring the library with no settings restored and it still does the same thing - crashes after 3 mins or so of video streamed, System doesn't crash. MC24 does crash & is non-responsive with a constant CPU load of 70% usage by MC24 alone, along with 30% GPU usage.

2) I've tried it on my custom gaming PC (8 core) different build of Windows but its also 1809 and same version of MC24; again with all the latest drivers to have the exact same problem, only MC24 is responsive with no crash but is pulling a constant CPU load of 80%, with a similar GPU load.

It appears that its still streaming the data to the external device but the video locks up and scrolling though the video doesn't to make it play again - it just shows a paused frame, play/pausing seems to work but jumps towards the end of where it last streamed the data (say crashes at 3min then upon a play/pause it will jump to 50mins or so) and again it will crash the video and carry on streaming the data.

Any thoughts now?

[swiv3d]

I have just streamed a movie from MC 24 to my LG Tv with no problems. Windows 10 1809 64 bit.

[markwrc]

Interesting...

Anyway; i have done some more tests;
 
1) I tried once again to put my system to stock and still the video playback on the remote device crashed - Host PC continued to stream the data; either MC24 crashed or it BSOD'd my system with (again) the same .dll's being flagged.

2) I tried MC23 and had the exact same issue.

3) I changed the hardware of my system to what i knew worked months ago (yes this has been an issue for me for a few months) - Google said that hll.dll normally crashes due to RAM sub-timings or Windows - BIOS polling so i swapped my RAM out and the exact same issue happened and every time its after 3-5mins of video streaming.

4) I updated the BIOS to a beta one - which has some patching for the Spectre/Meltdown bug (but system isn't patched against this fully cause' Asus havent brought out an official BIOS since Oct 17') Exact same issue, made no difference.

5) I re-installed Windows (Build 18262) and ran the latest version of MC24 (24.0.60) the exact same issue had happened and this was with a fresh install, with nothing else installed but MC24 and the drivers for my machine - system was stock with its BIOS settings.

6) Streaming video from my Mac using my host PC as a Library Server streamed fine with no issues - showing that its not a LAN or Networking issue with the playback device and the host PC - also shows that its not a problem with corrupt data on the storage drives or the my RAID arrays etc.

I still have a huge disbelief that this is a stability issue with my PC - yes its custom, yes its overclocked... but if 14hrs of an AVX2 load doesn't crash it, or hours of Small FFT Prime95 doesn't crash it; why is JRiver the only thing crashing it with what appears to be driver crashes?

[Spike1000]

In this fix for 8.1 it specifically mention the UFEI BIOS Overclocking setting - which I sort of didn't expect. Might be relevant. . .

Spike

https://windowsreport.com/fix-whea_uncorrectable_error-windows-8-1/

"Solution 1 – Disable Overclock in BIOS

The most common fix for whea_uncorrectable_error is disabling overclock from your BIOS. To do that, just follow these steps:

    Open Start Menu and go to Settings.
 
    Go to Update and Security, and then to Recovery tab.
    Click on Restart now in Advanced startup and your computer will restart.

    On Advanced startup, go to Troubleshoot, then Advanced Options and then UEFI Firmware Settings.
    Click Restart.
    After the computer is restarted, it should open BIOS automatically, so go Advanced tab.
    Navigate to Performance, and look for Overclocking.
    Make sure overclocking is disabled.
    Save changes to BIOS (by pressing F10 on your keyboard), and start your computer normally.

After disabling overclocking, you shouldn’t face this issue anymore, but if it’s still present, check out one of the following solutions."

[Hendrik]

A mixed workload like video encoding can be quite a different category of stress then artificial load that only targets one specific function of the CPU. Especially if also the graphics card is involved for decoding the source video.

Generally speaking, BSODs are either a driver, or hardware. There is no third category. In this case probably hardware.
If I can write user-mode software to cause such a high load that your system BSODs, then by definition your system is not stable.

If you are feeling adventorous, you could try encoding a video with FFmpeg directly, which is what Media Center uses to serve a video to you.

For example, this commandline reflects roughty the process we do when streaming in 720p ("High" preset)
You can get a binary from here: https://ffmpeg.zeranoe.com/builds/win64/static/ffmpeg-4.0.2-win64-static.zip

(Replace <source> with an actual filename to a video for testing)

Code: [Select]

ffmpeg -hwaccel dxva2 -threads 1 -i <source> -vf scale=1280x720 -c:a aac -b:a 192k -c:v libx264 -profile:v main -preset:v faster -crf:v 20 -b:v 5M output.ts


[markwrc]

When i mean stress testing - i mean everything not just the CPU, I've tested and verified the GPU, RAM, CPU, CPU's Cache & the CPU's FPU, both with rendering (images/graphics benchmarks) loads and compute loads - nothing came up. I used to do extreme overclocking a few years back so I know how to test these systems to destruction & I do know that they can degrade over time - however this isn't the case.

I will re-iterate once again - currently the system is NOT overclocked, its at stock settings, stock memory timings & stock freq timings (not even XMP applied) only thing I've changed in BIOS is to activate RAID mode and turned off CPU fan speed monitoring.

I also just swapped out the GPU to my spare GTX980 and the same issue happens.

I've tried FFmpeg before as its a HWBot benchmark as is some x265 encoders. both run fine on my system with no crashes, BSOD's or hangs. I just tried FFmpeg with your settings w/ a 540p TV ep and it is currently having no problems with a 4k mkv. Neither however use my GPU like JRiver does (30-50% util) though.