Jellyfin crashes during playback with no evidence in logs

[hugefinfan]

Ah, spoke too soon. This just crashed on chrome again:

Chrome console log:

Code:

HLS Error: Type: networkError Details: fragLoadError Fatal: false main.jellyfin.bundle.js:2:69163
hls.js response error code: 502 main.jellyfin.bundle.js:2:69369
Requesting url without automatic networking: https://sub.domain.tld/Sessions/Playing/Progress node_modules.jellyfin-apiclient.bundle.js:2:14622
Requesting url without automatic networking: https://sub.domain.tld/Sessions/Playing/Progress

[TheDreadPirate]

Is your router or something just going bad? Bad cable? Dropped packets?

[hugefinfan]

No. I have everything hardwired and a full ubiquity pro stack setup in a rackmount setup.

Interestingly, back on version 9 of JF I used to be able to reboot the docker container and the video cache would allow it to keep playing through the (20 seconds or so) reboot. Now when I just tested that today the Roku keeps playing through the cache but on the browser it crashes immediately. Is the cache not client-side when watching in the browser?

I find this strange because if even the slightest network jitter caused JF to crash wouldn't everyone have this problem all the time making web playing essentially unusable? Incredibly I frequently stream jellyfin on airplanes! Talk about a tenuous network connection of dropped packets....

[TheDreadPirate]

Are you using custom MTU sizes? Jumbo frames? Separate VLANs for clients and server? Special routing rules?

[Fate]

Are you using custom MTU sizes?  Jumbo frames?  Separate VLANs for clients and server?  Special routing rules?

Adding to what TDP said:
duplicate IP Addresses? asymmetric routing?

@OP
can you create a network capture with wireshark and recreate the crash?

[hugefinfan]

Are you using custom MTU sizes?  Jumbo frames?  Separate VLANs for clients and server?  Special routing rules?

This forum doesn't host images so I will have to just answer these from my dashboard:

No custom MTU sizes, yes to Jumbo frames, no to separate VLANS. I'm not sure what would qualify as special routing rules but I have no NAT or static routes set up, only the port forwarding through CloudFlare SSL to the Nginx container.

I will look into doing a packet capture and try to grab something there.

[TheDreadPirate]

Also disable Cloudflare proxying. As stated earlier it is against their TOS to server video through their proxy (they have a PAID tier for that), and we also want to eliminate that as a factor since we have no control over it.

[hugefinfan]

Also disable Cloudflare proxying.  As stated earlier it is against their TOS to server video through their proxy (they have a PAID tier for that), and we also want to eliminate that as a factor since we have no control over it.

On 10/31 I connected directly from my browser to the local IP (see my post from then on page 3 of this thread). By bypassing the DNS and SSL certs I got this error when it crashed:

Code:

Requesting url without automatic networking: http://10.xxx.xxx.xxx:8096/Sessions/Playing/Progress node_modules.jellyfin-apiclient.bundle.js:2:14622

HLS Error: Type: mediaError Details: bufferStalledError Fatal: false main.jellyfin.bundle.js:2:69163

Requesting url without automatic networking: http://10.xxx.xxx.xxx:8096/Sessions/Playing/Progress

[hugefinfan]

I just wanted to thank everyone for their help.

I have been running version 10.10.1 for a while now with no problems, even running it through my CloudFlare DNS and Nginx proxy without issue. Long story short I do not understand where the issue came from or what resolved it.

If I had one wish it would be for this problem to have shown up in a log somewhere and to understand a bit more about why (if this was indeed a network blip) the caching did not allow for effortless recovery.