Jellyfin
Replies: 17 comments 11 replies
|
This sounds less like a browser/client issue and more like Jellyfin is blocking on slow filesystem access to the SMB share. If the whole dashboard becomes unresponsive while starting playback, I would first check whether Jellyfin is waiting on network I/O, metadata probing, subtitles, trickplay/chapter extraction, or transcoding startup. A slow or blocking SMB mount can make the server feel “locked up” even if the network itself is 10 Gb. A few things I would test:
docker exec -it jellyfin bash
time ls -lah /media/path/to/movie
dd if="/media/path/to/movie/file.mkv" of=/dev/null bs=64M status=progress
The key comparison is not only “Plex/Emby work on the same share”, because Jellyfin may be doing different probing/transcoding/subtitle/library operations at playback start. If the same file works instantly when Jellyfin is installed on the main server, but stalls only when Jellyfin accesses the remote share, then I would focus on the Jellyfin server’s access path to the media: Docker mount, SMB/NFS mount options, permissions, latency, and whether Jellyfin is doing extra reads before playback. |
|
I had hosted everything on NAS - media and data and transcodes. After I realized all the round trips that forced Jellyfin to make I moved the database locally, mounted transcodes to tmpfs RAM only since I am I just a single user, and only media is remote. I now load instantly. The dashboard itself also loaded instantly after moving the database locally. Databases and NAS don't go well together. |
|
@ZD-FLoM @starkmarkus if i play video file on main server no issues but the remote cant... in unraid i setup a internal share to the host comp so the mount inside jellyfin docker goes /mnt/remote/share name in the end i did get it to work better.. i had to google a bunch.. the problem is jellyfin cant play videos directly properly like emby or plex.. seems jellyfin reads files differently and causes problems what you have to do is install Webdav on the main server.. and you gotta setup "rysnc" webdav then in jellyfin you gotta point to the webdav.. then the latency is pretty much gone i havent tested alot as i havent been home much.. but that seems to stop the freezing as for transcoding or not no transcoding just direct play in browswer... but you gotta do hurdles just to setup Webdav just to get Plex to work on a remote share as for my database and transcodes.. my transcodes it to /tmp so to the ram and the database for jelly is in the appdata folder so its on a nvme drive |
|
oh and then you gotta make a user script that mounts the Webdav path each time the array mounts so jellyfin can mount to |
|
That makes sense. If you go with the WebDAV workaround, I would make sure the WebDAV mount is created before Jellyfin starts or before Jellyfin scans the library. Otherwise Jellyfin may start while the path is still missing/empty and then cache or scan the wrong state. So the order should be roughly:
If this is on Unraid, a user script that mounts the WebDAV path when the array starts is a reasonable workaround. I would also make Jellyfin depend on that mount being available, or restart Jellyfin after the mount is established. For future readers, this confirms the workaround is not just “use WebDAV” but: mount the WebDAV path at array start, then point Jellyfin to that mounted path Still, I would treat this as a workaround for the remote filesystem behavior, not as proof that Jellyfin cannot direct-play remote media in general. SMB, NFS, and WebDAV can behave very differently for seek/stat/open patterns. |
|
@starkmarkus ya i cant seem to get google to reprint what i read i think it said dont quoute me but either jellyfin or its plex/emby they read files at a block level where the other doesnt and this is the cause of problems... as direct play will work over a samba share the webdav share will play instantly but when you use the weddav then it doesnt lock up the jellyfin... i honestly dont know why or how this weddav works for jellyfin etc ya its a work around but maybe jellyfin can adapt to it inside jellyfin for remote share so can you explain since you problably understand WebDav better then me but how is it different from the smb and why does it work better for jellyfin.. i only came across it by accident with constantl googling for help cuz i was getting frustrated how jellyfn just locked up where you just have to just X the browser and then when you try to reload the jellyfin site it wouldnt load it was hogging the prrogram or cpu like stuck in an endless loop till it just istelf broke loose.. |
|
@starkmarkus or is there a way in jellyfin if the mount isnt mounted dont scan? |
|
I would handle this mostly outside Jellyfin. Best approach on Unraid is probably:
Example idea: #!/bin/bash
MOUNTPOINT="/mnt/user/webdav_media"
CONTAINER="jellyfin"
# Mount WebDAV here.
# Use your real rclone/davfs command.
# Example:
# rclone mount myremote: "$MOUNTPOINT" --daemon
# Wait until the mount exists
for i in {1..30}; do
if mountpoint -q "$MOUNTPOINT"; then
echo "WebDAV is mounted. Starting Jellyfin..."
docker start "$CONTAINER"
exit 0
fi
echo "Waiting for WebDAV mount..."
sleep 2
done
echo "WebDAV mount was not available. Jellyfin will not be started."
exit 1If if [ -d "$MOUNTPOINT/Movies" ]; then
docker start jellyfin
fiI would also turn off Jellyfin's automatic startup in Unraid, otherwise the container may start before the WebDAV mount is ready. About the second question: I do not think Jellyfin has a perfect built-in option like "do not scan this library if the mount is missing". You can reduce the risk by disabling automatic scans / real-time monitoring for that library, but that is not the same as mount-aware scanning. So I would do both:
The important thing is to avoid this situation:
because then Jellyfin may cache or update the library based on the wrong state. |
|
i guess doing testing even with the webdav when you load up jellyfin the libarys load fast you can go into a video to play plays instantly probably seems if you exit using the back arrow couple times then it gets stuck circle spinning on the library folders if you go back to the folder like action folder all the libary posters are blurred out it takes a like 10 seconds to refresh itself so it doesnt keep them populated or whatever you wanna call it and when you go back further to the main page it also hangs with the circling so ya with the webdav work around works better for playing videos but backing out still hangs for the posters and the library main page posters.. but doesnt hang as long as a samba share |
that is my user script to start when array starts and i noticed if i run jellyfin on main server if i play a video and back out posters to videos refresh instantly and same with the library folders but in the remote server when it points to the main server.. the posters take up to 20 seconds to populate right away and same with locking up on the library folders but the libraries and folders are on the 2nd unraid server on a nvme drive so shouldnt they load instantly or are they scanning the network or something? |
|
i find the collections buggy using webdav on the remote i even got error as the circle spinning.. it just doesnt like remote.. if the collections on the main server its faster but few issues i found with the collections 1.. doesnt show up on the main page you gotta go into a library folder just to get it work.. and i still think collections should be also inside a movie.. so if you go into movie "48 hours" after the cast it should show collection "another 48 hours" but i figure all these posters should load instantly as its saved on the nvme like plex where when you go into any folder its instant loaded... unless jellyfin does it different i find plex seems to be the best to work over network,,, emby has issues and so does jellyfin... i guess really i need to have jellyfin on the main server running only reason i dont is and the logs for the above gui error i have and jellyfin is just broken now i going to try a vanilla jellyfin on my 2nd server.. incase these plugins i installed mess up for accessing webdav |
|
At this point I would simplify the test setup, because several different things may now be mixed together. The ChunkLoadError in the browser is probably not directly about SMB vs WebDAV. That usually means the Jellyfin web UI failed to load one of its JavaScript chunks. I would first try:
The log you posted also shows the PlaybackReporting plugin. Since you now have playback, UI navigation, collections, posters, and plugin logging involved, I think testing a clean Jellyfin instance is the right next step. I would test in this order:
That will tell you whether the problem is really the remote media path, or whether plugins / old config / web UI cache / collections metadata are adding a second issue. For the poster/library loading question: even if posters and metadata are stored on NVMe, Jellyfin may still do server-side checks or API calls when entering libraries, collections, or item pages. If the media path is slow/blocking, the UI can still feel slow because the server is waiting on backend filesystem calls. So “metadata is local” helps, but it does not fully eliminate remote-path latency. Your current findings sound like this:
So I would not call Jellyfin completely broken yet. I would isolate it with a vanilla container and no plugins first. If the clean instance still behaves the same, then the useful bug report would be much clearer: Clean Jellyfin, no plugins, local config/cache/db/transcodes, same file:
That would be much easier for Jellyfin maintainers or Unraid users to reason about than the current mixed setup. |
|
@starkmarkus ok i appreciate the feed back.. i currently running 3 library folders on webdav path i didnt try samba share.. with my slow 3megabit internet connection i find nothing against jellyfin as its free and people work hard.. but getting metadata doesnt work very well for someone with poor internet where plex does best and emby second.. not knocking it at all.. i guess they just work different.. so my internet is always hammered 100% that java script error never went aaway closing browser tab and re launching jellyfin didnt work i guess at the time plugins just crapped out jellyfin.. i had to restart the container to get it back.. as that was the last log errors too showing where jellyfin was just hanging.. all it did was spinning circle... i going to make a 3rd jellyfin where it uses the samba shares.. so i can get a somewhat accurate... so ill have 3 jellyfins if hard drives were cheaper prior this datacenter/ai explosion and big companys get the hard drives etc i could just made server just for hte media instead of meadia and my home files.. so nothing to speed up the metadata too bad you cant just turn off look serverside checks just load the metadata... does emby and plex read meta data differently cuz plex is instant? or is because it saves all the metadata in the appdata/plex-media-server and thats why it is so big it copys all the poster images there and jellyfin maybe doesnt? i dont know but ill do more testing and get the results |
|
and do you happen to know if its possible or not you know how plex while in the browser while it sit there if a video gets added the screen shows the added video or if you scan to fix metadata it fixes it there... but with jellyfin only way to see updates is hit F5 i do not know what you call that really and i dunno does jellyfin offer that style of refresh or something like that |
|
That test plan sounds good. Having separate instances for:
should make the result much clearer. For metadata: if your internet connection is only around 3 Mbit/s and metadata/artwork download is running, Jellyfin can definitely feel slow during scans. Poster/artwork download depends on remote metadata providers, so that part is not only NVMe speed. The local NVMe helps once metadata and images are already downloaded, but the first scan / refresh still depends on internet speed, provider response time, and how much artwork is being fetched. For the “why don’t posters instantly appear if metadata is local?” question: Jellyfin may still make backend API calls when entering a library, collection, or item page. If the server is also doing scans, checking files, loading collections, or waiting on the remote media path, the web UI can still show spinners even if the stored images are on NVMe. For the “Plex updates the browser automatically when new media appears” question: I think you mean live UI refresh / real-time library update. Jellyfin does have “Real Time Monitoring” for libraries, but it depends on filesystem notifications. On Linux that usually means inotify. Remote mounts such as rclone/WebDAV/FUSE and some network filesystems often do not provide those events in a way Jellyfin can use. In that case Jellyfin may not notice new files immediately, and you need a library scan or manual refresh. So for your WebDAV/rclone setup, I would not expect perfect Plex-style instant updates in the browser. I would separate the issues like this: Playback start: For your slow internet, you may also want to reduce metadata/artwork fetchers during testing. For example, start with one small library, one metadata provider, no extra image providers, no plugins, and let the scan fully finish before judging UI speed. If the vanilla WebDAV instance plays quickly but does not auto-update new files, that may simply be the limitation of real-time monitoring over rclone/WebDAV. If the vanilla SMB instance still freezes while local media does not, that points back to the SMB remote filesystem behavior. |
|
That test plan sounds good. Having separate instances for:
should make the result much clearer. For metadata: if your internet connection is only around 3 Mbit/s and metadata/artwork download is running, Jellyfin can definitely feel slow during scans. Poster/artwork download depends on remote metadata providers, so that part is not only NVMe speed. The local NVMe helps once metadata and images are already downloaded, but the first scan / refresh still depends on internet speed, provider response time, and how much artwork is being fetched. For the “why don’t posters instantly appear if metadata is local?” question: Jellyfin may still make backend API calls when entering a library, collection, or item page. If the server is also doing scans, checking files, loading collections, or waiting on the remote media path, the web UI can still show spinners even if the stored images are on NVMe. For the “Plex updates the browser automatically when new media appears” question: I think you mean live UI refresh / real-time library update. Jellyfin does have “Real Time Monitoring” for libraries, but it depends on filesystem notifications. On Linux that usually means inotify. Remote mounts such as rclone/WebDAV/FUSE and some network filesystems often do not provide those events in a way Jellyfin can use. In that case Jellyfin may not notice new files immediately, and you need a library scan or manual refresh. So for your WebDAV/rclone setup, I would not expect perfect Plex-style instant updates in the browser. I would separate the issues like this: Playback start: For your slow internet, you may also want to reduce metadata/artwork fetchers during testing. For example, start with one small library, one metadata provider, no extra image providers, no plugins, and let the scan fully finish before judging UI speed. If the vanilla WebDAV instance plays quickly but does not auto-update new files, that may simply be the limitation of real-time monitoring over rclone/WebDAV. If the vanilla SMB instance still freezes while local media does not, that points back to the SMB remote filesystem behavior. |
|
still testing but what i found so far testing across on jellyfins i setup 3 movie Libaries i did just the TMDB as the meta data and unchecked the rest /dashboard where you have scan all libraries restart shutdown
my internet is maxed out at 3megabit/s 100% and i only did 1 server at a time till my internet was released from jellyfin
|
|
and are you a bot some of the answers seem your not human that your a bot or AI |
|
Some of my replies were AI-assisted, but I’m not a bot account. I’ll step back from this thread so someone with deeper Jellyfin/Unraid-specific experience can take over. |
|
so when i did fresh installs and i let them all scan on there own like i mean i let 1 server do the scan till i seen in pfsense it wasnt using the internet anymore then i did the next server then 3rd i did only 3 movies,, action adventure cartoons, collection i find they work fast now maybe jellyfin was updating in the background that it buggers up while i try to play as i have poor internet... the posters seem to load fast now across all 3 and seems to play ok.. i had to figure out some videos dont play the audio i found under playback of my root user set from Auto to Stereo fixes it i guess jellyfin glitchs on auto mode cant work now it seems to work fine i dunno if the glitchs issues is it trying to update from the internet,, or the plugins, or cuz i have like 20 or so library folders when i set it up orginally |
|
my internet craps out too much to get posters to work right .. i guess plex offers a polished way i guess it updates the metadata until it finds the poster and stuff and doesnt move on till next till it finds it or a timetout... i find jellyfin not forgiving and just moves on.. to the next file for the poster... i guess you really need 1 gb internet is my finding so far |
|
so i tried only 1 plugin well not really plugin its the netflix style css @import seems to work but i find a video plays through windows media player ok... but through jellyfin the video jiggers on the screen and the audio too like how the vertical hold on old tvs would go up and down the specs of the video are guess jellyfin cant play this video? |
|
@starkmarkus so things seem better on my nothing installed jellys local or the remotes smb and webdav when i was testing on my full blown remote setup i get lock ups and circle spinning on the main page i get this .. so how i created it |
|
oh and is there a way drives me nuts since i have like +20 library folders is there no way to do the rescan for missing metadata that covers all |
|
and i guess cuz my interent is poor 3megabit and caps out i guess it fails alot with updaters? im gessing its cuz it cant connect cuz my internet maxed 100% |
|
so what i kinda seeing u probably know more |
|
when i try loading the collection page on my main remote jellyfin server i get this error it loads the collection page so slowl like takes a min or so before it populates it should be instant.. when i seem to semi get it work over the network not bad not the greatest then another issues pops up lol like dog chasing its tail lol |
|
i dont think collections likes network shares all i get in logs and collections keeps spinning won populate |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
so i have plex emby and jellyfin on a 2nd nas that uses a remote share both nas are unraid
and 10gig networking
they are all setup the same
if i load a movie on plex or a show it loads instantly
if i load a movie on emby or a tv show it doesnt load instantly takes like 20 seconds but loads
if i load a movie on jellyfin or a tv show it can take like a minute to load if you try to back out the who jellyfin is locked up you try to goto the dashboard thats locked up for a minute if you try to do any sub tabs cant do it,..
if you get the video to play it will play 10 seconds then stop then buffer or so then play then stop after like 10 seconds and buffer like 20 seconds before it play
how do you adjust jellyfin for samba share access to work instantly like plex or semi fast like emby is there some setting to turn on or off to get it to work? or does jellyfin not like network shares as if i install jellyfin on the main server it works fast and doesnt pause a video only network share
and for all 3 tests is i using MS Edge Internet Explorer as the client to watch the video no jellyfin desktop client as that app is broken
All reactions