Updated to Ubuntu 18.10, PMS to most recent version, before starting new version deleted EAE folder. Still won’t play EAC3 vids.
FYI, I just installed Emby to make sure it’s not some sort of limitation with my system. EAC3 vids play fine.
Do you have anything ‘extra’ loaded in 18.10?
My VMs are bare distro packages + Chrome. Nothing else added?
I don’t use this as a desktop so there’s not a lot of crap, but it also has Firefox, SABNzbd, Couchpotato, Sickbeard, Subsonic, VLC, Emby (just installed the other day to test), and MusicBrainz Picard, but not much else. These vids play fine with locally with VLC & remotely with Emby. It’s just behaving as if it doesn’t have access to the transcode.
And it is not always reproducing. Sometimes I can go for few days without error, but simetimes I get error next day after restart (I jus need restart, no deletion needed). Maybe we need some flag to debug EAE better.
At first I was thinking it could have something to do with my low entropy, but I fixed that and error still happens.
Now I think it could have something to do with swap, as I get some swapped momory time to time for containers. But that’s just a guess.
I had a similar issue on my Ubuntu 16.04 server. There were inotify errors in my log. If you are running linux run this command as root. Should help a bit. No restart needed.
sysctl -w fs.inotify.max_user_watches=100000
If you are running out of max_user_watches, the more permanent solution is to add it
sudo echo fs.inotify.max_user_watches=YourValue >> /etc/sysctl.conf
sudo sysctl -p
Common values for YourValue are multiples of 32K: e.g. 32768, 65536, etc.
The Linux default is 8192. This is why the table is easily exhausted.
One entry is allocated for each directory “watched” by PMS, including subdirectories.
Each Notification event table slot allocated locks 540 bytes of kernel memory.
1,000,000 notified = 540MB of locked (used) kernel memory which is not available to programs.
1 Like
Holy crap that did it! Tested Plex/Web & Roku so far. Thanks a lot. This was a pesky one.
The default kernel table size is 8192 .
When we have 30,000 directories, it’s a “bit” of a nuisance. 
I have mine set to 65535 and never have an issue
1 Like
I don’t get “inotify” errors in my logs, should I try this fix still?
I need to see the up-to-date logs of this happening to be certain.
Please recreate, using DEBUG logging, and attach the ZIP file.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.
