Server unreachable even in bundled web app

Server Version#: 1.28.0.5999-97678ded3
Player Version#: Latest Web Player

I updated my NAS to the latest Plex Server update today and ever since the crashing/going offline issues I’ve been having for the past week have gotten worse and worse. Anywhere from 5 - 15 minutes after restarting the server, the server goes offline to the internet, the LAN, and even itself. The process is still running and uses CPU time for sonic analysis, but the server cannot be accessed. Allowing insecure connections allows connections to continue for a few more minutes before it eventually faceplants again. Any suggestions?

Can you post your server logs? sounds like a corrupt database.

Plex Media Server Logs_2022-08-01_21-46-02.zip (2.7 MB)

I am seeing a lot of (28: No space left on device)

You’ve exceeded the number of directories Linux can monitor (the error message says “no space left” but in reality the inotify table size has been exceeded).

Aug 01, 2022 21:45:13.665 [0x7ff3223d0b38] ERROR - [Notify] Failed to add watch for "/volume2/Music/Cdot Honcho/Cdot Honcho - OTW" (28: No space left on device)

See FAQ 16 for the procedure to increase the table size.

FAQ 16: Increase the number of folders Plex can monitor

Note that this is a Linux, not Plex, setting. You may have to change it again when Synology issues DSM updates.

Thank you; this seems to have fixed the issue. Though why is there such a limit in the first place?

The default table size is either 8192 (older kernels) or 65535 directories on newer kernels.

There is a limit because this is Non-Paged, Non-Releaseable, Non-Swappable, KERNEL memory.

It makes no sense to let the kernel freely grow to consume all available host memory

Gotcha. What about Windows and macOS? Do they have similar watched folder behavior?

@GlacierIsland

I do not know anything about Windows.
MacOS came from Unix. Linux came from Unix. That is the extent of my MaxOS kernel internals knowledge LOL

Knowing what I do know of operating systems, it is logical for kernel tables to have size limits somewhere for the same reason Linux as these default limits.

Are you aware – You can increase the size of the inotify table on your machine?

As a secondard consideration, media storage with lots of unnecessary nested directories waste these precious slots … Curation of directory structures is important

e.g. /nas/Movies/Movie.JUNK.4K.UNKN/Movie (year)/Movie (year).mkv

Top of the library is /nas/movies.
this example wastes 1 slot. “Movie.JUNK.4K.UNKN”.

If all movies are this way, there is 50% directory waste.

Photos and Music are often the types which consume the most notify slots.
They tend to be over compartmentalized quite frequently

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.