Plex needs restart before playing certain media.

server-linux

#1

Hi there,

Was wondering how I can go about diagnosing an issue I keep having;

At random points during the day Plex refuses to play certain media. PlexPy will show the client Buffering and Transcode speed at 0.0. If I restart PMS then it works fine.

Server running Ubuntu 14 - PMS 1.9.3. Has happened on versions from 1.8.x upwards.

Around 80% of my media is mounted via plexdrive/EncFS - but I am sure it has happened to local media too.

What settings do I need to apply to capture sufficient logs for me to start diagnosing this issue?

Cheers!


#2

I will need to see a full set of log files (Settings - Server - Help - Download Logs) captured immediately after you've experienced this.

In the interim, I would test your fuse mounts (plexdrive) . Sockets being closed will manifest as you're describing


#3

I'll grab a set of logs when it next happens - however there is nothing on the fuse mount (plexdrive) logs. Also I can literally restart PMS and play the exact same media file - and it's pulling it from the RAM cached chunk and Plex plays it fine :/


#4

If PMS plays it just fine after the initial hit then the media isn't 'available' when PMS goes for it but is available after it actually does read.

Something in that mount is taking too long to complete and results in a read i/o failure


#5

Plex never says the media isn't available - it just sits there at 0x transcode speed.

Restarting PMS (without touching the mount) causes it to work

Additionally, if I download the file via the Plex Web interface it downloads fine.

Edit: just making sure I don’t come off as rude. I’ve talked to the plexdrive guys too and upgraded across three versions now and it persists. Additionally like I said I’m pretty sure it happens with local media too.


#6

Ok - happened again, although this time I did some testing whilst it was "Broke"

Media on Fuse mount - doesnt play
Media on local disk - doesnt play
Sync items - fail
Sync also gives me an error reading sync database (error on iOS 11.1 latest client)

Logs attached


#7

I noticed these;

Oct 06, 2017 18:03:04.644 [0x7f6a18fee700] WARN - Waited one whole second for a busy database.
Oct 06, 2017 18:03:05.651 [0x7f6a18fee700] ERROR - Failed to begin transaction (../Playlists/PlayQueue.cpp:852) (tries=9): Cannot begin transaction. database is locked

Testing my disk read/write I get;

/dev/sda:
 Timing cached reads:   13270 MB in  2.00 seconds = 6640.48 MB/sec
 Timing buffered disk reads: 260 MB in  3.12 seconds =  83.43 MB/sec

However there is a conversion occuring at this time;

Total DISK READ :      20.71 M/s | Total DISK WRITE :       2.23 M/s
Actual DISK READ:      14.73 M/s | Actual DISK WRITE:       0.00 B/s
  TID  PRIO  USER     DISK READ DISK WRITE>  SWAPIN      IO    COMMAND
20463 be/5 plex        5.98 M/s    2.23 M/s  0.00 % 17.24 % Plex Transcoder -codec:0 h264 -codec:1 ac3 -i ~5/7b8efb9d-0a9e-417d-be14-c1637c5d4df7/progress
20480 be/5 plex        0.00 B/s    0.00 B/s  0.00 %  0.00 % Plex Transcoder -codec:0 h264 -codec:1 ac3 -i ~5/7b8efb9d-0a9e-417d-be14-c1637c5d4df7/progress
    1 be/4 root        0.00 B/s    0.00 B/s  0.00 %  0.00 % init nomdmonddf nomdmonisw

#8

Thank you for those logs. I am on a priority task and will reply as soon as possible.


#9

Before I step away, there are some things you will need to do.

A. Optmize the database

Oct 05, 2017 15:51:35.360 [0x7f6a55fff700] WARN - SLOW QUERY: It took 430.000000 ms to retrieve 39 items.
Oct 05, 2017 15:51:35.400 [0x7f6a4dbff700] WARN - SLOW QUERY: It took 270.000000 ms to retrieve 39 items.
Oct 05, 2017 15:51:35.406 [0x7f6a557fe700] WARN - SLOW QUERY: It took 240.000000 ms to retrieve 39 items.
Oct 05, 2017 15:51:35.436 [0x7f6a4d3fe700] WARN - SLOW QUERY: It took 230.000000 ms to retrieve 39 items.
Oct 05, 2017 15:51:35.459 [0x7f6a587ff700] WARN - SLOW QUERY: It took 510.000000 ms to retrieve 1 items.
Oct 05, 2017 15:51:35.466 [0x7f6a54ffd700] WARN - SLOW QUERY: It took 370.000000 ms to retrieve 39 items.
Oct 05, 2017 15:51:35.468 [0x7f6a4e800700] WARN - SLOW QUERY: It took 660.000000 ms to retrieve 39 item

B. Increase the amount of entries which can be placed in the directory iNotify table (you have more than the default 8192)

Oct 05, 2017 15:52:28.160 [0x7f6a6e3ff700] ERROR - [Notify] Failed to add watch for /home/Plex/ACD/Plex/TVBoxsets/Devious Maids (28: No space left on device)
Oct 05, 2017 15:52:28.161 [0x7f6a6e3ff700] ERROR - [Notify] Failed to add watch for /home/Plex/ACD/Plex/TVBoxsets/Devious Maids/Season 4 (28: No space left on device)
Oct 05, 2017 15:52:28.162 [0x7f6a6e3ff700] ERROR - [Notify] Failed to add watch for /home/Plex/ACD/Plex/TVBoxsets/The Good Fight (28: No space left on device)
Oct 05, 2017 15:52:28.164 [0x7f6a6e3ff700] ERROR - [Notify] Failed to add watch for /home/Plex/ACD/Plex/TVBoxsets/The Good Fight/Season 1 (28: No space left on device)
Oct 05, 2017 15:52:28.169 [0x7f6a6e3ff700] ERROR - [Notify] Failed to add watch for /home/Plex/ACD/Plex/TVBoxsets/The Middleman (28: No space left on device)
Oct 05, 2017 15:52:28.170 [0x7f6a6e3ff700] ERROR - [Notify] Failed to add watch for /home/Plex/ACD/Plex/TVBoxsets/The Middleman/Season 1 (28: No space left on device)

This is becoming increasingly common. I've created a Linux Tip to show how to increase the kernel table size
https://forums.plex.tv/discussion/290621/increase-the-number-of-directories-linux-can-monitor-notify


#10

I'll re-run Database Optimisation but I ran it like 2-3 days ago.

The other bits I will also look into.

Let me know if anything else comes up in the logs.

Cheers


#11

Hi again, any update?


#12

Have you dated to current release?

Question: Have you updated to current version? When you have I will need updated log files


#13

I'm still on 1.9.3, but have had this issue for the last 4/5 versions.

Should I upgrade to 1.9.4? I didn't see any bugfixes regarding my issue, and I'm not using Live TV.


#14

Upgraded to 1.9.5 - I'll update this post if/when it happens again.

Did the logs I sent over show anything other than the db timeouts?


#15

I have the same issue. Did yo ever figure out a fix.