Since the latest update, Plex has been a complete dog - slow load times for music in the web browser and iOS displaying the error “couldn’t retrieve the play queue for this item.”
I’ve tried deleting statistics, as mentioned in another post and combined that with a database rebuild. The library displays at a normal speed and nothing seems to be going on with the machine itself. I see errors like this in the logs:
Dec 06, 2019 05:30:05.653 [14300] ERROR - Soci Exception handled: sqlite3_statement_backend::loadOne: database is locked
Dec 06, 2019 05:30:05.835 [1280] WARN - Waited one whole second for a busy database.
Dec 06, 2019 05:30:06.088 [16380] WARN - Held transaction for too long (…\Statistics\StatisticsMedia.cpp:29): 0.234375 seconds
Hadn’t noticed that option before in the app. I cleared the cache there but still getting the same error. On the browser, it’s playing but it’s slow to load between songs.
One thing that I don’t recall being high before is ‘power usage.’ Also, before I can say it was 100% Plex, I see that I had a Windows 10 update deployed on Nov 15 that had a bunch of junk. Not sure if that was exactly when my problems started.
Nothing added in the past 6 months. On the pictures side, I removed a few folders and it’s saying that I have 62k pictures but I don’t have anything displaying those currently.
I’m not sure if it’s metadata slowness per se. The library lists themselves, tracks, etc. all load quickly. It’s only during playback where there is an excessively long pause and then, eventually, it starts to play. On the iOS, I suspect that it’s giving up much faster and throwing the error. Once I start to play an album, the song will play, minutes-long pause between tracks and then the next one loads.
Fixed it - sort of. Another bomb from a Plex update. I downgraded to 1.18.1.1973 and now I’m not seeing the errors in the logs anymore. None of the StatisticsMedia.cpp errors, no database locked errors, no database errors at all. I’m not sure if I can draw Plex support to this post but there’s something wrong with the latest update folks - skip 1.18.2.2058, possibly 1.18.2.2029 as well. I’m not sure which caused but I’ll wait for a fix to be posted.
I’ve had these exact symptoms, include the errors in the logs: ERROR - Soci Exception handled: sqlite3_statement_backend::loadOne: database is locked AND also the Held transaction too long errors.
I can’t pinpoint exactly when it started, but it was probably sometime in December when I upgraded to 1.18.2 or a version around that. Here is what I know of this issue:
I think it impacts all clients, resulting in shows/movies SOMETIMES taking a long time to start or fail with the “couldn’t retrieve the play queue for this item”
I was seeing these exact symptoms with PMS 1.18.5.2260 yesterday. However I have not restarted my server in a long time, so this morning I applied PMS 1.18.5.2309 and restarted so maybe that will help. HOWEVER I’m already observing Held transaction for too long error in the logfiles after this restart.
My PMS server has an Intel Xeon Processor E3-1246 (4 CORES, 3.5GHZ) and 32GB OF RAM. It is running Ubuntu Server 16.04 LTS (will be upgrade to 18.04 LTS soon), and is pretty much dedicated to plex. All my media is stored in a ZFS zpool, however my plex DB is stored on an SSD volume (ext4). The server is never really that busy.
Something is not right here, I’ve used plex fine for many years before this issue. I’m hoping the reboot fixed this, but wanted to post here now in case it has not.
