Yeah having PC backups is important. I use Veeam as its free. I will say that just because a version update identifies database corruption does not mean that it caused it. I simi-broken database can limp along for a while before an update comes thru and needs to modify part of the broken section of the database, causing visible issues.
I downloaded an older version of Plex Media Server from here: Older versions of Plex Media Server (Mac) | Uptodown
I used the 1.25.8.5663 Apr 8th, 2022 version. All I did was drag the old version app to the Applications folder and overwrote the faulty version. Everything is now working perfectly again.
Yeah, experienced this exact problem on Media Server Version 1.27.0.5897 for OSX, immediately after updating. The Media Server could not recover, was not serving files, and could not rescan files after the db corruption.
I got back up and running by deleting all of the db files and installing an older version of PMS (1.26.X). But losing all the tracking data of what episodes have been viewed etc. is a pretty steep price to pay on such a large library.
I could find no dated backups of com.plexapp.plugins.library.db in the directory⦠but I did find a few recent backups of com.plexapp.plugins.library.blobs.db (not clear on what the latter contains) and tried to use itā¦
Was eventually successful mating the older PMS version to database files from a few days ago via Time Machine backups.
This worked immediately and was very easy.
Looked at the logs from @spawn666 and @plex_movix -appear to be different issues - in both cases debug logging was disabled so logs whilst do show the error - they do not show the sequence that led to the errors
In the case of @plex_movix which was on docker running 1.27.0.5897-3940636f2 it seems that the database file was incomplete - missing a table - do you still have that db file ?
In the case of @spawn666 - the logs again have no debug logging and show that the server on 1.27.1.5891-c29537f86 on a Mac. It appears to have run on this version without logging any corruption between 13:01 and 13:03 on the 15th and the corruption was logged after restrart on the same version - may be a db update was still running and got interrupted when the server was restarted - without debug logging it is difficult to say
On some updates where the db gets changed - the database gets optimized and this may take time
Please do run with debug logging enabled and attach the server logs when issues arise
Wow. Youāve been reading this thread and thatās your response?
Are there any professionals working for Plex that we can talk to?
Without diagnostics it is impossible to troubleshoot issues that do not arise in our testing or usage - my guess at the moment is that there is a vacuum being run on the database during startup that is taking a long time on some databases and on some platforms and aborting or restarting whilst this is taking place and the database update is incomplete, could lead to a corrupt database / we are changing some of the startup database update logging to be INFO to help when debug logging is disabled -
So would need to see debug logs and it may help to also have copies of the corrupt database and earlier backup copy to troubleshoot
Over the years Iāve seen sa2000 professionally help many Plex users solve their problems. What have you contributed other than snark?
I am the tech lead on an app that has ~20M users. If we launched a build that had 6 people complain that the update had deleted their configuration and prevented the app from working we would:
- immediately pull the update and create a new update based on the previous version.
- gather logs where we can.
- assign a senior developer to investigate.
- try to replicate the error in-house.
Thatās how a professional organization handles it.
I started a similar thread and I mailed my actual db to MovieFan.plex (as requested). I can repro this issue all the time, if I update to 27, no content and if I go back to 26 it is all normal.
agreed. I see several threads about this, so this doesnāt seem to be some rare instance. There should be a major warning to users (mac) to not install 27 yet.
I know thats how an organisation does it, since iāve worked in IT over 20 years. I take the time to report this issue to notify you, because I know it may be important. Maybe its a one-off, maybe its bigger, only you can review that on your level. Ive already troubleshooted and called in an sql expert before posting. A big problem usually starts with a couple of users and could be an indication that there is a bigger problem as you will certainly know.
Barking down at users like complaining little children with tiny problems is pretty rude and not what I expected. We are just trying to help and maybe get a bit of help. Since there is no more direct way to log a ticket we can only post on the forum in the hopes that someone helps. Sorry if our problems have inconvenienced you in any way.
Thank you - following that up with the devs
Could anyone experiencing issues starting up PMS after updating to 1.27.0 or 1.27.1 please send me their server logs and PMS database files, including:
- At least one backup database from before the issue started occurring
- The current database file
- The corresponding .db-wal and .db-shm files
Thanks much for your assistance in investigating this.
Thanks, @tramp78 I had the exact same thing happen to my server. I just downloaded the previous version and Iām running a library scan now. I hope it fixes my server as well.
@d1617 can you send Ridley a copy of the information they are requesting?
I can try but my server appears to be pretty messed up. I tried restoring back from two prior backups and still having issues even after rolling back.
Scratch that, some reason my restores are not working.
Thanks to the database file shared by @d1617, I was able to diagnose an issue in the schema migration process and implement a preliminary fix, which Iāll share here as soon as the build is ready.
Please note that itās not clear whether the issues other people in this thread are seeing are the same as the one @d1617 ran into, so it would still help to see more db files that people are having trouble with.
Please verify whether the issue still occurs when using the build linked below for your platform. Note that the server may take several minutes to start up the first time on a new version with database schema changes (such as this one), particularly if your database is stored on a rotational hard drive instead of an SSD.
Desktop Platforms
-
Windows
-
Ubuntu/Debian
-
RedHat/Fedora/CentOS
-
FreeBSD
-
macOS
NAS Devices
-
Asustor
-
Drobo
-
Seagate
-
Synology (DSM 6)
- ARMv7_Neon (x15 Series (excluding DS115j, RS815), x16 Series (excluding DS216se), x17 Series, x18 Series, and DS414j)
- ARMv7 (x13 Series, x14 Series (excluding DS414j), DS115j, RS815, and DS216se)
- ARMv8 (x18 Series, x19 Series, x20 Series)
- Intel - 32-bit (x10 Series, DS415play, and DS214play
- Intel - 64-bit (DSM 6.0 or newer)
-
Synology (DSM 7)
-
QNAP
-
Unraid
-
Netgear
-
Western Digital (OS 3)
-
Western Digital (OS 5)
Ive tried to do a rollback. Server didnāt want to start, no clue why. I also tried to recover/repair the db but all recover attempts resulted in empty recovery db.
Iām rebuilding from fresh now.