Ooh, that’s a good spot. Looks like some files got moved erroneously (probably as part of the tedious file-moving when doing the Plex Dance). I’ve moved them back up a level. Thanks for that.
Doesn’t explain the McMafia thing, but certainly not ideal. Would be nice if Plex flagged up inconsistencies like this…
A further look I see many network errors/warnings(unable to connect, download, etc), to access rights(can’t read, can’t write), slow database queries, etc…
Also I’m seeing the dreaded get_iplayer alot. I’m not sure if LMA is disabled here, or if it is, PLEX will not honor the setting but for what ever reason it is PLEX is trying to match a show call get_iplayer
I had to reboot my NAS earlier today, so the read/write errors may be from that. But Plex definitely has the correct permissions, so that’s probably a red-herring. Will wait for all of the metadata to populate (it’s still processing after yesterday’s Dance) and then maybe re-do a Plex Dance for a couple of specific seasons, and make a note of the time when I started that, to give a ‘clean’ log.
I think you have many, many issues going on at the same time. This is not as simple as one or two shows not matching up. Just for comparison… My logs vs your logs…
My com.plexapp.agents.thetvdb logs: 102 errors from 11448 entries.
Your com.plexapp.agents.thetvdb logs: 1368 errors from 22059 entries.
My Plex Media Server logs: 381 errors and 37 warnings from 216856 entries.
Your Plex Media Server logs: 7462 errors and 27624 warnings from 37705 entries .
So how does one a) identify these errors and b) resolve them, without trawling through Plex logs? Is there some way of saying “show me stuff I need to fix with my configuration”? Or is it all guesswork and low-level digging?
It could, of course, be that I have a more eclectic/obscure set of shows for which metadata is missing…
Here is a thought:
Could it be that the NAS is so busy that it never executes the Bundle cleaning operation successfully?
Your logs are only covering a very short time span, which might mean that the server has had no time to finish any of its operations.
The Plex Dance must be performed after all library updates and metadata fetching has been finished. It won’t succeed when there is still anything like this going on.
I’d say: leave this box alone for a few hours, so any activity has a chance to finish.
Disable any periodical and automatic library updates.
https://support.plex.tv/articles/200289526-library/
Only after that, try performing the Plex Dance again.
And after the dance, add only one (smaller) show back, for now. Wait until Plex has fetched every bit of metadata for it, before issuing any other commands. (This may take several minutes.)
I’ll try it. But this is extremely tiresome. Is there a faster way to rebuild the library/cache/bundles etc from scratch?
Is there an interim solution - i.e., manually deleting (say) the Metadata folder from within the plex folder, to force the same effect as a plex dance without having to keep moving files about etc?
I’m tempted to just delete the entire Plex installation and reinstall, disable LMA, and then add the library and run the media scan, to see if that solves it.
@“Mark Otway” said:
I’m tempted to just delete the entire Plex installation and reinstall, disable LMA, and then add the library and run the media scan, to see if that solves it.
If you delete the whole library. Followed by an ‘empty trash’ and a ‘clean bundles’ afterwards, you’ll achieve the same effect.
For good measure, empty out the ‘agent cache’ afterwards:
https://support.plex.tv/articles/202967376-clearing-plugin-channel-agent-http-caches/
But still, this might be really an issue with a very busy NAS device.
If you really go through with it, I recommend to leave the Plex Server alone afterwards for 20 minutes or so.
When you are adding back your tv shows, start by doing it one at a time. Give it plenty of time to fetch metadata etc.
I’m going to try a full, clean install of Plex.
Adding back shows one at a time is not really practical, given I have 300+ shows sitting on the NAS.
Why would a busy machine have this sort of effect though? I can understand it taking a long time for shows to be picked up and metadata to be loaded, but it makes no sense at all that the NAS being busy would result in shows being mis-identified - that would imply a pretty significant issue/bug in the Plex server code.
Grrrr. Do a fresh install but now I can’t sign into the Plex server - it keeps redirecting me back to the Plex site and then showing me “Looking for Server”… :’(
@“Mark Otway” said:
Why would a busy machine have this sort of effect though? I can understand it taking a long time for shows to be picked up and metadata to be loaded, but it makes no sense at all that the NAS being busy would result in shows being mis-identified - that would imply a pretty significant issue/bug in the Plex server code.
Because if PLEX at one time mis-identified a show and the NAS did not complete a full purge of the series(s) then you are adding back the same content PLEX has already indexed, so it will think you just moved the content around and it will reuse the existing info.
That on top of the shows you had in other shows’ folders. All these will compound the issue.
No, I mean when building a complete new library in a fresh Plex install with no previous files at all why would a busy NAS make a difference?
I can’t answer that. I just know that 9000 errors and 27624 warnings is no mismatch issue.
Since I’ve been summoned 
How can I help?
Mark,
What language are you using? What’re the settings for your server? When I opened your logs, I was profoundly surprised. I know these are written in English.
灁㌰〲㠱〠㨸㠱㌺⸱㤴‴せ㝸㑦攲㌴㠶〷崰䤠䙎⁏汐硥䴠摥慩匠牥敶ㅶㄮ⸲⸲㤴㤲㈭昹搶㜱㘹ⴠ甠畢瑮⁵䍐砠㘸㙟‴畢汩㩤氠湩硵甭畢瑮㡸弶㐶猠湹汯杯⁹䵇⁔〺ਰ灁㌰〲㠱〠㨸㠱㌺⸱㤴‴せ㝸㑦攲㌴㠶〷崰䤠䙎⁏楌畮⁸敶獲潩㩮㌠ㄮ⸰〱′⌨㔱㘲‶䵓⁐潍慍㘲ㄠ㨵㈱㌺′千⁔〲㠱Ⱙ氠湡畧条㩥攠单䄊牰〠ⰳ㈠‸㠰ㄺ㨸ㄳ㐮㔹嬠砰昷㈴㑥㘳㜸〰⁝义但ⴠ倠潲散獳牯†††䤠瑮汥刨
敐瑮畩⡭⥒䌠啐†㍎ㄷ‰䀠ㄠ㘮䜰穈䄊牰〠ⰳ㈠‸㠰ㄺ㨸ㄳ㐮㔹嬠砰昷㈴㑥㘳㜸〰⁝义但ⴠ⼠慶⽲慰正条獥倯敬⁸敍楤敓癲牥琯牡敧⽴汐硥䴠摥慩匠牥敶ੲ灁㌰〲㠱〠㨸㠱㌺⸳㈰‴せ㝸㑦攲㜳晦〷崰圠剁⁎牃獡㩨䌠慲桳爠灥牯楴杮搠獩扡敬灁㌰〲㠱〠㨸〲㈺⸲ㄶ‴せ㝸㑦昲愰㐰〸崰䤠䙎⁏䍏偓›畓捣獥晳汵祬爠瑥楲癥摥爠獥潰獮牦浯挠捡敨ਮ灁㌰〲㠱〠㨸〲㌺⸲㈸‵せ㝸㑦攲㜳晦〷崰䔠剒剏ⴠ䔠牲牯椠獳極杮挠牵彬慥祳灟牥潦浲栨湡汤⥥›灁㌰〲㠱〠㨸〲㌺⸲㈸‵せ㝸㑦攲㜳晦〷崰圠剁⁎呈偔攠牲牯爠煥敵瑳湩佐呓栠瑴㩰⼯㈱⸷⸰⸰㨱㈳〶⼰敤楶散⽳楤捳癯牥⠠ⰰ丠牥潲⥲⠠慆汩摥挠湯敮瑣琠㈱⸷⸰⸰㨱㈳〶㬰䌠湯敮瑣潩敲畦敳⥤䄊牰〠ⰳ㈠‸㠰㈺㨰㈳㠮㘲嬠砰昷㈴㍥昷㝦〰⁝剅佒⁒牅潲慰獲湩潣瑮湥䄊牰〠ⰳ㈠‸㠰㈺㨰㈳㠮㘲嬠砰昷㈴㍥昷㝦〰⁝剅佒⁒牅潲慰獲湩䵘⹌䄊牰〠ⰳ㈠‸㠰㈺㨱㐱㜮ㄴ嬠砰昷㈴ㅥ晢㝦〰⁝义但ⴠ匠湹㩣搠睯汮慯敤‱祳据氠獩⡴⥳眠瑩″祳据椠整獭猨㨩〠渠睥‰灵慤整Ɽ〠搠汥瑥摥䄊牰〠ⰳ㈠‸㠰㈺㨱㔱ㄮ㜴嬠砰昷㈴摤昳㝦〰⁝䅗乒ⴠ吠潯潴潬杮⠠⸰ㄱ〰〰猠捥湯獤
潴猠慴瑲愠琠慲獮捡楴湯漠⸮䰯扩慲祲䐯瑡扡獡䙥硩灵灣㩰㔱ਹ灁㌰〲㠱〠㨸ㄲㄺ⸵㐱‷せ㝸㑦搲㍤晦〷崰圠剁⁎牔湡慳瑣潩桴瑡眠獡爠湵楮杮眠獡猠慴瑲摥漠桴敲摡〠㝸㑦攲㤰ㄱ〷‰瑡⸠⼮汐杵湩⽳汐杵湩慄慴慢敳瑓瑡灣㩰ㄱਲ灁㌰〲㠱〠㨸ㄲㄺ⸵㠱‱せ㝸㑦搲㙢㌲〷崰䤠䙎⁏畁潴灕慤整›潮甠摰瑡獥愠慶汩扡敬䄊牰〠ⰳ㈠‸㠰㈺㨱㘱〮㤱嬠砰昷㈴晤昳㝦〰⁝剅佒⁒敎睴牯敓癲捩㩥䔠牲牯椠牢睯敳慨摮敬爠慥㩤ㄠ㔲⠠灏牥瑡潩慣据汥摥
潳正瑥ⴽ灁㌰〲㠱〠㨸ㄲㄺ⸶ㄱ‹せ㝸㑦搲㍦晦〷崰䤠䙎⁏敎睴牯敓癲捩㩥䄠慢摮湯湩牢睯敳猠捯敫ⱴ椠⁴慷汣獯摥ਮ灁㌰〲㠱〠㨸ㄲ㈺⸲㔸‵せ㝸㑦搲㥥ㄱ〷崰䔠剒剏ⴠ堠䱍›湅楴祴›楬敮ㄠ:䄊牰〠ⰳ㈠‸㠰㈺㨱㈲㠮㔵嬠砰昷㈴敤ㄹ㜱〰⁝剅佒⁒䵘㩌瀠牡敳r䄊牰〠ⰳ㈠‸㠰㈺㨱㈲㠮㔵嬠砰昷㈴敤ㄹ㜱〰⁝剅佒⁒䵘㩌攠牲牯㨠灁㌰〲㠱〠㨸ㄲ㈺⸲㔸‵せ㝸㑦搲㥥ㄱ〷崰䔠剒剏ⴠ堠䱍›瑓牡⁴慴硥数瑣摥㰧‧潮⁴潦湵d䄊牰〠ⰳ㈠‸㠰㈺㨱㈲㠮㔵嬠砰昷㈴敤ㄹ㜱〰⁝剅佒⁒䵘㩌猠慴畴㵳歯灁㌰〲㠱〠㨸ㄲ㈺⸲㔸‵せ㝸㑦搲㥥ㄱ〷崰䔠剒剏ⴠ堠䱍›^䄊牰〠ⰳ㈠‸㠰㈺㨱㈲㠮㔵嬠砰昷㈴敤ㄹ㜱〰⁝剅佒⁒牅潲慰獲湩潣瑮湥䄊牰〠ⰳ㈠‸㠰㈺㨱㈲㠮㔵嬠砰昷㈴敤ㄹ㜱〰⁝剅佒⁒卓偄›牅潲慰獲湩敤楶散猠档浥潦瑨灴⼺ㄯ㈹ㄮ㠶ㄮ㈮ㄴ㤺㠰ਰ灁㌰〲㠱〠㨸㈲㈺⸵㌲′せ㝸㑦攲㤰ㄱ〷崰䤠䙎⁏敒牦獥楨杮瀠瑡獨眠瑡档摥戠⁹楌牢牡啹摰瑡䵥湡条牥䄊牰〠ⰳ㈠‸㠰㈺㨴㠰㠮㈷嬠砰昷㈴慤㠲㝣〰⁝义但ⴠ䰠扩慲祲猠捥楴湯㘠⠠噔
楷汬戠灵慤整敢慣獵景愠挠慨杮湩⼠潶畬敭⼱楶敤⽯噔启敨䄠敭楲慣獮⠠〲㌱⼩敓獡湯〠灁㌰〲㠱〠㨸㐲〺⸸㜸″せ㝸㑦搲㉡挸〷崰䤠䙎⁏楌牢牡⁹敳瑣潩‵吨⁖桓睯⥳眠汩敢甠摰瑡摥戠捥畡敳漠档湡敧椠瘯汯浵ㅥ瘯摩潥启⽖桔流牥捩湡㈨⤳匯慥潳䄊牰〠ⰳ㈠‸㠰㈺㨴㠰㤮嬠砰昷㈴慤㠲㝣〰⁝义但ⴠ䰠扩慲祲猠捥楴湯㘠⠠噔
楷汬戠灵慤整敢慣獵景愠挠慨杮湩⼠潶畬敭⼱楶敤⽯噔启敨䄠敭楲慣獮⠠〲㌱⼩敓獡湯〠ਲ਼灁㌰〲㠱〠㨸㐲〺⸸〹′せ㝸㑦搲㉡挸〷崰䤠䙎⁏楌牢牡⁹敳瑣潩‵吨⁖桓睯⥳眠汩敢甠摰瑡摥戠捥畡敳漠档湡敧椠瘯汯浵ㅥ瘯摩潥启⽖桔流牥捩湡㈨⤳匯慥潳㌰䄊牰〠ⰳ㈠‸㠰㈺㨴㠰㤮㜱嬠砰昷㈴慤㠲㝣〰⁝义但ⴠ䰠扩慲祲猠捥楴湯㘠⠠噔
Are you running some kind of encryption?
Other files are ok…
I think you’ve got a file system corruption problem.
I did find, in your older logs (which were English), The database is overloaded. (too busy) for everything to work correctly. We can take care of that but we will need work through this step at a time.
Thanks for your time, Chuck.
I am running a standard Synology OS, UK English, with the Plex server set to UK English. Which log file was that unicode in?
This evening I did a full, completely clean brand new installation of Plex. Completely deleted the entire Plex folder and installed from scratch. I set up the agent with LMA completely disabled for both films and TV, and then kicked off the media scan. The same thing has happened - I have 32 seasons listed under McMafia, made up of a random selection of shows from various folders within the directory tree. As far as I can see this is just bad matching by Plex, pure and simple. The folders and shows it’s misidentified are all correctly named, and it shouldn’t be scanning them for any metadata, unless I’m misunderstanding.
I can’t see how the DB can be overloaded as nothing other than the initial scan is running, and Plex should manage that itself. I also can’t see how this can be file corruption - I’m running raid 6, do regular SMART scans and all my shows play correctly and are identified properly by both Sonarr and Kodi.
I’ll upload a fresh set of logs in the morning when the scan has completed and we’ll see if we can figure out what Plex is doing wrong.
Mark,
That was file Plex Media Server.log. Plex Media Server.1.log was ok. Whatever happened it was in those final moments before you captured that posted log.
The DB gets ‘overloaded’ because it’s a SQLITE3 db. The ‘overload’ is because of how SQLITE3 stores tables. It is a single-file database engine therefore as tables grow they will naturally fragment. The problem occurs when normal operations occur, such as adding media, and those tables become fragmented. Sqlite3 must follow the internal links to find the data. This causes delay. When PMS doesn’t get the reply within the expected time limit a warning is raised. It results in SLOW QUERY messages in the log files. This is the start of a potential avalanche within PMS.
To avoid this scenario on a Synology, you must first remember the CPU isn’t all that fast.
What you’ve done in this is maintain locality-of-reference. The different table rows are nice and compact . You don’t have, for example, ‘music’ records physically interspersed with ‘movie’ or ‘episode’ rows.
Making sense?
Hi Chuck,
Your description makes sense, although it’s a bit worrying. If that’s the case, it means that despite being a mature, and reasonably expensive product, Plex has some serious architectural flaws which means it can’t hope to work properly.
My Synology is a 916+ with 8GB of RAM, so it’s an x64 process running on a quad core Linux box, and really should not be having any issues with performance.
Regardless of that, a decently written app would see DB read/write operations fail under load and would immediately retry/back off to ensure transactional integrity is maintained. If Plex just ploughs on, spiralling requests until a transaction storm occurs (an ‘Avalanche’, in your words) then the product is clearly flawed.
Some things don’t add up here though.
All of this has me wondering whether I’ve made a mistake moving back to Plex - which I did after 2 years of flawless Kodi operation using the same NAS and set of media files. I switched back due to Plex finally supporting HW transcoding on Synology, but if what you’ve described is in fact the state of the Plex architecture, it feels like I should request a refund for my Lifetime pass, and move back to using Kodi. Plex is supposed to be a mature, stable product that makes everything Just Work.
PS: where are the Plex staff here to help diagnose the problem? It’s great being helped by forum members, but with a paid product I’d have expected somebody from the Plex support or development team to have looked into this by now.
Mark,
I am Plex staff and work primarily in Customer Support. With respect to Synology NAS systems and Linux at large, I am also on the development team.
My purpose / responsibility here is to support you and get PMS working again. Should we find an actual software problem, I am your advocate in getting it resolved as well as, if possible, providing you a workaround until it is properly corrected by the Engineering team.
I ask we not enter into the “paid support” discussion here. The software itself is free. The additional services available with a Plex Pass are what you have subscribed to. That’s how they structured it. It’s been that way for as long as I’ve been around.
Regarding your Syno,
With this apparent level of corruption, we need to make certain the machine and drives are solid. I would like you to do the following:
Thanks Chuck. I wasn’t aware you are Plex staff and a developer. Might be useful if your profile/tags were more clear about this.
I’ll update you with the Smart health results later. I run scheduled monthly data scrubbing and the disk was clear the last time it ran.
Happy not to enter into the paid support discussion; I understand exactly what Plex pass is for and the value behind it (although that might lead me onto the numerous bugs and issues with Android sync, but maybe let’s not go there here ;)).
