Plex doesn't always read media with multiple channels as anything other than stereo

You are correct on the Heisenbug principle.

As for databases, com.plexapp.plugins.library.db is 700MB, com.plexapp.plugins.library.blobs is 2.2GB

My system has been a mess since the plex update. I have more than 1,000 unmatched or mismatched movies. Everything was matched before the update, but the update completely trashed my database.
I’ve been working on re-matching movies for about 2 or 3 weeks now, but I figure it will take a few more weeks before I am done.
The new matching logic is beyond garbage.
Any movie with a short title (maybe 5 characters or less) just gets tossed to some random movie.
Finding all these broken an mis-matches movies is very difficult and time consuming.
When it’s done, maybe I’ll post something about long scans.
Right now, I know for a fact that my library is trashed due to Plex updates, and until that is fixed, I don’t want someone to spend time digging into it.

1 Like

That sounds really frustrating. You’re not the only person who was affected by recent changes in the scanner/matcher.

I noticed that you’ve got “letter” folders. What’s configured in the Library? Is it I:\Movies2\ or each of the I:\Movies2\M folders?

Do you know about the ability to add ID hints to movie folder (or file) names?

.../path/The Matrix Revolutions (2003) {imdb-tt0242653}/The Matrix Revolutions (2003).mp4

If you’re already spending time fixing naming and matching, adding IDs is powerful.

Library is configured with I:\Movies2\M not I:\Movies2
I use the .nfo files, not hints.

I didn’t always use .nfo files, they didn’t used to be required, and it seemed a lot of unnecessary work. With the new scanner, they are absolutely mandatory. With the new scanner, if you don’t use them, your media is hosed.

I do have a bit of heartburn that Plex didn’t make this a bit more ‘noisy’. They should have been screaming this out at the top of the mountain, that if you didn’t use hints, your library would get trashed.
But no use crying over spilled milk. It is what it is. Plex bought a ton of ill will with their customers on this, but that’s on them, not on me.

I think the .nfo files behave very similarly - an ID is extracted initially and used as a (very strong) matching hint.

90% of the time “my stuff got unmatched” happens to people who say “My files are named perfectly!” but it’s actually nonsense.

Your seem like they’re named really well. Might be worth sharing logs - could get some attention.

We are deviating from the thread, and I don’t really want to get into this, because this topic attracts trolls.

To answer your question.
There are 3 primary issues with the new matching logic. None of these issues existed prior to the update in January. Prior to January, I would have given Plex a 95%+ accuracy at matching on title and year only.
After January, I’d give it maybe 75% to 80%.
After January, I had to watch it like a hawk.
Hints fix this of course. But Plex documentation still says they are not required, so they are not required. So any answer of ‘you must use hints’, needs to start by updating documentation to say they are required.
I am now going through and adding hints to all of mine, but I have 16,000 movies. Adding hints to 16,000 movies is not something you do in a few weeks. Yea, I know there is a program that automates this, and I have that. That takes some of the heavy lifting, but not all.

#1) short names are a major crap show. The new matching logic is a total loss on short title names. If a title maybe 4 to 6 characters (or less), Plex no longer makes a serious effort to match it, even if you have the year included. I’ve seen matches that I sit back and say ???
I’m a programmer by trade. I try to find logic in this. My best educated guess, is that if the title is less than ‘x’ characters, Plex applies a randomization to the name, and randomly assigns some title out of a bucket. No other explanation makes sense.
The movie ‘3 (2017)’ will match to a different title every time you try it. 'Cars 3 ’ is the current match I have.
‘It (1990)’ plex will match to ‘Let it Ride (1990)’. To me, this is hugely lazy programming

There are some that make sense
Pet (2016) matches to Pete’s Dragon (2016) Yea, I get that they both have ‘Pet’ and ‘2016’. I am sure when Plex tries to search the database for ‘Pet (2016)’ that it gets back both ‘Pet (2016)’ an ‘Pete’s Dragon (2016)’. But honestly, can’t Plex just assume that ‘Pet (2016)’ is the correct match for search term Pet (2016). Is that really asking too much?
Pig (2021) Plex will match to Girl (2020) I can’t come up with a reason for that.
Polar (2019) matches to Polaroid (2019) (see my explanation on Pet)
Rent-A-Pal (2020) matches to The Rental (2020)
It’s really just a total S**t show. But worse, when I originally reported it, they blamed me for the problem. Said it was all my fault, and if I followed the ‘required hints’, that this wouldn’t have happened.

#2) common names are a major crap show. ‘Heist’ for example, or some other name that is in dozens of movies. Even if you use the year, you have no guarantees. But something like Heist is just as likely to match to a random name, as it is to Heist.
My best guess: When Plex tried to derive a match, if there is more than 1 possible result, it loses it’s mind and just goes random.

#3) There is a whole other grouping that I can’t find any pattern to. A long title name, with 2 or 3 significant words (plus the year) sometimes matches to some other movie. Some real examples: Paranormal Activity 4 (2012) plex will match to Paranormal Activity 3 (2011)
Amanda & Jack Go Glamping (2017) plex matches to Amanda and the Fox (2018). Oh, well, they
both have the word ‘Amanda’, so I guess that makes a perfect match, right?
American Poltergeist 3 (2014) plex matches to The Haunting of Alice D (2014)
In my opinion, the matching logic went from very sophisticated, to stupid and lazy.

1 Like

I agree that hints (filename or .nfo) shouldn’t be required, and when there’s an exact match I would sure expect it to be preferred.

A few of your examples do seem Real Dumb and would be very frustrating.

The extreme slowness, especially because it was a huge change, makes me wonder if something else is just wrong.

:person_shrugging:

Wouldn’t surprise me a bit.

One thing I would really like to see, is the ability to spilt meta data out to a different hard drive from Plex install.

After the January update, the amount of Meta data increased by 4 to 5 times.

I used to be on a 1 TB SSD, and was only using about 60% of that. So prior to the January update, my foot print was roughly 600 MB
After the January update, after the first scan, I ran out of space on that hard drive due to the increase of Meta data. I immediately had to go to a larger drive, I went to a 2 TB drive, and I blew for space. I finally had to go to a 5 TB drive.
My current footprint is 2.72 TB. Very little change in my actual media in the last 30 days, but my plex footprint is almost 5 times what it was a month ago.
I attribute all of that to the new update.
This is likely where the slow down is.

And yea, I was more than a bit pissed that Plex didn’t warn me that the update would require a 500% increase in disk space. That’s a rather significant change.

You can move the Plex data folder to another location on your system, separate from where the application is installed.

[HowTo] An extended guide on how to move the Plex data folder on Windows

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