Hi,
I really like the program and its possibilities, but I have some trouble with the performance.
I’m running the docker on a qnap NAS with 16GB of RAM. Around 32.000 titles.
After around the half of all songs, all available 8gigs are eaten up.
All other Containers on the NAS are starting to fail…
Is there a possibility to reduce the RAM allocation or even better to make it free some RAM during an active run?
Usually SongKong doesnt take up much memory, but are all your files in a single folder ?
I’m using the plex format, so using a basefolder /music with Artists/Album/*.mp3
Okay thats fine.
So the problem may be more Docker specific, Docker instances arent really aware of what else is running on the machine maybe you have too many things running on the server at the same time. When you launch the SongKong docker instance you can specify memory usage so I would suggest yo reduce the max memory allowed, SongKong only works on a few folders a time so it shouldnt need that much memory.
Actually, I used the memory limit for the docker, and it went even better.
With the set cpu and mem limit, the process went a lot faster and better leveled for the whole server.
A little off-topic but is there a possibility to prevent musicbrainz from tagging it with compilation albums?
Okay so I dont know that much about Docker but I think that probably it originally thought it had more memory than the machine physically had, and so it ended up draining the machine of memory because being a Java application although it can automatically garbage collect unused memory it doesn’t rush to do it if not running out of memory. Perhaps I need to look at this and whther i can configure a default lower memory limit when using within Docker.
If you have enforced lower memory consumption then it will garbage collect as neccessary to prevent running out of memory.
It should only tag with compilations albums if no match to the original album, but to fully understand your question it would be best for me if you run Create Support Files so I can check your options and give me an example of where it has matched a compilation album where you dont think it should have.
I think I found a part of the issue for that many compilations. I set the release country to mine Germany.
But as I just noticed on musicbrainz nearly all tracks are flagged with release country EX for Europe.
So as it searches for germany it will just end up with compilations…
That would not prevent it matching albums from countries other than Germany, it just means that if a potential match is country Germany that match will get a small boost compared to an equivalent release from a different country.
The best thing you can do is run in against one album that matches against compilation when you don’t think it should, then immediately run Create Support Files so I can look at how SongKong came to that decision.