cross-posted from: https://lemmy.today/post/39908979
Hi, I’m having a problem with my qBittorrent setup that I can’t quite debug or find information on.
I’m seeding from a hard drive and I get something like 5-15 MBps of seeding capacity depending on the day. However I noticed upon inspecting a system monitor (I am running Debian 13) the total disk IO read was about 4x the seeding speed.
I’ve tweaked the advanced settings all I can to no avail. I am using the version of QBit from Devian 13 main. It’s 5.Something.
Would anyone have insight on this problem? I’ll take anything at this point. The application is using about 250MB of RAM.
I think it’s just normal, I seed from a 10gbps SSD and barely get above 7MB/s
Don’t know if this helps you but did you already try changing Disk IO Type to Simple pread/pwrite (in advanced settings)? It was meant to address some issues with disk read/write + RAM usage. Not something that affected me all that much but could be worth a look.
Was mentioned in the 5.0.1 release notes and in the github pages
https://www.qbittorrent.org/news
https://github.com/qbittorrent/qBittorrent/pull/21300
Beyond that I’d echo the other comment, try to just change settings you actually need to change. Oftentimes people make a whole ton of settings changes in their torrent client and then can’t figure out how to get it back to normal.
This helped! Disk IO read is now a bit less than double the speed bandwidth! Much more reasonable. Thank you/:))))))
I just changed it. I’ll get back to you in a bit.
Total speed is going to be bottlenecked somewhere, which could be drive read speed, upload bandwidth, or the peer’s download speed, their write speed, or any number of other factors.
If your side is capable, don’t worry too much about it. You could be seeding to someone on their phone in a shack in remote Mongolia. That’s never going to be fast.
Also, reset all the advanced options back to default. You don’t need to touch any of them, and changing things might make it worse.
Yeah but youd think I would fill some sort kf buffer up eventually, if I’m limited by bandwidth, right?
Am I actually just limited by DiskIO so I’m always behind? My memory usage is around 250MB.
No. There’s no point in filling more than the protocol buffers if you can’t empty them. Loading a huge file into memory that won’t be used right away is a waste.
You’ve already said your read speeds are better. Memory usage doesn’t seem relevant here.
Isn’t port binding in the advanced settings? If you reset advanced settings to default, don’t forget to re-bind your client to your VPN to avoid leaks!
If you’re seeding more than one file to more than one concurrent user, chances are, your drive needs to switch around and buffer quite a bit left and right to get the material you’re seeding cached. That sounds like fairly formal behavior.
