Wardy

/mnt/user/Storage - cache is set to Yes.

How would I be sure? I can see that they are written to the cache if I go exploring the cache drive from Main. I've tried giving rTorrent a lot of ram in the settings, but it never seems to use it in the way I'd expect / at all. The drives are ~1 year old and parity checks / mover generally run at around 120-180MBs (IIRC). Diags attached. unraid-diagnostics-20210113-1532.zip

I posted a few weeks ago in the binhex/qbittorrentvpn thread about weird issues when downloading, and high IO wait essentially locking my server up. Since then, I've done a complete rebuild and have moved over to linuxserver/rutorrent. However, I'm seeing similar issues again. This is now the 3rd client I've tried, and I can't seem to figure this out. (FYI: My server is 8c/16t with 32gb ram, 1tb sata SSD cache and 5*4TB array (1 parity)) Basically, I have a gigabit ISP and everything is hardwired over a gigabit lan. I can download games through steam on my gaming laptop at full speed, regardless of the size of the download. (This laptop has a very fast nvme ssd, so I'm not sure if that makes any difference here). I can also download torrents on the laptop at full speed too - so no traffic management from the ISP. On the server however, rTorrent will download at full speed for about 5 minutes, then the CPU usage (on rTorrent) shoots to about 20-30%, and the speed will tank down to a steady 25-30MB/s. The CPU usage on unraid seems relatively low, with occasionally 1 thread at 100%, but overall it's sitting at 5-10% usage. Previously, using deluge and qbittorrent, a few minutes into the download, the CPU usage would be at around 70-80%, and the entire Unraid UI would become unresponsive. When I looked into it a bit more, the IO wait was 30-50% while this was happening. This time, it seems to be 20-30%. I've recently swapped the SSD out for a different one, that I know is good, and it hasn't solved the problem. I don't know enough about Unraid and how it deals with writing data to the cache / array to even begin to understand what might be the issue here. Having said that, I can transfer large files across the network to and from the server, and the transfer will happily saturate the line for however long it needs to, and the server doesn't even care. With that in mind, can anybody tell me if this is hardware or software? Or can anyone give me any advice as to how I can start to figure this out once and for all? Any help is greatly appreciated!

For anyone interested, I solved my IO wait issue by changing my cache fs to xfs. Everything's working as expected now!

I'm having issues with write cache overload / high IO wait. I used to own a seedbox, but then moved to a place with a gigabit line, so decided to bring everything in-house. I'm running on an array of 5x 4TB disks (~< 1 year old), with a 500GB SSD cache sitting on top. The server itself is an 8c/16t xeon with 64GB of ram. For some reason now, when I'm downloading large files, it'll hit max speed for a bit, then drop down to ~40MBs, and unraid UI will lock up. If I SSH in, top says my wait is 40-50, at the same time, qbt is reporting around 80% write cache overload. None of this seems particularly normal to me, and I would assume that any downloads go straight to the cache (considering that's where my docker image sits), so there should be no huge wait times there. I've tried different combinations of pre-allocating or not, playing with the IO threads and cache lifetime values, but I can't seem to figure out what the problem is here. If anyone has any idea where I can start looking, that would be amazing, thanks.