Unraid 7.2.0-beta.1 (though issue occurred on non beta as well)

Hey all,

I'm struggling to solve my download performance issues, so I'm coming to you all in hopes of a fix.

When using p2p clients (transmission, qbittorrent, deluge) I can't seem to maintain stable speeds. My configurations worked fine on windows before I made the switch to Unraid, but now I'm struggling. The speeds will start around 125-150mb/s but within a minute drop down to <10mb/s where they will remain.

Downloads are going to the cache first (2x 4tb 990 pros). I've troubleshooted my network performance and the hardware is performing fine. I've done iperf tests,tried different raid configurations, downloaded to a separate ssd not in the cache or array, and the problem persists. I've installed a 10g fiber nic and it didn't help either. I've stopped all docker containers besides the p2p client and the speed drops are still persistent. I'm mostly out of ideas now.

I appreciate any help

phobolog-diagnostics-20250817-1453.zip

Just to confirm, you can do large transfers over SMB without issue, i.e., the issue is just with app downloads?

Setup smb to the same share and transferred a 14gb file. It sustained speeds of 285mb/s for the whole transfer.

Assuming 14GB is enough to see the issue with a download, it rules out the devices as the problem, possibly some config issue with the containers, or a problem with the WAN. Does an internet speed test, for example, work normally when you see the slow speed?

Yeah the p2p speeds usually drop significantly within 10 seconds. My testing would point to the container/docker configurations, I think.

I've done extended downloads (just yesterday) outside of unraid and have pretty easily hit the sustained 2gbps cap on my WAN.

I did download the Ookla speed test script within Unraid and the DL speeds maintain at about 1220 mbps which is much better than what I am seeing in p2p performance. I also downloaded the same script inside a few of my docker containers and saw similar performance via WAN. p2p containers are on a vlan that go to a round robin balanced VPN connections and its a bit slower 500mbps (since single threaded test). However I've tested these containers in host mode, and bridged mode on the main (non VPN) subnet and the issue persists in both scenarios. VPN setup has also performed stellar before I transferred off my windows server so I have no reason to suspect that.

I suggest asking in the support thread for one of the affected containers; there may be some known issues or config suggestions.

Coming back just to say the issue was related to docker networks I manually created using the console. Deleting them and recreating them via Unraid's UI fixed the issue.