Better, I guess, but still weird. Now two flows get started and a few more get started sooner in a few more tests than those attached. Still, during the test I am basically locked out of getting in via a new ssh until it completes. I guess the problem here is deeper than this, perhaps higher up the stack? Also, that it's "good" that it exists on the omnia to get more eyes on it. Pls file bug there too? Not so good performance for a NAS though. On the plus side... cake CAN shape to 900mbit on this platform, only eating 80% of cpu or so. On Sun, Nov 20, 2016 at 2:50 PM, Toke Høiland-Jørgensen wrote: > Toke Høiland-Jørgensen writes: > >> Marcin Wojtas writes: >> >>> Hi Toke, >>> >>> Yes, I've been overloaded heavily with another projects and this one >>> is my hobby activity I haven't had time for. I should have some spare >>> time though in the beginning of December and will send v2. >> >> Awesome! >> >> I'm building a LEDE image with the BQL patch; will post it when it's >> finished building... And then try to get the patch into LEDE and the >> Turris Omnia repo once someone has tested it. Figure the xmit_more patch >> can come later, once you get around to fixing that up :) > > WRT1200 image with BQL patch here: https://kau.toke.dk/lede/mvebu/generic/ > > -Toke -- Dave Täht Let's go make home routers and wifi faster! With better software! http://blog.cerowrt.org