> > Unfortunately I have to report that the crashing has resurfaced. I'm > > currently using kernel 5.0 with Heiner's fix applied. In the last few > > days I've had the crash occur 4 times now. I'm not sure how to further > > investigate this but I'm guessing a patch that adds more debugging > > output will be needed to track down where exactly things go bad and > > hopefully what exactly is triggering it? In 2 cases there was > > bi-directional traffic, in the other 2 cases there was only incoming > > traffic. In all cases the nic became unresponsive, all connections to > > the box died/timed out. After roughly 10 minutes the nic came back to > > life on its own without my intervention. The odd thing I noticed was > > the dmesg log showing the problem only shows up once in dmesg instead > > of 4 times to coincide with the 4 crashes/freezes. I don't think the > > box reboot itself, and if it did then dmesg should be clean. > > > Instead of having to wait for the issue: Can you trigger it with iperf3 > in either direction?
I'm not familiar with iperf3 but installed it and looked up a quick example. I started the iperf3 server on the server and then ran the test on the client a bunch of times. And then also tried the test with the `-d` bi-directional flag. I couldn't get the crash/freeze to trigger though. Is there a certain command line I should run the test with, or anything specific I should do? -Derek
