Date: Fri, 14 Sep 2018 10:48:24 +0800 From: in2 <in2@gamesofa.com> To: freebsd-net@freebsd.org Subject: aws ena flipping workaround on 11.2-R Message-ID: <CAB9hPKVr4Gpte0nTRnou-fzu0s%2BGoQvAvzsoJC7_sjuhpVGBOQ@mail.gmail.com>
next in thread | raw e-mail | index | archive | help
Hi,
I'm running 11.2-RELEASE on aws c5/m5/r5 and seeing ena resetting every 30
minutes
this causes around 0.5% packet loss overall.
Sep 14 09:19:57 ip-10-251-24-249 kernel: ena0: device is going DOWN
Sep 14 09:19:57 ip-10-251-24-249 kernel: ena0: device is going UP
Sep 14 09:19:57 ip-10-251-24-249 kernel: ena0: queue 0 - cpu 0
Sep 14 09:19:57 ip-10-251-24-249 kernel: ena0: queue 1 - cpu 1
Sep 14 09:49:57 ip-10-251-24-249 kernel: ena0: device is going DOWN
Sep 14 09:49:58 ip-10-251-24-249 kernel: ena0: device is going UP
Sep 14 09:49:58 ip-10-251-24-249 kernel: ena0: queue 0 - cpu 0
Sep 14 09:49:58 ip-10-251-24-249 kernel: ena0: queue 1 - cpu 1
Sep 14 10:19:57 ip-10-251-24-249 kernel: ena0: device is going DOWN
Sep 14 10:19:58 ip-10-251-24-249 kernel: ena0: device is going UP
Sep 14 10:19:58 ip-10-251-24-249 kernel: ena0: queue 0 - cpu 0
Sep 14 10:19:58 ip-10-251-24-249 kernel: ena0: queue 1 - cpu 1
this bug has been fixed on head r333454 but is not merged back to 11.2-R
yet.
since this issue happens when renewing dhcp but we do not need dhcp anymore
after instance booted, may I suggest a simple workaround: kill dhclient,
for example
echo '(/bin/sleep 60; /usr/bin/killall dhclient)&' >> /etc/rc.local
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAB9hPKVr4Gpte0nTRnou-fzu0s%2BGoQvAvzsoJC7_sjuhpVGBOQ>