------6F542WX18VLPXEJTSAO8OF9DUPZW5N Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=UTF-8 Hi, I'm not near the server for a while but network is set up 2x broadcom nic with whatever driver works out of the box. Both set as slaves to bond0 which is in mode=4 with no explicit options. Switch is a fairly basic tp-link but works almost identically to a Cisco. Has the 2 ports set up in a portchannel. Short and long, this config works in earlier kernel, but not with kernel shipped in 7.2 Release notes for RH7.2 suggest some work on bonding has been done, wonder if default options (LACP speed?) have changed? Jon On 30 December 2015 09:44:02 GMT+00:00, Dan Kenigsberg <danken@redhat.com> wrote:
Hi Stefano,
It's definitely not the switch, it seems to be the latest kernel
On Tue, Dec 29, 2015 at 09:57:07PM +0000, Jon Archer wrote: package
(kernel-3.10.0-327.3.1.el7.x86_64) which stops bonding working correctly, reverting back to the previous kernel brings the network up in 802.3ad mode (4).
I know, from reading the release notes of 7.2, that there were some changes to the bonding bits in the kernel so i'm guessing maybe some defaults have changed.
I'll keep digging and post back as soon as i have something.
Jon
On 29/12/15 19:55, Stefano Danzi wrote:
Hi! I didn't solve yet. I'm still using mode 2 on bond interface. What's your switch model and firmware version?
Hi Jon and Stefano,
We've been testing bond mode 4 with (an earlier) kernel-3.10.0-327.el7.x86_64 and experienced no such behaviour.
However, to better identify the suspected kernel bug, could you provide more information regarding your network connectivity?
What is the make of your NICs? Which driver do you use?
Do you set special ethtool opts (LRO with bridge was broken in 7.2.0 kernel if I am not mistaken)?
You have the ovirtmgmt bridge on top of your bond, right?
Can you share your ifcfg*?
-- Sent from my Android device with K-9 Mail. Please excuse my brevity. ------6F542WX18VLPXEJTSAO8OF9DUPZW5N Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: 8bit <html><head></head><body>Hi,<br> <br> I'm not near the server for a while but network is set up<br> <br> 2x broadcom nic with whatever driver works out of the box.<br> <br> Both set as slaves to bond0 which is in mode=4 with no explicit options.<br> <br> Switch is a fairly basic tp-link but works almost identically to a Cisco. Has the 2 ports set up in a portchannel.<br> <br> Short and long, this config works in earlier kernel, but not with kernel shipped in 7.2<br> <br> Release notes for RH7.2 suggest some work on bonding has been done, wonder if default options (LACP speed?) have changed?<br> <br> Jon<br><br><div class="gmail_quote">On 30 December 2015 09:44:02 GMT+00:00, Dan Kenigsberg <danken@redhat.com> wrote:<blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;"> <pre class="k9mail">On Tue, Dec 29, 2015 at 09:57:07PM +0000, Jon Archer wrote:<br /><blockquote class="gmail_quote" style="margin: 0pt 0pt 1ex 0.8ex; border-left: 1px solid #729fcf; padding-left: 1ex;"> Hi Stefano,<br /> <br /> It's definitely not the switch, it seems to be the latest kernel package<br /> (kernel-3.10.0-327.3.1.el7.x86_64) which stops bonding working correctly,<br /> reverting back to the previous kernel brings the network up in 802.3ad mode<br /> (4).<br /> <br /> I know, from reading the release notes of 7.2, that there were some changes<br /> to the bonding bits in the kernel so i'm guessing maybe some defaults have<br /> changed.<br /> <br /> I'll keep digging and post back as soon as i have something.<br /> <br /> Jon<br /> <br /> On 29/12/15 19:55, Stefano Danzi wrote:<br /><blockquote class="gmail_quote" style="margin: 0pt 0pt 1ex 0.8ex; border-left: 1px solid #ad7fa8; padding-left: 1ex;">Hi! I didn't solve yet. I'm still using mode 2 on bond interfac e. What's<br />your switch model and firmware version?<br /></blockquote></blockquote><br />Hi Jon and Stefano,<br /><br />We've been testing bond mode 4 with (an earlier)<br />kernel-3.10.0-327.el7.x86_64 and experienced no such behaviour.<br /><br />However, to better identify the suspected kernel bug, could you provide<br />more information regarding your network connectivity?<br /><br />What is the make of your NICs? Which driver do you use?<br /><br />Do you set special ethtool opts (LRO with bridge was broken in 7.2.0<br />kernel if I am not mistaken)?<br /><br />You have the ovirtmgmt bridge on top of your bond, right?<br /><br />Can you share your ifcfg*?<br /><br /></pre></blockquote></div><br> -- <br> Sent from my Android device with K-9 Mail. Please excuse my brevity.</body></html> ------6F542WX18VLPXEJTSAO8OF9DUPZW5N--