Thanks for the investigation, I guess what I'm wondering the most right now is if the patch from Oleg is a good change in general, so any others that can test and give me results would be appreciated.
Thanks, Jack On Wed, Sep 25, 2013 at 11:14 AM, Rumen Telbizov <telbi...@gmail.com> wrote: > Hi Jack, > > Thanks for the suggestions and looking into this. > Here are a few additional bits of information that you requested: > > 1. We did disable spanning tree on the switch port and the result of that > is that basically now creating/destroying a vlan on the ix interface makes > it freeze for about 3 seconds from previously 6. > 2. I brought manually the physical interface down (ifconfig ix1 down) and > then up and measured the time it took for the interface on the switch to > show as up - and that was again about 3 seconds. So basically it seems to > me like that's how long it normally takes for a link to be negotiated. > 3. We tweaked a setting on the switch which instructs the port to be down > for 5 seconds before it's considered down. Then repeated the test again and > the result was that now the "freeze" period got reduced to 1.5 seconds and > no ping packets were lost but one of the packets returned with 1500 ms > delay. My guess on this one is that since the switch ignored the flapping > of the interface, bringing it back up was much faster. So it does mask the > problem somewhat but it is still there - the interface seems to go down. > > Next steps: > 1. I will reinstall this machine with the latest 9.2 and repeat the tests > see what happens. > 2. If the previous one fails - I'll try the patch that Oleg sent. Thanks a > lot for providing that patch. > > I'll update this list with more findings. > > Thank you, > Rumen Telbizov > > > > On Wed, Sep 25, 2013 at 10:01 AM, Jack Vogel <jfvo...@gmail.com> wrote: > >> Rumen, >> >> I'd like to know if you can check the spanning tree setting as Daniel >> mentioned and if that solves it, >> I do know that that can cause considerable delay in link transitions. >> Also, can you see if you see >> different behavior by going to the latest 9.2 bits? >> >> Oleg thanks for the patch, I will check it out and have my validation >> engineer do some tests and >> we'll get to the bottom of this. >> >> Jack >> >> >> >> On Wed, Sep 25, 2013 at 9:57 AM, Rumen Telbizov <telbi...@gmail.com>wrote: >> >>> Thanks for the patch Oleg. >>> I'll give it a try but I would also like to hear back from Jack. >>> >>> Oleg, can you confirm that you're experiencing the exact same >>> problem/behavior? What kind of switch is connected on the other end? >>> >>> Thanks, >>> Rumen Telbizov >>> >>> >>> On Wed, Sep 25, 2013 at 9:40 AM, Oleg Bulyzhin <o...@freebsd.org> wrote: >>> >>> > >>> > I'm running my ixgbe servers with attached patch. >>> > It does solve this problem for me. >>> > >>> > -- >>> > Oleg. >>> > >>> > ================================================================ >>> > === Oleg Bulyzhin -- OBUL-RIPN -- OBUL-RIPE -- o...@rinet.ru === >>> > ================================================================ >>> > >>> > >>> >>> >>> -- >>> Rumen Telbizov >>> >>> Unix Systems Administrator <http://telbizov.com> >>> _______________________________________________ >>> freebsd-stable@freebsd.org mailing list >>> http://lists.freebsd.org/mailman/listinfo/freebsd-stable >>> To unsubscribe, send any mail to "freebsd-stable-unsubscr...@freebsd.org >>> " >>> >> >> > > > -- > Rumen Telbizov > Unix Systems Administrator <http://telbizov.com> > _______________________________________________ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to "freebsd-stable-unsubscr...@freebsd.org"
