Date: Thu, 11 Sep 2014 10:05:14 -0700 From: Freddie Cash <fjwcash@gmail.com> To: Steven Hartland <killing@multiplay.co.uk> Cc: FreeBSD Stable <freebsd-stable@freebsd.org> Subject: Re: Using CARP with multiple IP aliases (FBSD 10.0) Message-ID: <CAOjFWZ5qVePjG8Wuor7NS1__QcOv29EH7fVtxLgA7QZa6SSqZg@mail.gmail.com> In-Reply-To: <7925563B043E419996CD7FEE8C7DFDB6@multiplay.co.uk> References: <CAOjFWZ5_M_d3AStmkUJKk0TgesowJH-jWsKBbvZshwQ278o8ow@mail.gmail.com> <FE2CFC78656D40D996B1970649E20CCF@multiplay.co.uk> <CAOjFWZ7L619Ms=r=G7vPnA9E-5EGc-tQcgf19NbV-bxCU5TJvw@mail.gmail.com> <7925563B043E419996CD7FEE8C7DFDB6@multiplay.co.uk>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Sep 11, 2014 at 9:34 AM, Steven Hartland <killing@multiplay.co.uk> wrote: > I can't say I've used it in that way and I'm not sure how carp decides ho= w > to fail over when it has multiple IP's available. > =E2=80=8BI'm hoping, and my testing appears to corroborate, that it fails b= ased on the interface state, and all IPs transfer over at once (CARP systctl set to fail everything at once if any one interface state changes).=E2=80=8B > I can confirm you don't need all the params when adding an IP to vhid. > so you can for example configure the vhid and then add the IP's or do > as you have done and configure it on the first IP. > =E2=80=8BThat's good to hear. Will simplify things a bit.=E2=80=8B > Best thing to do is try it and see. > =E2=80=8BThat's scheduled for tomorrow morning. :) I'll try it first with= only setting pass/advskew on the vhid once, and just adding the alias IPs to the vhid. If that doesn't fix things, then I'll try with a separate vhid per IP. The reason I was asking about this is that I have a pair of systems in place now (sys1 and sys2, with sys1 configured with advskew 1 to make it always master) where everything works wonderfully for between 5 and 15 minutes. If I down an interface on sys1, or physically remove a cable from sys1, everything fails over to sys2 and traffic continues normally.=E2=80=8B =E2=80=8B If I bring the interface back up on sys1, then everything fails = back over to sys1 and traffic continues. After 5-15 minutes, though, igb0 on both boxes switches to master state. :( igb1, igb2, and igb3 on sys2 all stay in backup state. And then traffic slows to a crawl as the upstream switch gets confused and sends packets=E2=80=8B randomly between the two hosts. Manually changing state to backup on igb0 on sys2 fixes things for about 3 seconds, and then it switches back to master. Once this happens, tcpdump on both systems only shows VRRPv2 packets from sys1, nothing from sys2.=E2=80=8B I have to reboot sys2 in order to get th= ings working again. As I said, this is the first time I've used CARP with multiple shared IPs on an interface (NAT firewall), so I may be doing things "wrong" or non-optimally. :) --=20 Freddie Cash fjwcash@gmail.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAOjFWZ5qVePjG8Wuor7NS1__QcOv29EH7fVtxLgA7QZa6SSqZg>