|
One, physically separate load balancer. (ubuntu 6.06). Two physically
separate real servers. This is all brand spanking new, high end intel
hardware. Running DR ( with the loopback adapater on the 2k3 servers).
The real servers are using the same physical interface to respond to the
load balanced requests as to the direct requests.
The more hops out the more the issue presents itself. I am trying to by
pass the loadbalancer in the problem case to see if it is even the issue.
On 8/3/07, Graeme Fowler <graeme@xxxxxxxxxxx> wrote:
>
> On Fri, 2007-08-03 at 09:17 -0600, Dan Baughman wrote:
> > Just to follow up on this, the packet loss appeared to be an unrelated
> > issue. If I connect directly to a realserver, the page loads fairly
> fast.
> > If I connect through the loadbalancer ( and get the same real server) I
> see
> > a lot of tcp retranmissions and such (as sniffed from the real server).
> I
> > can't figure it out. The delays appear to be from the RTO (retrans
> > timeout). I saw a patch from miccy soft about the mtu discovery and have
> > installed, I also tried disabling mtu discovery.
> >
> > Is lvs-nat "more compatible"?
>
> With what? :)
>
> Anyway: you have *something* wrong with your topology or infrastructure
> somewhere, clearly. I'd put my money on it being at layer 2 - a duff
> cable, bad switch port, bad cable run (adjacent to power), wrong type of
> cable (Cat5e where Cat6 should be, for example) and that you remove the
> problem by going direct.
>
> Is (Are) your director(s) [a] physically separate machine(s) from the
> realservers?
>
> Remind me again - this is DR, isn't it?
>
> Graeme
>
>
> _______________________________________________
> LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
> Send requests to lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
> or go to http://lists.graemef.net/mailman/listinfo/lvs-users
>
|
