Uhg, I think I found what I was doing wrong.
The route statement on the Director should have been:
route add -net xxx.xxx.134.144/29 eth0
route add -host xxx.xxx.134.150 dev eth0:1
This makes sense in hindsight since my Director IP is on a different IP
block than the Realservers and the Virtual IP are.
I noticed because I was not able to see ARP of the Realserver's Real IP,
when the first (host) route was in place. Packets from the Director for
the Realserver must have been getting sent to the Director Gateway which
is a bad thing for LVS-DR.
Also strangely enough I did setup my new Realservers on the same IP
block as the Director at one point in troubleshooting, but it didn't
work then. Oh well..
This was indeed a DUH! problem.
Thanks all for your help.
On Wed, 2006-09-20 at 20:18 +0100, Graeme Fowler wrote:
> On Wed, 2006-09-20 at 20:57 +0200, Vittorio R. Tracy wrote:
> > When I run the same tcpdump command on the Realserver I don't get
> > anything.
> OK, so now we know that the traffic isn't getting out of the director.
> Or, at least, across the wire to the realserver...
> > IP Virtual Server version 1.2.0 (size=4096)
> > Prot LocalAddress:Port Scheduler Flags
> > -> RemoteAddress:Port Forward Weight ActiveConn InActConn
> > TCP xxx.xxx.134.150:25 rr
> > -> xxx.xxx.134.147:25 Route 1 0 1
> > IPVS connection entries
> > pro expire state source virtual destination
> > TCP 00:57 SYN_RECV xxx.xxx.134.237:1424 xxx.xxx.134.150:25
> > xxx.xxx.134.147:25
> > IP 00:31 ERR! xxx.xxx.134.237:0 0.0.0.5:0
> > xxx.xxx.134.147:0
> Umm... I haven't ever seen a connection in state ERR! before. Anyone
> else know why that should/could be the case? Evidently this is the
> source of your problem, Vittorio. Possibly :)
> LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
> Send requests to lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
> or go to http://www.in-addr.de/mailman/listinfo/lvs-users
Vittorio R. Tracy <vrt@xxxxxxxxxxxxxxx>