[Xen-users] Using Xen on load balancer keeps loosing connection

[Tom Potwin <wxman2@xxxxxxxxx>]

Hi

I just joined this list after I tried looking for help with the HAProxy, and
HA heartbeat people. They're both at a loss as to what's causing my problem.
The only place I haven't tried is here with the Xen experts. I checked xm
info for the version and got this:
Release : 2.6.24-23-xen
version : #1 SMP Wed Apr 1 23:47:10 UTC 2009

I'm using HAProxy and Heartbeat on two Ubuntu 8.04 servers. I have two Xen
nodes on each of my physical machines. One is the load balance and Heartbeat
(LB1), the other is the actual LAMP web server (WEB1). Testing
HAProxy/Heartbeat setup seems that it's working fine, by that I mean that
shutting off one of the web servers, it switches to the other one. My
problem is I keep loosing access to the HAProxy stats page, which only
should show up on the LB nodes where HAProxy is. I know that isn't a huge
problem, but I'm worried it might be a sign of a bigger problem somewhere. 

The stats show up fine for about 15-20 minutes, then I get a apache generic
404 error page. I also see: "File does not exist:
/var/www/apache2-default/haproxy" show up in the apache error log on WEB1
node as soon as I loose it. If I go back to my LB1 node and restart
Heartbeat, it all comes back for another 15-20 minutes. There's nothing in
any of the logs that I can see, other than it stops logging when it happens.
I use http://192.168.31.100/haproxy?stats to get to that stats page. The
.100 is the shared address between the the load balancers. If I use
192.168.31.201, which is LB1, I get the browser's 404 notice. If I use .100,
it shows my apache generic 404 page. So somehow it stops seeing LB1, and
goes to port 80 on my web server on the WEB1 node. That's where I see the
apache error saying it can't find the HAProxy stats page.

When I used the "tcpdump -q -i eth0 tcp port 80 and src host 192.168.31.100"
command, it showed me looking at the stats, and the test web page:
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on eth0, link-type EN10MB (Ethernet), capture size 96 bytes
11:23:16.106664 IP 192.168.31.100.www > 192.168.30.64.2289: tcp 0
11:23:16.254209 IP 192.168.31.100.www > 192.168.30.64.2289: tcp 0
11:23:16.254409 IP 192.168.31.100.www > 192.168.30.64.2289: tcp 262
11:23:16.254501 IP 192.168.31.100.www > 192.168.30.64.2289: tcp 0
11:23:17.460534 IP 192.168.31.100.www > 192.168.30.64.2290: tcp 0
11:23:17.628385 IP 192.168.31.100.www > 192.168.30.64.2290: tcp 0
11:23:17.628590 IP 192.168.31.100.www > 192.168.30.64.2290: tcp 2712
11:23:17.839448 IP 192.168.31.100.www > 192.168.30.64.2290: tcp 2712
11:23:17.839460 IP 192.168.31.100.www > 192.168.30.64.2290: tcp 524

Once I couldn't see the stats page again, the output stopped completely. I
watched it on LB2 as well. It seems like it stops listing to the .100 IP
address. If I use "tcpdump -q -i eth0 tcp port 80" I see LB1 checking web1
and web2, but nothing on the .100 address. The HAProxy people said they
thought it might be a Heartbeat problem, because after they checked my
HAProxy setup, they couldn't find any problems there. The heartbeat people
say the same thing. Sorry for the long post, I'm just getting desperate for
some help. 

Thanks, Tom



_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users