|
Hello,
My setup is:
1. Two CentOS 5 servers running on a Virtual Rack at ThePlanet.
2. Each server has 10 addresses assigned to it (for sake of privacy
I'll use rfc1918 prefixes):
2.1 one server (boffo) is on 192.168.0.80/28, has address 192.168.0.84
2.2. second secer (crivens) is on 192.168.0.64/28 has address 192.168.0.68
3. The virtual rack router allows both servers to communicate on both
subnets - i.e. I assigned
crivens with 192.168.0.85 on eth0:0 and it can be contacted through
that address.
4. I picked 192.168.0.82 as a virtual IP that I'd like to serve from
both servers.
5. The service is a simple HTTP (port 80) for now.
6. Eventually I'll run Xen guests on these servers which will use more
IP addresses from these ranges, but for now I just want these servers
to run Apache on Dom0 to do some http redirection. I also suspect it
might be a good idea to run the ldirectord on Dom0.
I tried to follow the instructions in a few pages, the most relevant
one seems to be:
http://www.ultramonkey.org/2.0.1/topologies/sl-ha-lb-eg.html
The packages I use are:
heartbeat, heartbeat-ldirectord from centos (i.e. not the Piranha
thing mentioned in some places).
I'm still working through the "Waht if there is a problem" section in
the mini-HOWTO but just wanted to ask a quick question -
Since one of the servers is on 192.168.0.64/28 I though that maybe I
need to assign it an address on the same network as the VIP, so I
assigned it 192.168.0.85 (mentioned in point 3 above). I did this
using eth0:0:
# ifconfig eth0:0
eth0:0 Link encap:Ethernet HWaddr 00:1E:4F:0F:B1:37
inet addr:74.54.241.85 Bcast:74.54.241.95 Mask:255.255.255.240
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
# ip a s eth0
7: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue
link/ether 00:1e:4f:0f:b1:37 brd ff:ff:ff:ff:ff:ff
inet 192.168.0.68/28 brd 192.168.0.79 scope global eth0
inet 192.168.0.85/28 brd 192.168.0.95 scope global eth0:0
inet6 fe80::21e:4fff:fe0f:b137/64 scope link
valid_lft forever preferred_lft forever
Could this be an issue with the setup?
Also, it looks like the heartbeat processes on both servers manage to
see each other in terms of the heartbeat packets and can connect to
each other (verified with tcpdump and telnet to port 80 on both
servers) but still think that the other one is dead and I suspect both
try to serve the same address.
The VIP address 192.168.0.82 dies when I stop httpd on the
192.168.0.68 (/.85) server "crivens".
I'm still trying to go through the docs but was wondering if the
aliasing of eth0:0 could be the problem and if so then what's the
proper way to solve this?
Thanks,
--Amos
|
