Hello
This is my current setup. I am keeping it simple while getting my feet wet.
I am following the LVS mini-HOWTO for VS-NAT.
I have a director, client, and two real servers.
Director and real server software is:
RH Linux 6.2 Kernel 2.2.18 from kernel.org compiled with options from the
HOWTO
ipvs-1.0.5-2.2.18 patch from virtualserver.org
The director is using the configure_lvs-0.7 package
ipvsadm-1.14-1.
I had to upgrade my piranha packages to > 0.4.14 to install ipvsadm-14.1 So
I installed the 0.4.17-7piranha from RH HA. I am not using piranha at this
time but am keeping the option open for now.
The setup is backwards from the example as my corporate LAN is at 10.0.1.
This is where clients for the test would be located.
I can isolate completely and follow example if needed but do not see reason
why (yet)
So I have this setup:
Machine IP
client CIP=10.0.1.120 (should this just be the
default route for the clients network?)
director VIP VIP=10.0.1.168 (eth0:1)
Director internal interface DIP=192.168.1.1 (eth0)
realserver1 RIP1=192.168.1.2 (eth0)
realserver2 RIP2=192.168.1.3 (eth0)
The lvs.nat.conf looks like this:
LVS_TYPE=VS_NAT
INITIAL_STATE=on
VIP=eth0:1 10.0.1.168 255.255.255.0 10.0.1.255
DIRECTOR_INSIDEIP=eth0 192.168.1.1 192.168.1.0 255.255.255.0 192.168.1.255
DIRECTOR_DEFAULT_GW=10.0.1.120 (this doesn't seem right?)
SERVICE=t telnet rr 192.168.1.2:telnet 192.168.1.3:telnet
SERVER_NET_DEVICE=eth0
SERVER_DEFAULT_GW=192.168.1.1
When I run the script on the director I seems to be OK when I run on the
real-servers it hangs and complains about my route.
I have changed IP's default routes etc I get the same message. ipvsadm -l
shows what is expected I was even able to see that I was making connections
to the VIP on the real servers although the were all inactive but I could
telnet to the vip from machines on the clinet network and even though they
would hang at the login ipvsadm -l would show new connections and it was
doing rr.
Tried it with http as well as the service again I could see connections
being made but I would get http timeout.
I am now isolating these machines to their own switch/hub and I am going to
set up exactly as in the mini-HOWTO except eth0:0 192.168.2.110 instead of
eth0:110 192.168.2.110 as I don't see how that makes a difference. Again I
have 2 real servers a director and a client. I will post the setup after I
make the changes and run through the configure process.
Off to start over again.
thanks
This is my current setup. I am keeping it simple while getting my feet wet.
I am following the LVS mini-HOWTO for VS-NAT.
I have a director, client, and two real servers.
Director and real server software is:
RH Linux 6.2 Kernel 2.2.18 from kernel.org compiled with options from the
HOWTO
ipvs-1.0.5-2.2.18 patch from virtualserver.org
The director is using the configure_lvs-0.7 package
ipvsadm-1.14-1.
I had to upgrade my piranha packages to > 0.4.14 to install ipvsadm-14.1 So
I installed the 0.4.17-7piranha from RH HA. I am not using piranha at this
time but am keeping the option open for now.
The setup is backwards from the example as my corporate LAN is at 10.0.1.
This is where clients for the test would be located.
I can isolate completely and follow example if needed but do not see reason
why (yet)
So I have this setup:
Machine IP
client CIP=10.0.1.120 (should this just be the
default route for the clients network?)
director VIP VIP=10.0.1.168 (eth0:1)
Director internal interface DIP=192.168.1.1 (eth0)
realserver1 RIP1=192.168.1.2 (eth0)
realserver2 RIP2=192.168.1.3 (eth0)
The lvs.nat.conf looks like this:
LVS_TYPE=VS_NAT
INITIAL_STATE=on
VIP=eth0:1 10.0.1.168 255.255.255.0 10.0.1.255
DIRECTOR_INSIDEIP=eth0 192.168.1.1 192.168.1.0 255.255.255.0 192.168.1.255
DIRECTOR_DEFAULT_GW=10.0.1.120 (this doesn't seem right?)
SERVICE=t telnet rr 192.168.1.2:telnet 192.168.1.3:telnet
SERVER_NET_DEVICE=eth0
SERVER_DEFAULT_GW=192.168.1.1
When I run the script on the director I seems to be OK when I run on the
real-servers it hangs and complains about my route.
I have changed IP's default routes etc I get the same message. ipvsadm -l
shows what is expected I was even able to see that I was making connections
to the VIP on the real servers although the were all inactive but I could
telnet to the vip from machines on the clinet network and even though they
would hang at the login ipvsadm -l would show new connections and it was
doing rr.
Tried it with http as well as the service again I could see connections
being made but I would get http timeout.
I am now isolating these machines to their own switch/hub and I am going to
set up exactly as in the mini-HOWTO except eth0:0 192.168.2.110 instead of
eth0:110 192.168.2.110 as I don't see how that makes a difference. Again I
have 2 real servers a director and a client. I will post the setup after I
make the changes and run through the configure process.
Off to start over again.
thanks