LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: Help debugging director problems after a period of time

To: "LinuxVirtualServer.org users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: Help debugging director problems after a period of time
From: nick garratt <nick-lvs@xxxxxxxxxxxxxx>
Date: Thu, 8 Apr 2004 08:59:21 +0200
with ifup it will definitely matter. my DR service aliases use the same netmask as the primary interface on the director and i use ifup/ifdown to control them without an issue. i'm using FWMARK for all my services, DR and NAT, which may well make a difference.

i use DR for cluster internals (web services, etc) only, so the service IP is on the same subnet as the real server primary interfaces and the client hosts. this has been working well for me for the last year or so doing healthy volumes.

                       |
                  internet facing interfaces
                  --------------
                  | director   |
                  --------------
                  internal interfaces
                  eth0 192.168.10.1/16
                  eth0:0 192.168.20.100/16
                       |
                       |
                       |
                       |
                       |        eth0 192.168.20.1-n/16
                       |         ----------------
                       |-------- |real server1-n| dummy0 192.168.20.100/32
                       |         ----------------
                       |
   client hosts  ------|
 all on subnet 192.168.0.0/16


director configured:
iptables -t mangle -A PREROUTING -d 192.168.20.100 -i eth0 -p tcp -m tcp -j MARK --set-mark 8888
ipvsadm -A -f 8888 -s wlc
ipvsadm -a -f 8888 -r 192.168.20.1:0 -g -w 5
ipvsadm -a -f 8888 -r 192.168.20.2:0 -g -w 5
ipvsadm -a -f 8888 -r 192.168.20.n:0 -g -w 5


nick


On Wed, Apr 07, 2004 at 07:07:35AM -0400, Joseph Mack wrote:
 nick garratt wrote:
 >
 > hi
 >
 > your problem is is that your alias eth2:0 has a 255.255.255.255 mask
 > whereas it should have the same mask as eth2, being on the same
 > subnet.

 he has the correct netmask for LVS-DR

As the interface is on eth2 (as opposed to lo) it should probably
have the netmask of the network. Though it most likely
does not matter.

--
Horms
_______________________________________________
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

<Prev in Thread] Current Thread [Next in Thread>