LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: LVS, Squid and heartbeat ?

To: "LinuxVirtualServer.org users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: LVS, Squid and heartbeat ?
From: Alexis Januskiewicz <ajanuskiewicz@xxxxxxxxxxx>
Date: Tue, 01 Mar 2005 13:49:37 +0100
Hi,

i'm sorry, i've not been verry precise in my request... I'll try to be more exact in my answers ;)

Mack.Joseph@xxxxxxxxxxxxxxx wrote:

are you using LVS-NAT or LVS-DR?
LVS-DR. I needn't NAT in my configuration.

Of course, I've read a lot of docs but nothing works on my config.
I assume "nothing" is described below?
Not really ;)
Many of docs I've read used at least 3 computers : 2 real servers, and 1 director. In my case, I'd like to use only 2 computers. Each runs squid, heartbeat could switch between them in case of network failure, and ldirector could switch between RIP in case of squid failure.

Heartbeat is running perfectly, and when I switch off a
computer, the second takes hand. When I kill squid,
ldirectord delete server from the
cluster (ipvsadm -Ln shows me this) but nothing happen
with heartbeat and, consequently, there's no failsafe :(
What is "nothing"? What did you expect it to do?

I'd like than (if squid is killed on the main node) connections to my VIP be automatically switched on the second real server.

I assume you have heartbeat setup to failover the directors.
If a realserver fails, the director doesn't failover too.
I wasn't clear, i'm sorry.

I've tried to run ldirectord withtout heartbeat in debug mode.
Real servers are correctly discovered and squid monitored.
IPVSAdm give the following result :

rs1:~# ipvsadm -Ln
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
  -> RemoteAddress:Port                      Forward Weight  ActiveConn      
InActConn
TCP  10.0.0.234:3128 dh
  -> 10.0.0.233:3128                 Route   1       0       0
  -> 10.0.0.232:3128                 Local   1       0       0

But when I try to telnet on 10.0.0.234 (or to ping the VIP), I can't (packets lost). Of course, if I enable heartbeat (which creates a virtuel eth0:0 with 10.0.0.234 IP), i can telnet on my virtual server and the first node correctly answers. If I kill squid on the first node, ldirectord correctly discover the failure and the node is deleted from IPVS. But if I try to connect to 10.0.0.234 :

rs1:~# telnet 10.0.0.234 3128
Trying 10.0.0.234...
telnet: Unable to connect to remote host: Connection refused

Do you understand my problem ? I hope I'll be more clear !
--
Alexis

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