LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

strange behavior

To: <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: strange behavior
From: Ricardo Kleemann <ricardo@xxxxxxxxxxx>
Date: Fri, 26 Oct 2001 11:47:58 -0700 (PDT)
Hi,

I'm using the ultramonkey set of packages to implement LVS. Here's what I
have:

heartbeat-ldirectord-0.4.9-1.um.1
ipvsadm-1.14-1
kernel-2.2.18pre11-va2.1

Operationally I have no reason to believe that lvs is not working. It
seems to do its job quite fine.

However, analytically, ipvsadm output seems to tell me that something
isn't working right.

Here's a sample output, and I'll explain what I mean:

IP Virtual Server version 1.0.2 (size=4096)
Prot LocalAddress:Port Scheduler Flags
  -> RemoteAddress:Port          Forward Weight ActiveConn InActConn
TCP  192.168.1.102:80 wlc
  -> 192.168.1.32:80             Route   10     3          142
  -> 192.168.1.31:80             Route   10     4          80
  -> 192.168.1.30:80             Route   10     3          91
TCP  192.168.1.101:80 wlc
  -> 192.168.1.22:80             Route   10     21         65
  -> 192.168.1.20:80             Route   10     19         93
TCP  192.168.1.101:443 wlc
  -> 192.168.1.20:443            Route   10     12         21
  -> 192.168.1.22:443            Route   10     13         10


Again, seems fine. However, I can go to 192.168.1.22 and stop apache
altogether, and ipvsadm will continue to show it as in the loop. It's not
that the data stays the same, it will decrease the connection numbers
somewhat, but I would think that it should completely remove it from the
cluster, no? However it will only really remove it from the cluster if I
restart ldirectord.

Does that make any sense?



<Prev in Thread] Current Thread [Next in Thread>
  • strange behavior, Ricardo Kleemann <=