LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

LVS Failover, Failback behavior

To: "LinuxVirtualServer.org users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>, piranha-list@xxxxxxxxxx
Subject: LVS Failover, Failback behavior
From: pb <peterbaitz@xxxxxxxxx>
Date: Fri, 12 Mar 2004 11:16:07 -0800 (PST)
Hello all,

We were previously running 

   Red Hat Linux 7.3 
   Piranha 0.7.0-3
   Ipvsadm 1.21-4
   Cost:  Free :) 
   Support: Self

And now are running

   Red Hat Enterprise Linux AS v3 
   Piranha 0.7.6-1
   Ipvsadm 1.21-9.ipvs108
   Cost:  Not Free :( 
   Support:  EDU Pricing - no phone support
             Fulll Price - Full Support


Question:

What is the failover / failback behavior 
of LVS supposed to be?   

In the olden days with the free version (above) it
appears whenever the PRIMARY/BACKUP get disconnected
due to network goes down and back again, they
auto-negotiate via PULSE daemon, and the BACKUP kills
its LVS daemons and routing table so the PRIMARY is
the only one routing.  But how about if we fail the
PRIMARY to the BACKUP, and reboot the PRIMARY, then it
seems the BACKUP remains active only, and does not
fail back to the PRIMARY.  

In the current days with the not so free version
(above) it appears to match the above behavior.

The question is how is LVS supposed to failover/back
in the above two scenerios, (1) network disconnect
between PRIMARY and SECONDARY, and (2) purposfully
failover to BACKUP, then boot PRIMARY again.  How come
auto-negotiates and PRIMARY wins in case 1, but not
case 2 ?  


Thanks
Peter



__________________________________
Do you Yahoo!?
Yahoo! Search - Find what you?re looking for faster
http://search.yahoo.com
<Prev in Thread] Current Thread [Next in Thread>