Re: [CentOS] Floating VIP...

2012-05-14 Thread John Doe
From: Tait Clarridge t...@taiter.com

 Also, use IPaddr2 as the resource type as it will do an gratuitous arp,
 significantly decreasing the failover time.

From: Dennis Jacobfeuerborn denni...@conversis.de
 Hearbeat isn't developed anymore so if you want to go the pacemaker route
 you probably want to go for corosync+pacemaker instead.

Thx for the tips!
JD
___
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos


[CentOS] Floating VIP...

2012-05-11 Thread John Doe
Hi,

right now I am using only one external server as a gateway for the internal 
servers.
I would like to enable a fail-over on a second server.
To implement the floating VIP, should I use heartbeat+pacemaker?
Or is there something more lightweight?
Basically, I just need server2 up the VIP when server1 is down, and server2 
down the VIP when server1 is back up (or server 1 does not up the VIP seeing 
server2 has it already).

Thx,
JD
___
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos


Re: [CentOS] Floating VIP...

2012-05-11 Thread Tait Clarridge


On Fri, 2012-05-11 at 05:40 -0700, John Doe wrote:
 Hi,
 
 right now I am using only one external server as a gateway for the internal 
 servers.
 I would like to enable a fail-over on a second server.
 To implement the floating VIP, should I use heartbeat+pacemaker?
 Or is there something more lightweight?
 Basically, I just need server2 up the VIP when server1 is down, and server2 
 down the VIP when server1 is back up (or server 1 does not up the VIP seeing 
 server2 has it already).
 
 Thx,
 JD

Heartbeat is a pretty lightweight way of doing this. With any
failover/VIP setup there is always the possibility of a split-brain but
for the most part I haven't seen this happen in my environment.

Also, use IPaddr2 as the resource type as it will do an gratuitous arp,
significantly decreasing the failover time.

eg. [haresources file]

node.fqdn IPaddr2::10.10.10.10/24/eth1

Substitute your IP, mask and network interface.

There are some options you can set as well that tell it not to fail back
automatically, just so you can check the box that failed before pushing
services back to it.

Tait

___
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos


Re: [CentOS] Floating VIP...

2012-05-11 Thread Dennis Jacobfeuerborn
On 05/11/2012 02:40 PM, John Doe wrote:
 Hi,
 
 right now I am using only one external server as a gateway for the internal 
 servers.
 I would like to enable a fail-over on a second server.
 To implement the floating VIP, should I use heartbeat+pacemaker?
 Or is there something more lightweight?
 Basically, I just need server2 up the VIP when server1 is down, and server2 
 down the VIP when server1 is back up (or server 1 does not up the VIP seeing 
 server2 has it already).

If you only need IP handling then you can use keeplived which is more
lightweight than pacemaker but can only deal with networking and nothing else.

Hearbeat isn't developed anymore so if you want to go the pacemaker route
you probably want to go for corosync+pacemaker instead.

Regards,
  Dennis
___
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos