Re: [Keepalived-devel] Maximum number of IP addresses?

[Todd F. <to...@fl...>]

Tijn,
You mention using just 1 vrrp instance and that you can ping the master from the backup node. If you are not seeing a transition, then I suspect you will find the master node is still sending advertisements with a higher priority. You can confirm this with tcpdump. This may or may not be relevant to your issue, but if these nodes have multiple interfaces and are acting as gateways between multiple networks then I'd highly recommend running a vrrp instance on each interface and then put them in a vrrp_sync_group if appropriate.

You mention the master node becomes unreachable from the outside but is pingable from your backup node. Is ping the only thing that responds? What happens when you try to contact other tcp ports that should be listening? Does keepalived manage the default routes of the nodes? Can you console onto the master node while it's in this state? After rebooting it do you find anything in the logs that might indicate what state it was in prior to the reboot? It also might help to see a diagram of your nodes & their configurations.

As for the maximum number of IP addresses you can have, I'm not sure if there is a limit. My configurations have up to 20 virtual_ipaddresses defined and then any new ones go in the excluded section. The most I've currently got is around 50. But I suspect if you are able to load all of your addresses in at first and it takes a few hours before things get weird then you are not hitting any limit in keepalived.

-T

On Mar 16, 2013, at 1:13 AM, Tijn Buijs wrote:

> Hello everybody,
> 
> At the moment I'm busy configuring 2 gateways that will go in front of an entier network. These machines are only the default gateway for the network. Now I want to load my entier iprange (I've got a IPv4 /24 to play with) to keepalived, in order to have all the external ipaddresses always available. I know the VRRP protocol can only send over 20 IP addresses at a time, so for a while I used multiple vrrp_instances, to have all the ip addresses fail over to the other host, in case of a problem. The other day, I switched over to using 1 vrrp_instance with 1 external ip address in the virtual_ipaddress section, and the rest in virtual_ipaddress_excludes, so they aren't included in the vrrp packages anymore.
> 
> The problem that I'm having now is that the machine itself becomes unreachable after a while (it takes several hours). I can't SSH to it anymore, the webserver on it is down, nothing. But there isn't a failover to the second gateway. I can SSH to the second gateway and from there I can ping to the first one. But I can't from my desktop (which isn't on the same network, the connection between my desktop and the 2 gateways goes over the internet).
> When I reboot the server and don't let keepalived start during the boot, everything is working fine and I can keep reaching the server.
> 
> What could cause these kind of problems?
> 
> With kind regards,
> 
> Tijn
> 
> Send from my tablet
> 
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_d2d_mar_______________________________________________
> Keepalived-devel mailing list
> Kee...@li...
> https://lists.sourceforge.net/lists/listinfo/keepalived-devel