openstack team mailing list archive

[Brent Roskos <brent.roskos@xxxxxxxxxxx>]

I stand corrected.  Mostly confused since the keepalived didn't actually
need addresses in the multicast IP range.  It does use it - as I can see
with ifconfig.  We minimized the  impact of this by creating a small subnet
that just had the switch address, host addresses and vrrp address in it.
 All the chatter was contained within that block.

We avoided pacemaker in this particular instance because the keepalived
setup and configuration was so very simple - only a couple of lines in a
config file, and because we didn't need any of the other available HA
features.

Brent


On Wed, Jun 26, 2013 at 10:03 AM, Jesse Pretorius <jesse.pretorius@xxxxxxxxx
> wrote:

> On 26 June 2013 15:42, Brent Roskos <brent.roskos@xxxxxxxxxxx> wrote:
>
>> I've also used keepalived for services that did not scale laterally.  In
>> this case I put two horizon servers behind an active/passive virtual IP.
>>  This was also pretty simple as there was no need to maintain state
>> information in for active passive. That wouldn't work quite as well when
>> capacity thresholds started to become a concern.
>>
>> Neither of the above required multicast support - which really helps with
>> deployment options.
>>
>
> *ahem* keepalived most definitely requires multicast support for its
> vrrp... and it's quite noisy. If there's a way to make it use unicast
> instead, I'd definitely like to know.
>
> corosync & pacemaker can do a virtual IP failover between as many nodes as
> you like using unicast instead of multicast.
>
> _______________________________________________
> Mailing list: https://launchpad.net/~openstack
> Post to     : openstack@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~openstack
> More help   : https://help.launchpad.net/ListHelp
>
>