|
On Fri, Jun 27, 2003 at 06:47:16PM +0200, Torvald Baade Bringsvor wrote:
> Hello.
>
> We have a LVS cluster using ldirectord and heartbeat. This has been working
> ok, but I have one problem: when we do a failover from primary to secondary
> node and back again, the ldirectord processes on the secondary node aren't
> shut down. I think this behaviour started after the upgrade to Heartbeat
> 1.0.2, but I am not sure. We are using IPVS 1.0.9 in case that matters.
>
> Have you got any suggestions as to what might be the problem here?
Ldirectord running on both nodes shouldn't be a problem.
But if you want it to only run on the active Linux Director
heartbeat should be able to manage this.
1. Make sure that your haresource file looks something like this:
http://www.ultramonkey.org/2.0.1/topologies/config/ha-lb/non-fwmark/linux_director/haresources
2. Make sure that the haresources (and other heartbeat configuration
files) are _identical_ on both linux directors.
3. Make sure that ldirectord is not configured to be started by init.
> Also we have seen a couple of times that the secondary node ARP's one of the
> virtual addresses, but it doesn't have an interface for the address (the
> other node has), herefore the service is unavailable. Have any of you seen
> this one?
No that is a new one to me. The bringing up of the interface and
sending of arp is done by the same resource (IPaddr) so it supprises
me that one happens without the other.
--
Horms
|
