|
Sid Stuart schrieb:
> To answer my own question, after a bit more research I find that the
> heartbeat configuration file will accept and entry without a service. The
> entry,
>
> lb1 172.16.1.1
>
> Will cause heartbeat to assign the 172.16.1.1 to the correct interface and
> to reassign it on the secondary server, should the primary server fail.
> Thanks for the help.
>
> Sid
>
> On 4/15/08, Sid Stuart <sid@xxxxxxxxxxxxxx> wrote:
>
>> Hi,
>>
>> I am using heartbeat and ldirectord in a NAT configuration to support HTTP
>> services. The problem with this configuration is the real server needs to
>> route traffic back through the active LVS server. This means the router
>> address assigned to the real servers needs to failover between LVS nodes. Is
>> there a standard mechanism to do this? I have not found it in the
>> documentation or on this mailing list.
>>
>> My first thought is to configure a dummy service in heartbeat that will
>> cause it to generate register the virtual IP. For instance, if I use
>> 172.16.1.1 as the virtual gateway address, I could create an haresources
>> entry,
>>
>> lb1 172.16.1.1 dummy
>>
>> Where dummy is a script that takes stop and start arguments but doesn't
>> actually do anything.
>>
>> Is there a better solution than this?
>>
>> Sid
>>
>>
Yes. Use heartbeat version 2. Make a group of the CIP, the DIP and the
ldirector service. Then you can monitor these resources from heartbeat
and failover if there is a problem.
Michael.
|
