|
First, I want to say that I'm really amazed why people always post
questions about heartbeat to the LVS mailing list. Must be something
to be with heartbeat being included in ultramonkey.
Second, I'm also amazed why people fail to search the mailinglist archives
and read the documentation :-(
"Schillaci" <lvslog@xxxxxxxxx> wrote:
> I am using heartbeat 0.4.9 in two machines in our network just for ip take
> over(may be for web-service). All my machines are in the same subnet. I
> have the following problem. When I stop the heartbeat in the node which has
> the take-over ip, it realeases the ip and the other machine takes it. And
> the take over is very slow. But if i remove the ethernet cable it is not
> working.
I suppose you have also a serial cable for heartbeating? If yes, then the
backup node can still communicate to the active one and will not fail
over.
> Sometimes when releasing the ip, the heartbeat shuts down, which
> prevents this machine from taking the ip again when the active node fails.
> Which node will act as an active-node? I could n't find any
> configuration specifying this.
> Should I add anymore hardware?
> Is the node name in the /etc/ha.d/haresources files in two machines should
> be same or it should be the corresponding one?
The node name in haresources is the node who should be the default master.
> One more thing is that, somemore machines are also running heartbeat in
> the network. And these machines are checking that also. why this. How to
> avoid?
Heartbeat broadcasts per default on port 694. Each machine running
heartbeat in the same subnet will hear this heartbeat. Just change the port
in ha.cf for other clusters pairs.
>
> Somebody please help me to overcome this problem. And I want this ip to be
> changed to the other machine when one machine fails and vice versa.
Please send further questions to linux-ha@xxxxxxx
See also http://www.linux-ha.org/
Juri
--
Juri Haberland <juri@xxxxxxxxxxxxxx>
|
