Oracle listener issue on HP-UX B.11.31 serviceguard...

[max0907]

Hi

We have a SG on HP-UX B.11.31.

We have multiple Oracle databases running with separate homes and listeners.

As of now all the listeners listen on VIP on different ports

For a new project, we wanted one of the listeners to also listen on the static host IP on the same port as it is listening on for the VIP.

I modified listener.ora to include the static IP and tried to reload the listener. After reload the listener would still not listen on the static IP.

So I stopped and started the listener after which I observed that the listener started listening on the static IP. However as soon as I did it the entire cluster crashed.

Now this is a bit weird for me, why would the whole cluster crash just because the listener was stopped and started, on listener stop the expected behavior would only be for the cluster to restart the lisener on the same node, isn't it?

Has anyone observed this behavior at their end?

Secondly, why would the listener not listen on statci IP as well only with a reload, why is a bounce required?

[Laurent Menase]

Hi,

I can only answer to the second question.

If you have 1 listener bound on inaddr_any/port1 and then try to bine an other listener on ip1/port_1, this won't work (bind will fail) except if you set the setockopt(SO_REUSEPORT)

The only thing which could work would be to have the 2 listenr bound explicitely on a different couple

listener1   ip1/port1

listener2 ip2/port1