TCPIP cluster application, DNS, and Node accessibility

[Sergey Akimov]

Hi, All !

I have a DNS-cluster (Windows 2000 DNS server) with some Alpha servers working under OpenVMS Cluster V7.3-2 (TCP/IP V5.4). One instance of my application working ON EACH cluster node. It is not a classic VMS TCP service (inetd -> how many connects - as much processes) but usual VMS application listen for some TCP/IP port and processing all client connections/queries in single process (common loop with ASTs).
Availability of every node through DNS controlled by LOADBROKER+METRIC (DNS dynamic update). When node failed, no client connections established to it (it is normal).
BUT! When only my application failed and node still working, LOADBROKER+METRIC (i.e. DNS server) allow client to connect to node, but my server application not responding (because it is dead on this node)!
QUESTION: How can I tune LOADBROKER+METRIC to allowing client connections (access) to node only if some TCPIP ports opened/listened (i.e. my application working) on that node?

Thanks! And sorry my English. :)

[Ian Miller.]

Sergey,
(don't worry about your english - there are plenty of non-native english writers here allthough I don't know if there are anybody who can read Russian.

You have supplied the tecnical detail, thats the main thing.)

I guess this is related to this thread.
http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=786224

Some Background reading
http://h71000.www7.hp.com/openvms/journal/v2/articles/tcpip.html


Ideally you would somehow update the system metric if your application is not running.
Parhaps you could shutdown the metric server when your application is shutdown?