Unanswered: RPC settings cause Cluster Service to fail
My customer has an application running on a web/app server which currently connects to a stand alone database server. The app uses RPC and according to the following MS KB article the registry was edited to allow RPC to pass through the firewall on ports 5000 - 5020: http://support.microsoft.com/default...b;EN-US;250367
They are trying to move the database onto a cluster and made the same registry changes to the two nodes of the cluster. As soon as the primary was re-booted, Cluster Services (which use RPC to connect to the secondary node) failed to start. Only when the registry changes were backed out would Cluster Services start and the cluster came back up okay.
Clearly the RPC configuration crippled Cluster Services, however the customer app still requires RPC and I have been unable to find any fix to this on MS or any forums. The quick fix is to open the firewall to allow RPC to use its default setting of all high ports (above 1024) but this is too much of a security risk.
Is there anybody who has come across this before? If so what was the answer?
Sussed it thanks. The Cluster Service also uses RPC, and pretty intensively. I added the registry change again but this time allowed RPC to use ports 5000 to 5100 (100 ports instead of 20). The cluster came back online straight away.