Hi!

We have a Oracle RAC for our Integration Server, and a virtual ip in front of the two nodes of our Oracle RAC.
We have configured our JDBC pools in a very common way :
jdbc:wm:oracle://$VIRTUAL_IP:$DATABASE_PORT;serviceName=$OUR_SERVICE_NAME

One night, the first node had troubles, and the cluster fell over on the other node.
But the Integration Server still tried to connect to the first node.

Is it a Url problem? JVM dns cache parameter to set? Other?

Thanks in advance for your answers!

If you just connected the IS to the RAC pointing it to a loadbalancer, open JDBC connections won’t be switched, only new ones (afaik). So you need to set Minimum Connections to 0, but even this will only help if exsiting connections are closed due to inactivity and new ones created.

For connection to an Oracle RAC there is a specific database URL mentioned in the webMethods_Installation_Guide, which contains the adresses of all RAC nodes, so the JDBC driver can do the failover.

We are using an Oracle RAC with no load balancer and have no issues if there is maintenance on one node or the other. As long as one node is up we are good. We do use a slightly modified url in the JDBC connections with both nodes of the RAC listed. It can be done.