By Jstiles-Oracle on May 15, 2015
By Brad Herren, Senior Principal Instructor, Oracle University
If 100 questions are asked during a week-long WebLogic class, about 75 of them will be about Node Manager.
It's interesting - people come to the class with WebLogic experience, yet have no idea what this Node Manager thing is or does.
A common misconception about Node Manager is that it actually does more than it does (a few people think it doesn't do half of what it actually does). Basically, you can use Node Manager to start your servers, while also involving it in the shutting down side of your servers.
The big value: if Node Manager started the server and the server crashes, or the server fails for some reason, it will automatically restart that failed server.
Make a WebLogic domain interact with Node Manager: 11g versus 12c
In a multiple host environment, let's say we've got two computers with our WebLogic installation on both of them and we have domains scattered across the two hosts.
From the admin console, an application that runs on a WebLogic admin server, the admin server is able to figure out how to get a computer started over here on a remote host. We accomplish this remotely with the Node Manager utility.
In a multi-host environment, we would have Node Managers on both hosts, such as Node Manager on Host01 and a Node Manager on Host02, running out of port 5556. And again, assuming you went with the default configuration, they would both be sitting here, specific to each host. That's how we did it in WebLogic 11g.
Version 12c introduced a new way of handling this, where the Node Manager is now domain-specific. Instead of running out of WebLogic home, the Node Manager will run out of the Domain Home. It's still going to the be the same JVM used to run out of WebLogic Home, but now the Node Manager is domain-specific and the domain would be configured to interact with Node Manager.
Failed servers and stuck threads
Node Manager can help with failed servers due to stuck threads (thread running 10 minutes or longer) and out-of-memory exceptions. If stuck threads are caused by faulty code, configure WebLogic server to have the server "kill" itself after a defined number of threads.
Node Manager can restart the server to solve the issue. In the webinar, you will see how Node Manager starts a server and how it restarts a failed server. With a free simulation application, StuckThreadForFree, you can get Node Manager to restart failed servers by creating a simple "kill" of the process and a server with stuck threads.
I hope you can join me for the full webinar, when I will demonstrate a server start/restart/shutdown flow with Node Manager, as discussed in this post. The webinar, WebLogic's Most Misunderstood Feature, is available through Oracle Learning Streams.