What is Node Manager?
Reference
Node Manager is a process which runs outside of WebLogic
Domains. The run time and operations of Node Manager is not depending upon a
domain security system and the run time of Administration and Managed Servers
What are the features of Node Manager?
Node Manager has the following three features
- Remote Start
- Auto Restart
- Auto Kill
Remote Start
Remote Start of WebLogic Server Instances is one of the
useful features for WebLogic Administrators. When the managed servers of a
domain are distributed across various physical servers, the Node Manager
provides a facility to start/stop all together from one location.
Auto Restart
The node manager brings up the failed server Instances
(failed for normal reasons like server reboot, power outage, unexpected PID
crash) automatically and it has built in ping mechanism to check the health of
server Instances.
Auto Kill
It’s a wonderful feature of Node Manager and it kills the
Managed Server Instances which performs low in performance. It helps the users
not to send the requests to the hanging server instances anymore.
Observations
- The administrator do not need to login into each physical server or VM to start/stop the server Instances.
- There is no need to have startManagedWebLogic script for each Managed Server Instance and the Node Manager will start those services.
- It is not required to keep several entries at rc scripts for auto restart and it is the job of the Node Manager to bring the server Instances during server reboot.
- When the Administrator updates the environment settings of start-up scripts for third party products, the node manager remote start tab (at the WLS Console) should also be updated with the same settings. During critical, pressurized deployment times, it is often possible the administrators forget to update the remote start tab and use the scripts to start. Later time if someone tries to start the server Instances using Node Manager, it will create a problem ticket. So if someone wants to use the Node Manager, a stricter change control mechanism must be followed and documented.
- When the Node Manager goes down for a longer time and the Administrator takes time to fix it, it will affect the life cycle operations of server Instances (Both planned and unplanned).
I would suggest not using Node Manager if you have 2-3
physical servers and let the Administrators login into each shell and operate
the server Instances. By not using, they may pose few operational inconveniences but the deployed applications will run without user complaints and also the
change management document becomes smaller. On the other note, recent Oracle
Fusion Middleware products (Ex: OEM, OID, OAM, OSB, WLS Portal) configures the
Node Manager by default during the install process and do not try to alter
those.
