Now I have two servers, in the first one I have the admin server and
two managed servers. In the second one, two managed servers. When I
start my linux server, I execute the StartNodeManager.sh script in each
computer and startWeblogic.sh in the first one. After, I start my
managed server by the Web Admin Console.
The problem is that I want to write linux boot scripts, to start all
the stuff automatically.
My question is: can the Node Manager start a Managed Server at boot? I
can make a daemon with startNodeManager, but how I have to configure to
start at this time, the four managed servers?
I looked the doc and the forum, but I can't find an exactly solution...
Thank you very much,
As i remember well, you have to configure Node manager on Admin server.
Then allow port 5555 or define some of your own, allow "starting via
node manager".
There is "tab" in WeblogicConsole. Before, you better to test if your
node manager port is open on all instances/boxes to avoid some issues.
After this "admin server" will keep basic (core) information for all
the servers and he can bring them up or down or to standby on you will
(there will be created file/folder structure on the admin server)
Anyway you need to have adminserver up and running. So usually it is
better to have it on another machine (not in cluster, not in let's say
paralle running on same box -> in case of any issues with java there is
possibility you need to bring down managed but you will be forced to do
so on admin one).
If you are using "loadbalancing" option in weblogic, node manager allow
you to bounce instances or manually load/unload some resources over the
"domain/cluster".
It is helpfull, when you want to re-deploy application, just say via
node: go down, via admin you push there application and say :go up, and
users just will not notify any difference :)) That's the power of
weblogic.
After this you just care of admin server and cover all stuff from it
(one box with more admin servers, to support more domains..:),
So you put some own script to "crontab" or some "inet.d" or "rc.d"
whatever you like.\
Those scripts will take care about the life (start, stop, restart
...etc) of the admin one.
and you just login to web console, that's it. You can abandon CLI.
There is way to put admin server down after you have all "working"
servers up, coz you need him only due some "managing/deploying" task,
that's why i recommend to separate managed ones from admin one. If you
check my old thread, there is little bit about clustering and stuff
around it.
I have to say, that if you don't need this feature, don't use it. The
communication over the node manager port is quite massive and if your
"weblogics" are not so wide over the machines (let's say 12 servers
with 24 instances ....4 cluster layers and so on...) you can do same
work via the CLI and scripts...sometimes i say it is "gui wizarding"
instead of running some regular scripts...
I wrote this from scratch, so i hope i does not forgot anything
important, but as small push it is ok i guess.
cheers
-vh-