Tips for rolling out a new EM11g site

Here are a couple of things to do and/or think about when doing a new EM 11g rollout:


  • Keep the Weblogic up-to-date

    The EM11g release notes mention the WDJ7 patch. Down the line other patches might be needed for the Weblogic Server (WLS). So keep an eye on this, and make sure all the relevant patches are applied.
    The tool to update the Weblogic components is BSU. It can be found in this location: <ORACLE_BASE>/utils/bsu.sh

  • Make an export of the OMS configuration

    A lot of config work goes into the OMS, all done by the install automatically. And as long as no disaster strikes (like a disk crash), it will all work just fine.
    But once Murphy decided to pay a visit, you'll need the OMS configuration in order to recover the OMS and get the 'original' configuration back online.
    To get the config exported, run this:
    $ emctl exportconfig oms

  • Have a copy of the emkey.ora file available

    On the 1st OMS installed in the environment, a copy of that file will be kept in the directory <OMS_HOME>/sysman/config.
    This key is absolutely crucial in order to do all the SSL traffic, and work with credentials for jobs, target logins, etc...
    And it's the same story again: As long as no disaster strikes, there is no need for this file. But once lightning strikes the Weblogic server, you'll need this file to get back on track.
    If you are security-conscious, you can keep an 'offline' copy of this file (and the exported OMS config from the previous point too for that matter) in a 'secure location' somewhere else: Just as long as a backup of these files exists, and can be easily retrieved in case of trouble.

  • Plan for growing your site

    Sometimes you are lucky, and now exactly how the site will grow, and how many targets will be there. But this is not always guaranteed to be the case... For those sites, where it starts small with one OMS, but you can 'smell' those opportunities down the line that will make the site grow, make sure you have the growth plan already in place for your infrastructure!

    For the repository database, growing and scaling out means going to RAC. And this means the additional CRS software and shared storage for the database. Make sure the hardware the repository database resides on, is capable of making that transition to RAC (and more importantly: Is it practically do-able?). You can always move the repository database to 'better' hardware, but if you can predict the work, and prepare the site, you can make things a lot easier down the line...

    For the OMS's, there are two things to consider:
    • As soon as there is more than one OMS, the Agents will either have to decide which OMS to talk to (which is NOT a HA-way of doing things!), or you have to put some kind of load balancing (SLB) technology in front of the OMS's.
      But putting an SLB in front of the OMS means a new 'machine' (Read: Virtual hostname) enters the infrastructure. And that means the communication between the Agents and OMS's will change. Result: You will have to re-secure ALL the agents again. Not so much of a problem if you have only 5 Agents. But that will become quite painful if you have 500 of them in your environment...
      To prevent this headache, install and configure the 1st OMS already using a virtual hostname using the ORACLE_HOSTNAME parameter during the install. By doing so, once you add a 2nd OMS and put an SLB in front of the OMS's, the SLB can use the virtual hostname, forwarding the traffic to the two (or more) physical OMS hostnames behind the scenes. But more importantly: This change is completely transparent to the Agents, eliminating the need to make any kind of change on the Agents.
    • The second thing to consider is shared storage, for both the XML file loading, and the software library (used for patching and provisioning). So make sure you can add a good shared storage solution to the OMS machines when needed to handle the shared files.

  • Discover the entire infrastructure

    There is more to the Grid Control infrastructure that an OMS (middle tier). To make sure you get all the right information, and receive all the appropriate alerts, make sure all the OMS's, as well as the repository is discovered properly in Grid Control.
    While doing an OMS install, you will get an Agent installed with it automatically: Just make sure to refresh the Grid Control Weblogic domain each time you make a change to the OMS's!
    For the repository database though, you will need to install an Agent on the machine (or all the cluster nodes in case of RAC) in order to make the repository database information available in Grid Control.

  • Configure the access to 'My Oracle Support'

    Besides the CPU (Critical Patch Update) information, it will also allow you to package 'incidents' on the 11g databases and the 11g Fusion Middleware software, and send them to the support website to create an SR for those automatically.

  • Add a second loader thread (if not done so already)
    For the bigger enterprise sites, this will already be done (and will be most likely using more than 2 loader threads).
    The default install out-of-box sets up the OMS's with just 1 loader thread. And this works fine for small sites and for staging/test sites. While the general rule is NOT to waste resources by creating pointless additional loader threads, the idea is to have spare capacity in case of a spike, or a burst of incoming information.
    And following the HA (High-Availability) mantra, the rule of 2 applies: Always a 'spare' at hand to increase availability!
    To setup the second loader thread, use this command:
    $ emctl set property -name em.loader.threadPoolSize -value 2 -module emoms


Comments:

Just curious why the 11g documentation talks about Oracle Web Cache and 10g. Was it that Web Cache is so awesome, Oracle just couldn't quite leave it out of the 11g documentation? Oracle® Enterprise Manager Administration 11g Release 1 (11.1.0.1) Part Number E16790-02 Starting and Stopping Oracle Application Server Web Cache By default, when you install Oracle Enterprise Manager 10g, the Grid Control Console is configured to use Oracle Application Server Web Cache.

Posted by jeff wiegand on June 24, 2010 at 05:35 AM PDT #

Werner, A very good article. We are installing Oracle Enterprise manager 11g at our customer site and are running into following issues - 1. In our DIACAP environment (a secure environment for federal government compliance), when our adminstrator is trying to install OMS, the installer cannot find WLS installation. WLS installation exists and is working as I have created test domains. We cant seem to find any log file that can provide us a clue as to why EM thinks WLS is not there or would like to know what EM checks to validate WLS exists. Any help appreciated. Thanks, Vijay

Posted by Vijay Mandava on July 01, 2010 at 10:01 PM PDT #

Webcache is not relevant anymore for 11g. All references for that will get removed from the documentation. Stay tuned: Refresh of the docs coming...

Posted by werner.de.gruyter on July 02, 2010 at 08:15 AM PDT #

When is the Documentation going to be refreshed? It is very confusing currently and causes tremendous problems when we are trying to sell this to customers. Also, anyone know how to safely cleanup the DB Repository? My customer sees this grow over 200GB in 2 weeks. They refuse to go in production without a cleanup process they can use to remove older data. Thanks in advance. A PS: Great Article

Posted by Afzal Memon on July 19, 2010 at 05:07 AM PDT #

Thank you for your article,It's well written, in depth posts like these that have aided me in becoming a good forex trader.

Posted by shaiya gold on August 02, 2010 at 11:33 AM PDT #

I simply wanted to let you know, I really like the posts on your website. But I'm using my Firefox Browser and the display does not appear to be quite right. It's certainly not a big deal as I am still able to see and read the postings but just thought I would inform you about that. Keep up the superb Cheers.

Posted by Kamagra on August 27, 2010 at 04:03 AM PDT #

Thank you for a great post

Posted by Eye test on September 16, 2010 at 11:15 PM PDT #

Post a Comment:
  • HTML Syntax: NOT allowed
About

Latest information and perspectives on Oracle Enterprise Manager.

Related Blogs




Search

Archives
« April 2014
SunMonTueWedThuFriSat
  
3
5
6
7
9
10
11
12
13
14
15
17
18
19
20
23
24
25
26
27
28
29
30
   
       
Today