Friday, March 14, 2008

ITM AIX Premium agents - an Overview

I recently got a chance to implement AIX Premium agents for one of our customers in a production environment. This article briefly discusses about our experience with these agents and also discusses about the pros & cons of these agents.

Installation

Installation of these agents is similar to Unix OS agents. There is a minor gotcha though. IBM currently offers "System P" SE edition as a free download with 1 year support, but don't confuse them with the "AIX Premium" agents which are available for current ITM customers. The "System P SE Edition" consists of agents for AIX Base, VIO and CEC whereas the "Premium" one consists of "AIX Premium", CEC and HMC agents. (Check with your IBM Sales rep about your entitlement).

Use the "AIX Premium agents" C101HIE.tar from Passport advantage and installation is very similar to usual agent installation. Make sure that you install the agent support files on your TEMS/TEPS servers.

What is the difference?

So what is new with "AIX Premium" agents? Of course the workspaces are different and the attributes provide AIX specific information such as LPAR, entitlement, Volume Group information and paging space that are NOT available with the generic Unix OS agents. You should be able to define situations to monitor these resources just like you would do for Unix OS Agents. This information could be very useful for AIX administrators.

You can find more information about this agent at the IBM Information Center for AIX Premium agent.

Rollout Considerations

The advantage of getting AIX specific information is really nice and most of admins would like it better than the Unix OS agents. However, there are couple of factors that you might want to look into before deciding whether to go forward with System P agents. One thing is the level of support and fixes available right now. Currently the UNIX OS agent is part of the "core" ITM release strategy and gets updated with every fixpack whereas AIX Premium agent is pretty much like the special type of agents such as DB2 agent, SAP agent etc. Since this is a fairly new type of agent, we don't know whether it will be integrated into IBM Fixpack release strategy.

One other issue is the added complexity of managing another type of agent. If you are happy with the current UNIX OS agent, then you could probably experiment it in your test environment and see if you need the features of AIX Premium agents.

An Introduction to Netcool Omnibus Components

We have already discussed lot of things about Omnibus in our previous articles but here we are going to explain the basic components of Netcool Omnibus and their functions.

What is an ObjectServer?

ObjectServer is the central piece of Omnibus and can be considered analogous to the TEC Server. The ObjectServer receives events from various probes
and processes them according to automation rules. The ObjectServer, however, is a light-weight process when you compare with TEC. Moreover, Omnibus provides
easy to use high availbility features out of the box, so it is normal to have two or more ObjectServers running in an environment.

What is a Probe?

Probes can be thought as an equivalent of Tivoli Event Adapters. They receive events from various sources, such as SNMP agents, Unix log file, etc. Omnibus
probes are much more intelligent than TEC event adapters in the sense, it can detect/filter duplicate information at the source and it is capable of advanced programming capabilities such as looping, regular expressions, etc.

Gateways

Gateways in Omnibus are different from the Gateways in Framework. In Omnibus, they act as a bridge between an Object Server and other ObjectServers, third-party applications or databases. Their function is to replicate ObjectServer data from one ObjectServer to other components such as another Object Server (for redundancy), or to third-party applications such as Remedy (for trouble-ticket management), or to databases such as Oracle (for event storage).

License Server

License Server is a central piece of Omnibus environment that dispenses necessary server and client licenses as and when requested. Without the license server, the Netcool components will not start. If the license server goes down AFTER the component is started, it should be back up within 24 hours or the Netcool components will shutdown. There are High Availability features supported for License Server and most of the time it is pretty stable. However, IBM understands the pain of managing one more component and ever since the acquisition, IBM is focussing on moving away from license server and expect this to be out in future releases.

Process Agent

The Process Agent plays a very important role in the overall architecture. It is responsible for starting and stopping Netcool components and it also restarts these components in case they died abnoramlly. Process agents are also responsible for executing automation requests coming in from remote systems. However, ironically the Process Control Agent does not manage processes on Windows, it is used only for executing requests received from remote systems. The process control agent works very well and it is very nice to see all the components started and managed by a single program. May be ITM could take a leaf out of it and use a similar solution!

Proxy Server

The Proxy Server is an optional component that can be used for better scalability and for firewalls. The Proxy Server acts as a proxy for ObjectServer, receives events from various probes and sends them over a single connection to the real ObjectServer. This will reduce the number of connections that the real ObjectServer has to handle. An equivalent in Tivoli world, Gateway comes to my mind!


These are the basic components you should know about. Stay tuned for more articles on Netcool Omnibus in the coming days.

Situation Status Field in ITM6.1

The situation_status field in TEC uses single letter status to denote the current status of the situation. Understanding the different values of this field is important if you need to write TEC rules for incoming events so that you don't end up taking multiple actions for the same event. This article lists the different values for Situation_Status field. The article credit should go to IV Blankenship.

According to IV in one of our earlier blog articles, the following are the valid values for situation_status field.

N = Reset
Y = Raised
P = Stopped
S = Started
X = Error
D = Deleted
A = Ack
E = Resurfaced
F = Expired

Here is the link to the earlier blog aricle/comments.

ITM FP05 Fixes & Omnibus Fixpacks

There are few interesting patches released at Tivoli Patches site including an interim fix for ITM 6.1 FP05 and a fixpack for Netcool Omnibus/Webtop. Here are the links to readme files.

ITM Interim Fix to Fixpack 05

Netcool Omnibus Fixpack 03

Thanks to Martin Carnegie for bringing this to our attention.

Using the APDE with JDBC Type 4 drivers

The APDE (Application Package Development Environment) can either be used on the TPM server or can be installed on a remote computer for the development of workflows. In Fix Pack 2, documentation was provided on how to configure the APDE to use JDBC Type 4 drivers instead of having to install the DB2 client on the remote computers. The problem was that this documentation jumps all over the place. So I thought I would document the steps I used to install the APDE and now I will share them with you.

Background
The APDE is the development environment in TPM for building custom workflows (and other things). Even though you could develop these workflows in the Workflow Editor or even notepad, the APDE provides an excellent IDE that allows for easier creation of custom workflows.

Steps required for DB2 connectivity on Windows
These instructions are for installing the APDE on a Windows system. The same principles should apply for other OSs.

1. Install Java 1.4.2+ confirm that the environment is setup by opening a command prompt and type java -version (Download from www.java.com
2. Download Eclipse from www.eclipse.org. This has to be 3.1.2. Version 3.2 does not work. (Download from www.ecplise.org
3. Extract the eclipse zip file to C:\Program Files (actually directory could be anywhere)
4. Copy the apde.zip and apde.upgrade.zip to remote computer
5. Extract apde.zip and apde.upgrade.zip
6. Copy/move the contents of apde.zip to the eclipse directory and overwrite existing files
7. Copy/move the contents of apde.upgrade.zip to the eclipse directory and overwrite existing files
8. Create a directory under the C:\Program Files\eclipse directory called TPMCFG
9. Copy the files db2jcc.jar, db2jcc_license_cisuz.jar, db2jcc_license_cu.jar from DB2_HOME\java on the TPM server to the TPMCFG directory
10. Copy the crypto.xml and dcm.xml files from TIO_HOME\config to the TPMCFG directory
11. Edit the new dcm.xml file and modify the file to the following values:
...



...
jdbc:db2://:/tc

The server name will be the name of the server where your TPM TC database resides (should be the TPM server). The db2 port is the port db2 is on, most likely this is 50000 or 60000 (this seems to happen more on Linux)
12. Run the eclipse.exe from C:\Program Files\eclipse
13. Go to Window -> Preferences -> Automation Package -> Database
14. Press the Import button and select the dcm.xml file from the TPMCFG directory. Confirm all the settings are correct. I had to modify the password as this did not seem to import correctly.
15. In the Import driver, I had to select the db2jcc.jar file and press the Add button or I received db connection errors
16. Press OK and restart the APDE

If you want to set up the APDE to allow for dynamic installs of the automation packages. The Deployment Engine information needs to be configured. Go to the Deployment Engine section and change the Host Name to the TPM server name. The remote and local directories will also need to be set.

Now it should be all good to go!

Martin Carnegie

Tivoli Provisioning Manager - Fix Pack 2

I am currently going through the Fix Pack 2 install for TPM (should be the same for TPM for Software) and thought I would make some notes.

The fix became available on June 3rd, but I was too fast on the draw and found out that the files were not uploaded totally. Oops. Then on the 5th, I saw that the files were now quite a bit bigger. Each of the files for the various OSs are about 1.8GB, OUCH!

For the patch information, go to http://www-1.ibm.com/support/docview.wss?uid=swg24016022&loc=en_US&cs=UTF-8〈=all

The install was quite easy and I did not see any errors. One thing to note, make sure you are logged on as the tioadmin account. This is a lesson I learned when installing FP01. I know the docs say this, but I thought that as Administrator (yes only Windows so far) I should be able to do it. Boy was I wrong. At some point the install of TPM changes the security to various files that does not even allow Administrator access.

So what is new?

Endpoint Tasks
This seemed to be something missing from GA and FP01 that was available in TCM. This allows for the creation of a Task much like we saw in TCM. In fact I think that it is much easier. The basic steps are:
1. Create a script and put it in the LocalFileRepository (and a sub-directory if you want)
2. In the TPM web interface, go to Task Management and there is a new section called Endpoint Tasks.
3. Select Edit -> Create Endpoint Task
4. Enter the task name and description
5. Select the files to send from the repository. You can define different files for different operating systems in the same task.
6. Define and parameters for the task
7. View the summary and then press Finish and the task is ready to go!

One thing to note, this only works if SOA is enabled. In my quick search of the docs, I did not see this as a requirement, but the message I saw when I tried to send without SOA enabled sort of pointed to the problem. The message was "COPDEX123E The workflow threw a NotSupportedInDE exception. The message is This workflow is a template workflow for TMF and TCA EndPointTask only."

"Automatic" agent upgrade
In previous versions, upgrading the TCA was a bit of a pain. Usually it was better just to remove the existing agent and reinstall with the new version. This would obviously be a nightmare in any real environment. So now there is a vorkflow called "TCA_Agent_Upgradeable.wkf" that can be wrapped as a Favorite Task that allows for the upgrading of agents. For the couple workstations I tried this on, the upgrade worked. For more information on the process check out Upgrading the common agent" on Info Center

Patching for HP-UX and AIX
I have not been able to try either of these, but I thought I would list them anyway. I know that AIX patching was there before, but according to people I talked to at the TTUG, it did not work. I was told that FP02 is where you have to be to make this work. HP-UX patching is new though and I am sure that people are looking for this.

Java Web Start for the Software Package Editor (SPE)
This is a nice improvement. Now you do not need to have Eclipse installed on your computer to use the SPE.

You do require Java to be install with the minimum level of 1.4.2. Once Java is installed, open your web browser and go to "https://:9045/SPEWebStart/spe.jnlp". This will start the Java Web Start, download the required files and then start the SPE. Once the SPE is started, you will have to go to the Settings -> Preferences to configure the Web Server Name, port and path. You will also have to configure if you are using SSL and default paths to use. Check out Launching and configuring the Software Package Editor on Info Center for more information.

New SOAP commands for Packaging
Some new SOAP commands were created to allow for the creation, distribution, install and uninstall of a software package. There does not seem to be much information in the docs about this yet. The only one I have found is around creating the SPB. I will keep looking and post more later.

Unified Patch Management
You can now manage patches in TCM from TPM and be rid of that "Automation Server" that TCM used. This is a possible scenario that I can see using TPM as a first entry point. I personally think that the patch management facility works fairly well in TPM. The one that came with TCM was a real pain to install and use. So by implementing TPM and importing your TCM environment into TPM you can manage all the patches from TPM and still do all the other stuff from TCM.

Note: This is only for Windows right now.

Defects Fixed
There are around 720 defects that have been fixed! For a complete listing, go to http://www3.software.ibm.com/ibmdl/pub/software/tivoli_support/patches/patches_5.1.0/5.1.0-TIV-TPM-FP0002/5.1.0-TIV-TPM-FP0002.DEFECTS.HTM

Extra Notes
Now that I have FP02 installed, it is time to start working through some of the features in more detail. Hopefully this blog has given you a little help in seeing what is new.

I see that IBM is putting a lot of work in this product and I think they are making progress in many areas.

The one area that still needs work is the installer. Unfortunately, this cannot be addressed in a fix pack. I have heard rumors that there will be a new installer in the next version (5.2??), but these are only rumors.

Some things that I see help improve the success of an install are pre-installing the DB2 and WebSphere products. These seem to be the two trouble areas, especially if you are using machines that are not up to the minimum requirements (like a test lab).

I also think that a fully manual process should be available to install TPM. I have done this and pretty much got everything to work, but it was a fight as I was working in areas that are not even close to documented. All I can say is thank you VMWare for snapshots :)

So now you are wondering if TPM is ready for your environment. Well here is the straight answer, yes and well no.

If you are new to the Tivoli environment and you want TPM or TCM, go to TPM now. There is no point in using TCM. There are some issues with TPM, but learning TCM now and then learning TPM later is going to really hurt. Learn TPM now, get the users on it and don't look back.

If you are a current TCM user, then I would definitely have TPM in the test lab right now at least. The biggest problem for current TCM users that I have seen is that there has been so many processes built around TCM (web pages, reports, data sharing/use) that it is going to take a long time to move everything over. Also the whole push vs pull thing is a big difference to TCM people. We have spent years managing the client expectations that when we submit something, it will start almost right away. Now with TPM, this is no longer true. There is some configurations that can be done to decrease the waiting (polling) but this can severely impact the performance of the product. So there is going to be quite a bit of work to change these expectations, unless something is done to allow for a "push now" option.

Ok this is enough for now, my fingers are bleeding from all the typing. I will keep you posted on anything I find new and interesting. If you have any questions, feel free to comment or send me an email (martin dot carnegie at gulfsoft dot com)

PERL Postemsg Script and Module

This has been around a long time, enjoy.

PERL Postemsg and Module...