- Identify which cluster node is being accessed by a client
- Be able to access a particular cluster node
There are many things which can cause problems that are only apparent on one cluster node, such as OS problems (full disk, hardware trouble, etc.) and application problems (synchronization problems).
1. HowTo: Identify which cluster node is being accessed by a client
The easiest way to identify which cluster node a client is accessing is to use the web-browser’s built-in cookie handling.
You can now find the name of the server by looking at the first few characters of the JSESSIONID cookie. In the example above the name is portal1.
This might help you to get a bit closer to the problem, but to uniquely identify the cluster node being accessed you need the cluster ID. You can find the number by looking the numbers after the ID characters in the JSESSIONID or looking using the value of sapj2ee_* cookie. Here the value is 354385894
2. HowTo: Be able to access a particular cluster node
In order to access a particular cluster node, there are three things which must be done:
- Delete all cookies
- Access the server directly and not through the cluster address
- Add the sapj2ee_* parameter to the request, not as a standard HTTP get parameter, but separated with a ; from the URL.If you have any HTTP get parameters, add them after the clusterId (include the ? before the values)
a.)To delete all the cookies use the Internet Options->Delete Cookies.. functionality
b.)If you only have the cluster address of the portal, add a DNS alias (or just put it in your local host file) for each portal server and use this in the url
c.)Use an URL of the following syntax http://portal1.bouvet.no/irj/servlet/prt/portal/prtroot/com.sap.portal.navigation.portallauncher.default;sapj2ee_*=354385901
If you don’t have the cluster ID, you can among others find it in the config tool.
You can use this information to create a simple dynamic page which provides links to all the cluster nodes as well as functionality for deleting the cookies. But that is left as an exercise to the reader 🙂
Thanks to Sigi for information on how to do the second part of this weblog.