WebSphere 6.1 problem: waiting for other servers to start

0
votes

I have noticed that our WebSphere is acting weird. It has been doing this since ever, so it is not due to topology changes:

  1. It appears When a server starts, it needs to have a ping to all other active computers in cell.

  2. If there's another server that is still starting, the current server may decide to wait for it until it is finished.

We tried a few different things, for example it appears that putting the different servers in different core groups eliminates the problem.

I think this means that WebSphere expects the servers to be HighAvailable for each other, but they are not part of the same cluster, or have any applicative relationship between them.

It appears that WebSphere wants to create a connection between every two servers in my cell.

IS this a known bug? Is it fixed in later versions of 6.1 and 7?

Thank you.

1
websphere

1 Answers

1
votes

As you mentioned this is High Availaility functionality that is used by WebSphere. Think of this as a heartbeat functionality that members of a "Core group". Members of a core group (by default there is a core group called the Default Core Group" ping each other.

If you haven't set up any Core groups then there is only one core group and all the members will attempt to keep an eye on each other (so that components/functionality can be transitioned to another member of the core group if the current provider fails)

This should not affect the server from starting up or doing it's functionality just because another server is down or starting up.

Have a look at this to get a reasonable idea of what HA Manager and Core groups are and what they are trying to do/achieve

http://www.ibm.com/developerworks/websphere/techjournal/1003_col_kepros/1003_col_kepros.html

HTH

Manglu