2

I am wondering wherther or not I need CoreGroups in my WebSphere Application Server. I'll explain:

I have different kinds of servers. However, they are all in the same DefaultCoreFroup. By my understanfing, I can seperate them to different CoreGroups, according to type. Yet, I understand that this seperation can be good if I want High Availability possibilities. I am not using the WebSphere High Availability possibilities.

However, I have noticed that occassionally, on WebSphere 6.1 a server on startup shows messages of DCS stack and that it cannot get in touch with another server in its CoreGroup, and so it is stuck that way, until that other Server is restarted / shutodwn or whatever.

If I seperate the two to different CoreGroup, the problem doens't happen. However, I want to know if this is what they meant to do in WebSphere or is this a bug in WAS v6.1 and may have been resolved in WAS v7.0

Thank you.

user967710
  • 1,815
  • 3
  • 32
  • 58

2 Answers2

0

The recommendation of IBM for core group size is 50-100 processes (depends on WAS version). Consider that process is also a node agent.

If your core groups are bigger that could cause the issues you are seeing because at startup each member tries to contact the others.

If they are smaller the problem is probably different, usually taken ports or problem opening a tcp socket between the processes.

Aviram Segal
  • 10,962
  • 3
  • 39
  • 52
0

Please, be very, very careful when / if touching CoreGroups. If you're a novice user, leave them alone. Just because you think you don't use HA services, that doesn't mean you're not using other WAS services that depend on HA.

If you're running 6.1.0.0, your start up problem is most likely a bug that has already been fixed. If you're running on a fairly recent fix pack, it is most likely a configuration / environment problem. If you're still having the start up problem, I'd advise you to give support a call.

Rick
  • 3,830
  • 1
  • 18
  • 16