PK29781: WLM ROUTING FAILS AFTER DMGR RESTART | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
![]() |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
![]() APAR status Closed as program error. Error description Description: The deployment manager has a set of reconnect logic that fires off when the rest of the cell remains up while the deployment manager is restarted. This logic was rewritten to be multithreaded with PK16480, as the single threaded version would block WLM routing until it completed. The multi threaded version has a timing window in which reconnections will not happen when a customer has nodes with clusters members from more than one cluster on it.Local fix Problem summary **************************************************************** * USERS AFFECTED: Websphere Application Server Network * * Deployment users restarting their deployment * * manager on versions 5.0.2.16+ and * * 5.1.1.9-5.1.1.11 * **************************************************************** * PROBLEM DESCRIPTION: After the a Deployment Manager restart * * with the rest of the cell running, * * Workload Management (WLM) routing may * * not function. * **************************************************************** * RECOMMENDATION: * **************************************************************** In a clustered topology in WebSphere Application Server Version5, when the deployment manager is restarted, JNDI lookups of EJBs that are deployed to the cluster may fail with the following error: . org.omg.CORBA.NO_IMPLEMENT: No Useable Targets minor code: 1229066304 completed: No at com.ibm.ws.wlm.client.selection.WLMLSDRouter.getNextTarget(WLMLS DRouter.java:214) ... . This occurs because when the deployment manager starts, there is logic that tries to reconnect to all of the nodes and cluster members within the cell. With the additition of PK16480 this logic was made multi-threaded, but internal and customer testing did not reveal at the time an additional bug on this path only found after PK16480 was applied. If the customer has a single node with cluster members from more than one cluster, only one of those clusters will end up reconnected, which can cause the above NO_IMPLEMENT error on some lookups, but not others.Problem conclusion The code was rewritten to ensure that all clusters will be reconnected, regardless of the number of different clusters there may be associated with one node. The fix for this APAR is currently targeted for inclusion into the 5.1.1.12 fixpack. Please refer to the recommended updates page for delivery information: http://www.ibm.com/support/docview.wss?rs=180&uid=swg27004980 No more fixpacks are currently scheduled for version 5.0.2 so if a customer needs this fix they will need to acquire an iFix through Level 2 support.Temporary fix Comments
APAR is sysrouted FROM one or more of the following: APAR is sysrouted TO one or more of the following: Modules/Macros
Publications Referenced
|
Product categories: Software > Application Servers >
Distributed Application & Web Servers > WebSphere Application
Server > General
Operating system(s):
Software version: 10A
Software edition:
Reference #: PK29781
IBM Group: Software Group
Modified date: Aug 28, 2006
(C) Copyright IBM Corporation 2000, 2008. All Rights Reserved.