Bill, Is the command server running on both nodes? If the command server is not running then the QMGR will appear to be down because there is no response to the PCF command.
-----Original Message----- From: Bill Anderson [mailto:[EMAIL PROTECTED] Sent: Monday, October 04, 2004 3:23 PM To: [EMAIL PROTECTED] Subject: Support Pack for AIX HACMP I recently implemented the HACMP support pack for MQ on AIX 5.2 / MQSeries 5.3 CSD07. One of its features is an application monitor that checks a queue manager to see if it is healthy and if not, it switches the resources to a second node. The script it uses is fairly simple. It does a "runmqsc" and pings the queue manager. If the ping fails, it checks to see if the queue manager is in a start up mode, which would explain why the ping failed. If no start up process is detected, it fails over to a stand by machine. Last week, one of the production queue managers began to flop back and forth between two nodes repeatedly. I had to stop HACMP services to stop it from moving around and then brought the queue manager up more or less manually. A review of the HACMP logs showed evidence that on the primary node the queue manager would not start up because processes belonging to it were still running ( from another switch two days before). I found one instance of the endmqm command, and three instances of amqrmppa still running on the primary node. On the secondary node everything seemed fine. So, that leaves me with two questions unanswered. 1. Why would amqrmppa and or endmqm being "hung" processes have stopped a queue manager from starting ? The scripts are fairly robust and never assume the state of a queue manager. They look for processes belonging to a given queue manager and if they exist, it kills them. It is all done in the same order as out lined in the system admin manual on how to stop a queue manager manually. But, amqrmppa and endmqm are not among the process names that are searched for. 2. On the secondary node I see in the MQ logs that the queue manager started up with no errors at all. But yet the application monitor seemed to switch it back to the other node. I can only assume that was because it did not respond to a ping, AND was not in start up mode. If anyone has seen this before, talk to me please! I have re-enabled HACMP, but de-configured the application monitor for now. It is a very "cool tool" and I would like to get it back up and running. Bill Anderson SITA Atlanta, GA Standard Messaging Engineering WebSphere MQ Service Owner 770-303-3503 (office) 404-915-3190 (cell) This e-mail contains information which is SITA - Company Confidential All sita.int addresses have changed to sita.aero [EMAIL PROTECTED] http://www.mconnect.aero/ Instructions for managing your mailing list subscription are provided in the Listserv General Users Guide available at http://www.lsoft.com Archive: http://vm.akh-wien.ac.at/MQSeries.archive Instructions for managing your mailing list subscription are provided in the Listserv General Users Guide available at http://www.lsoft.com Archive: http://vm.akh-wien.ac.at/MQSeries.archive
