What to do when a mainframe IPL stalls out without error messages
When initial program load stalls out on a mainframe logical partition, the culprit is often consoles. Connectivity is one of many potential causes.
I am trying to IPL a new z/OS isolated LPAR. The mainframe IPL starts up OK until the MASTER address is supposed to start running. There, it halts with no messages or wait state codes. I'm using z/OS 1.9 on an IBM z/114 2818 mainframe.
What's causing the IPL on the z/OS LPAR to hang up? How can I fix it?
Problems with initial program load (IPL) on an isolated logical partition (LPAR) almost always have something to do with z/OS consoleswhen the symptoms are like the ones described in this question. If you aren't in a wait state, then the system is actually running.
There could be a physical connectivity issue to a console or a console definition problem with an IBM 2074, Open Systems Adapter-Express Integrated Console Controller (OSA-ICC) or other OEM hardware in the environment.
There are a few other possible causes when the mainframe IPL hangs.
If you are getting Nucleus Initialization Program (NIP) messages on your IPL console, this isn't a console connectivity problem or a problem with the software definition for NIP consoles in your I/O definition files. The issue could lie in parmlib CONSOLxx member specification or the consoles defined within it.
The MVS console could be coming up on an alternate console that you do not have opened when IPL appears to stall. This also indicates that the console is not defined properly in the CONSOLxx parmlib member specified at IPL. Check for any messages during NIP related to console processing.
If no valid or working console exists in your parmlib CONSOLxx member, use the hardware management console (HMC) as a console. After IPL, you'll reach a point where you expect an MVS console. Go to the HMC for the LPAR. Under Daily Tasks, select Operating System Messages. This will bring up the HMC graphical user interface to the OS. From the command box, issue V CN(*),ACTIVATE. This should enable MVS commands from the console.
If all else fails, take a standalone dump and send it to IBM in a service request. If you want to look at the dump and try to diagnose the mainframe's problem, open it in the MVS Interactive Program Control System (IPCS) and analyze the basics. Use the IPCS VERBX MTRACE command to display the master trace table, which shows NIP messages, console messages or wait state messages.
For more information, review the MVS Planning Operations manual, in particular the chapter on initializing the system for z/OS 1.9.
This problem was originally posted on TechTarget's IT Knowledge Exchange, where users can ask and answer questions on mainframes and other technologies.
About the author:
Mark Zelden is an accomplished senior consultant with more than 25 years of experience providing solutions for corporate IT organizations. Mark is ITIL V3 certified and is an industry recognized leader in the OS/390 and z/OS operating environments and has a successful track record as a systems programmer and consultant with expertise in operating systems, JES, networking, system software installation and support, and technical applications support.
