IPL or Shutdown Types and Consequences (PWRDWNSYS, MSD, White Button/DPO, F3, F8)

Resolving The Problem

This document was written to clarify confusion about the effects of different types of IPLs and shutdowns available to IBM iSeries users. Some of these effects have serious consequences that should not be ignored. This information is being provided to provide information on the risks of the various types of IPLs and shutdowns that may be performed. A normal shutdown is accomplished when all processes on the system are able to clean up and terminate in a normal manner. Therefore, the ideal shutdown mechanism relies on each job having a built-in end mechanism that can be user-invoked prior to terminating the system with one of the following functions.

PWRDWNSYS OPTION(*IMMED)

This power down option causes the system to issue ENDJOB *IMMED against all processes in the system, followed by System Termination. An IPL after this type of shutdown may or may not be flagged as abnormal due to the ending status of the jobs that were ended. If an abnormal IPL occurs, the joblogs of all ended jobs should be checked to determine why the IPL was marked abnormal. Another cause for an abnormal IPL could be due to jobs ended abnormally prior to the PWRDWNSYS command being issued.

PWRDWNSYS OPTION(*CNTRLD) DELAY(3600)

This power down option causes the system to issue ENDJOB *CNTRLD against all processes in the system. Any processes that are not ended when the delay time expires receive an ENDJOB *IMMED against them. Once all processes are ended, System Termination is performed. Some systems may not be able to end all jobs in 3600 seconds. The DELAY value should be adjusted for these systems. An IPL after this type of shutdown is least likely to be abnormal. If an abnormal IPL occurs, the joblogs of all ended jobs should be checked to determine why the IPL was marked abnormal. Another cause for an abnormal IPL could be due to jobs ended abnormally prior to the PWRDWNSYS command being issued.

User-Initiated Control Panel Function 22 (F22) Main Storage Dump

Control Panel F22 is used by service to request a main storage dump be initiated. To collect an MSD, the system processors are stopped, and control is transferred to the service processor for the partition (or system if non-LPAR). The contents of the first 240 MB of Main Storage (RAM) are copied from RAM to a reserved area on disk. Once complete, a special type of IPL called a MSD IPL is initiated. This IPL will progress to a special IPL point called the Main Storage Dump Manager screen. On this screen, options allow copying of the Main Storage Dump to an Auxiliary Storage Pool (ASP) or to media (DVDRAM or TAPE). As soon as the Main Storage Dump Manager screen is reached, the system begins special processes to minimize the impact of the system termination by performing a special Storage Management step called Reclaim Main Storage. This process finishes writing changed pages from RAM to disk. Next, Storage Management Recovery takes place so that the dump can be copied to the ASP. Finally, the MSD is Auto-Copied to the ASP. An error is displayed at this point if an Auto Copy already exists in the ASP, allowing the user to manually copy the dump to ASP or to tape.

Delay Power Off causes the System Licensed Internal Code to issue a Machine Interface (MI) boundary request. This type of request allows processes to reach an MI boundary before taking the jobs out of the Machine Process Layer. This type of IPL will be marked abnormal. There is a higher than average chance that IPL recovery steps must be completed for a successful IPL.

Control Panel F3 issues a full system reset. As a result of F3, all processes and tasks in the system are halted. No system cleanup is performed. Damaged objects may result when this option is used on a system that is at or beyond the Storage Management Recovery step of IPL. This type of IPL will be marked abnormal and will incur significant IPL recovery.

Control Panel F8 issues an immediate power shut off. As a result of F8, all processes and tasks in the system are halted. No system cleanup is performed. Damaged objects may result when this option is used on a system that is at or beyond the Storage Management Recovery step of IPL.