Problem diagnosis for shared file system

This section provides additional diagnosis and repair procedures to use when there appears to be a problem relating to the z/OS UNIX System Services function for shared file system. The types of problems that this section addresses relate to file system availability on one or more systems in a parallel sysplex environment where the root cause of the problem is probably in shared file system processing, rather than, for example, a hardware failure or configuration problem. The two goals of the procedures described here are:

1. To prevent a sysplex-wide restart by either correcting the problem or limiting the scope of the restart to a single system or a subset of systems
2. To provide enough information about the problem to enable the IBM® Support Center to identify and resolve the root cause of the problem as expediently as possible

This section includes example recovery scenarios for the following problems:

1. One or more file systems are mounted in the shared file system but are not accessible (locally mounted) on all systems in the sysplex.
2. A file system appears to be mounted in the shared file system but is not accessible on any system in the sysplex. The file system cannot be mounted or unmounted from any system.
3. A file system appears to be delayed in an UNMOUNT state.
4. Mounting, unmounting, or quiescing of file systems on one or more systems seems to be hung.
5. File system initialization on a restarting system is delayed indefinitely. The delayed system issues message BPXF076I.
6. For whatever reason, you need to reinitialize the file system on all systems without performing a sysplex-wide IPL. (You can perform this reinitialization without any system outage.)

The diagnostic and repair procedures use the following system commands:

• D OMVS,F displays the file system state on any single system in the sysplex. This command displays file system information from the perspective of the system on which the command runs. In a sysplex environment, the file system state may not be consistent on all systems, which is an unusual condition for an active file system.
• D GRS,C and D GRS,LATCH,C display global resource serialization resource contention. Of particular interest for the shared file system is any latch contention for a latch in the SYS.BPX.A000.FSLIT.FILESYS.LSN latch set.
• MODIFY BPXOINIT,FILESYS=[DISPLAY, DUMP, FIX, RESYNC, REINIT, UNMOUNT, UNMOUNTALL] provides diagnostic information about the shared file system, analyzes and repairs certain problems, unmounts one or all file systems, and reinitializes the shared file system. Use this command with caution, only as suggested in the scenarios or under the direction of an IBM Service representative.