Replacing defective pubset disks

If, during operation or FDDRL saves, pubset disks prove to have partial disk defects, these disks should either be reformatted (VOLIN utility routine, see the “Utility Routines” manual [2]) or replaced.

The description below is not relevant for EMC disk subsystems because the integrated RAID technology means that no disk defects occur on these.

The extent of the disk defects can be determined in various ways:

• by a large defect-garbage file or by reaching the extent limit in this file (DMS060C).

• by a large number of EXC0857 messages when the disk is accessed or when it is saved using FDDRL.

Procedure for replacing disks

• Identify the disks affected by the defect using the following command:/SHOW-FILE-ATTR SELECT=*BY-ATTR(STATUS=*PAR(DEFECT-REPORTED=*YES)). A list of files with defects that have already been detected is then displayed. The defective blocks are also referenced by the extent list in the defect-garbage file.

• If you have logical saves of these files, restore the files with these saves.

• Export the pubset.

• Use FDDRL to copy the disk to a new disk or to save the disk to tape; reformat the disk with VOLIN (FDDRL formats the disk automatically when the MN is specified) and use the save to restore it.
  When the defective disks are read by FDDRL the I/O errors are skipped with “CONTINUE” or “CONTINUE ALL”. The numbers of the defective blocks are output (EXC0857, FBD3106, FBD3108); make a record of these numbers.

• Import the pubset with the new or reformatted disk.

• Using the block numbers you recorded, identify the files affected by the defect. The following command can be used to do this: /SHOW-BLOCK-TO-FILE-ASSIGNMENT

• Restore any files that are now still affected by means of their logical saves or by some other means.