Special Problems During the Restore Cycle

Four special problems can be encountered during a restore cycle. If these problems are not handled effectively, the data area of the hard disk drive could become corrupted. For detailed information about system restoration, see Restoring Files in the Guide to mvBase Files and Accounts.

Wrong Backup Media

The most common type of problem arises when an attempt is made to restore the data area using the wrong backup media; that is, accidentally using an incremental filesave tape (or diskette) when performing a full file-restore, or using a full filesave tape (or diskette) for an incremental file-restore. The system cannot differentiate between the various types of filesave tapes and filesave diskettes, and so the restore will corrupt the data area.

To avoid this problem, always label each filesave tape and diskette with the type of save contained on it as well as the date and time of its generation. Also, be sure to use the proper tapes and diskettes in the proper sequence for the type of restore you are performing. When restoring an incremental filesave diskette, be sure to restore the full filesave diskette first, then restore the incremental filesave diskette.

Changed D-pointers

All File Definition items contain a definition code (D/CODE) in Attribute 1; this code can be D, DX, or DY, and the letter C and/or U can be added to any one of these codes. File synonym definition items (Q-pointers) contain a Q in this attribute.

During a filesave, the system saves the contents of each file having a D-pointer of D, DC, DU, or DCU. If a file has a pointer of DX, DCX, DY or DCY, however, the data file pointed to is not saved; only the Q, D, DC, DY and DCY pointers in the file dictionary are saved. Saving Q-pointers preserves the synonym relationship without having to save files more than once. Saving DY and DCY pointers produces an empty file on the backup medium.

If no activity has taken place in a file, the user may want to change its D-pointer to DX or DY (or DCX or DCY) before a weekly full filesave to reduce the time required to back up the data area. Any file whose D-pointer has been altered in this way is not saved on backup media. To change a D-pointer, use the TCL command CHANGE-FILE.

CAUTION

Be sure to change this pointer back to D after the filesave has been completed!

Problems can occur when D-pointers are changed during an incremental filesave cycle. For example, suppose that the backup procedure consists of weekly full filesaves and daily incremental filesaves and that these diskettes are then used to restore the system. The full file-restore does not load the contents of a file that was DX-ed for the full filesave, but the incremental file-restores restore those groups which were updated during the past week since the DX was changed back to D before the incremental filesave diskettes were made. After the restoration procedure is complete, the restored file contains only the updated groups. The rest of the file will be lost.

To avoid this problem, do not change any D-pointers to DX or DY (or DCX or DCY) before a full filesave if an incremental save cycle is used to back up the data area. If a DX or DY is changed back to D after a full filesave, the account and file which it defines (and all files under it in the file hierarchy) must be backed up using some method other than an INCR-SAVE.

Not Enough Disk Space

Two other problems can occur that prevents the restore processor from completing the restoration cycle. These problems occur when there is not enough disk space. The system either runs out of disk space altogether, or it does not have a large enough block of contiguous available space to create a file during the restoration procedure. These problems can arise during any stage of a restore cycle.

When a restore processor aborts, one of the following four error messages displays to indicate the reason for the abort:

BLOCK OF xxx FRAMES NOT AVAILABLE FOR FILE 'filename'

BLOCK OF xxx FRAMES NOT AVAILABLE FOR BASIC OBJECT 'item '

NO DISK SPACE

NOT ENOUGH DISK SPACE

Then a second message displays. How you recover from the abort depends upon which restore processor was being used.

AGAIN/QUIT/REALLOCATE (A/Q/R)?

If this message displays after the BLOCK OF xxx FRAMES NOT AVAILABLE FOR FILE 'filename ' message, the restore aborted because the processor could not find enough contiguous available space to create the file named in the message. There are three options at this point:

Try to create the file again (A).
Quit the restore and return to TCL (Q).
Try to create the file using a new modulo (R). If you type R, the system prompts for a new modulo with which to create the file.

If the :FILELOAD processor aborts a second time at the same file and displays the A/Q/R? prompt, either delete some files (use another terminal’s process) and type A to attempt the restore again, or type R and select a different modulo.

If either the ACCOUNT-RESTORE or the SEL-RESTORE processor aborts a second time, perform a full filesave and restore before you repeat the ACCOUNT-RESTORE or SEL-RESTORE. When the TIME= prompt displays, abort the COLD-START Proc by typing EXIT. This returns control to TCL so that the next step in the restore cycle can be performed immediately.

AGAIN/QUIT (A/Q)?

If this prompt appears following the BLOCK OF xxx FRAMES NOT AVAILABLE FROM BASIC OBJECT 'item ' message, then a full filesave and restore must be done before you try the restoration again. When the TIME= prompt displays, abort the COLD-START Proc by typing EXIT. This returns control to TCL so that the next step in the restore cycle can be performed immediately.

If this prompt appears following the NO DISK SPACE or NOT ENOUGH DISK SPACE messages, there is no more available disk space. More disk capacity must be added or enough files must be deleted to permit the restore to continue.

Recovery from Destroyed Pointers

If backup media information identifying a file is destroyed, it may not be possible for the restore to create that file and subsequent files in the proper sequence. The following message displays, giving the frame (fff) and hexadecimal displacement (ddd) of the software location at which the error was detected:

ERROR IN DSEGMENT

@fff.ddd

LEVEL (1 - 3) ?

A decision must be made whether to:

Search for and continue with the next account on backup media.
Search for and continue with the next file dictionary on backup media.
Search for and continue with the next data file on backup media.

The positioning of the files on the backup media must be taken into consideration before a decision is made.

NOTE	One of the above three courses of action must be taken; the restoration procedure cannot be aborted without starting over from the beginning.

Often restores resumed in this manner cannot be completed due to other problems on the backup media.