HP OpenVMS Systems Documentation

Changes the width of Buffered I/O and Direct I/O fields in a report from 8 to 10 characters.

Format

/WIDE

Description

The /WIDE qualifier corrects a problem that users have had with ACCOUNTING reports: the Buffered I/O and Direct I/O fields were too small and displayed asterisks (*) when numbers exceeded 8 characters.

The /WIDE qualifier changes the widths of the Buffered I/O and Direct I/O fields in reports to 10 characters.

Example

$ ACCOUNTING /PROC=BATCH /TYP=PROC - 
     /REPORT=(RECORDS,PROCESSOR,DIRECT_IO,BUFFER)  - 
     /SUMM=MONTH /SIN=1-JAN /WIDE
 
 



  

    

      

      
 
  
 
  
 
 
       MM    TOTAL    PROCESSOR       DIRECT   BUFFERED 
            RECORDS     TIME             I/O       I/O 
  ------------------------------------------------------ 
       01      2043  19 06:52:40.97  532675222 551986091 
       02      1767   9 00:14:34.00  183290432 420000532 
 
  ------------------------------------------------------ 




      

Without the /WIDE qualifier, the Direct I/O or Buffered I/O fields print ***** if the field overflows. With the /WIDE qualifier, these fields print correctly.

You can use the Analyze/Disk_Structure utility (ANALYZE/DISK_STRUCTURE) in two important ways:

• Use the utility on a regular basis to check disks for inconsistencies and errors, and to recover lost files.
• Use the utility with the /SHADOW qualifier to examine the entire contents of a shadow set or a specified range of blocks in a shadow set.

These two uses are explained in the following sections.

Checking Disks

ANALYZE/DISK_STRUCTURE detects problems on On-Disk Structure (ODS) Levels 1, 2, and 5 Files--11 disks; hardware errors, system errors, or user errors can cause these problems. By using ANALYZE/DISK_STRUCTURE to identify and delete lost files and files marked for deletion, you can reclaim disk space.

ANALYZE/DISK_STRUCTURE performs the verification of a volume or volume set in eight distinct stages. During these stages, ANALYZE/DISK_STRUCTURE collects information used in reporting errors or performing repairs. However, ANALYZE/DISK_STRUCTURE repairs volumes only when you specify the /REPAIR qualifier. For a complete description of each of the eight stages, and an annotated example of an ANALYZE/DISK_STRUCTURE session, see Appendix D.

ANALYZE/DISK_STRUCTURE allocates virtual memory to hold copies of the index file and storage bitmaps. With larger bitmaps introduced in OpenVMS Version 7.2, the virtual memory requirements increase correspondingly. To use this utility on volumes with large bitmaps, you might need to increase your page file quota. On OpenVMS VAX systems, you might also need to increase the system parameter VIRTUALPAGECNT.

Virtual memory size requirements for the bitmaps are in VAX pages (or Alpha and Integrity servers 512-byte pagelets) per block of bitmap. Note that the size of the index file bitmap in blocks is the maximum number of files divided by 4096. For ANALYZE/DISK_STRUCTURE, this requirement is the sum of the following across the entire volume set:

• 3 times all the storage bitmaps plus the largest bitmap in the volume set
• 117 times the index file bitmaps
• An additional 96 times the index file bitmaps if /USAGE was requested
• Approximately 600 pages additional fixed scratch space

Examining Shadow Sets

The ANALYZE/DISK_STRUCTURE/SHADOW command is especially useful if a shadow set was initialized with the INITIALIZE/SHADOW command but without the /ERASE qualifier.

Another use of the ANALYZE/DISK_STRUCTURE/SHADOW command is if an error is logged on a member device, and you do not know whether the error was caused by a disk error or by some other hardware component such as a disk controller or cable. When you use the ANALYZE/DISK_STRUCTURE/SHADOW command, every block of every member is read and compared.

For further details, see the Section 3.1.2 and to the /SHADOW qualifier documentation.

3.1.1 Disk Error Reporting and Repair

You can invoke the Analyze/Disk_Structure utility to operate in any of the following three modes for disks errors:

• Error reporting with no repairs
• Error reporting with repairs
• User-controlled selective repairs

By default, ANALYZE/DISK_STRUCTURE reports errors, but does not make repairs. For example, use the following command to report all errors on device DBA1:

$ ANALYZE/DISK_STRUCTURE DBA1:

When you issue this command, ANALYZE/DISK_STRUCTURE runs through eight stages of data collection, then, by default, prints a list of all errors and lost files to your terminal. One type of problem that ANALYZE/DISK_STRUCTURE locates is an invalid directory backlink; a backlink is a pointer to the directory in which a file resides. If your disk has a file with an invalid directory backlink, ANALYZE/DISK_STRUCTURE displays the following message and the file specification to which the error applies:

%VERIFY-I-BACKLINK, incorrect directory back link [SYSEXE]SYSBOOT.EXE;1 

To instruct ANALYZE/DISK_STRUCTURE to repair the errors that it detects, use the /REPAIR qualifier. For example, the following command reports and repairs all errors on the DBA1 device:

$ ANALYZE/DISK_STRUCTURE DBA1:/REPAIR

To select which errors ANALYZE/DISK_STRUCTURE repairs, use both the /REPAIR and /CONFIRM qualifiers:

$ ANALYZE/DISK_STRUCTURE DBA1:/REPAIR/CONFIRM

When you issue this command, ANALYZE/DISK_STRUCTURE displays a description of each error and prompts you for confirmation before making a repair. For example, the previous command might produce the following messages and prompts:

%VERIFY-I-BACKLINK, incorrect directory back link [SYS0]SYSMAINT.DIR;1

Repair this error? (Y or N): Y

%VERIFY-I-BACKLINK, incorrect directory back link 
[SYSEXE]SYSBOOT.EXE;1]

Repair this error? (Y or N): N

Consider running ANALYZE/DISK_STRUCTURE twice for each volume. First, invoke the utility to report all errors. Evaluate the errors and decide on an appropriate action. Then invoke the utility again with the /REPAIR qualifier to repair all errors, or with the /REPAIR and /CONFIRM qualifiers to repair selected errors.

For message descriptions, use the online Help Message (MSGHLP) utility or see the OpenVMS system messages documentation.

Recovering Lost Files

A lost file is a file that is not linked to a directory. Under normal circumstances, files do not become lost. However, files occasionally become lost because of disk corruption, hardware problems, or user error. For example, in cleaning up files and directories, you might inadvertently delete directories that still point to files. When you delete a directory file (a file with the file type .DIR) without first deleting its subordinate files, the files referred to by that directory become lost files. Though lost, these files remain on the disk and consume space.

When you run ANALYZE/DISK_STRUCTURE specifying the /REPAIR qualifier, the utility places lost files in SYSLOST.DIR.

For example, to report and repair all errors and lost files found on the device DDA0, issue the following command:

$ ANALYZE/DISK_STRUCTURE/REPAIR/CONFIRM DDA0:

If it discovers lost files on your disk, ANALYZE/DISK_STRUCTURE issues messages similar to those that follow:

%VERIFY-W-LOSTHEADER, file (16,1,1) []X.X;1 
        not found in a directory 
%VERIFY-W-LOSTHEADER, file (17,1,1) []Y.Y;1 
        not found in a directory 
%VERIFY-W-LOSTHEADER, file (18,1,1) []Z.Z;1 
        not found in a directory 
%VERIFY-W-LOSTHEADER, file (19,1,1) []X.X;2 
        not found in a directory 
%VERIFY-W-LOSTHEADER, file (20,1,1) []Y.Y;2 
        not found in a directory 
%VERIFY-W-LOSTHEADER, file (21,1,1) []Z.;1 
        not found in a directory 
%VERIFY-W-LOSTHEADER, file (22,1,1) []Z.;2 
        not found in a directory 
%VERIFY-W-LOSTHEADER, file (23,1,1) LOGIN.COM;163 
        not found in a directory 
%VERIFY-W-LOSTHEADER, file (24,1,1) MANYACL.COM;1 
        not found in a directory 

All lost files in this example are automatically moved to SYSLOST.DIR.

Erasing Old Home Blocks

When you initialize a volume, the initialize operation might not erase old home blocks. These are blocks that were created by previous initialize operations. If a volume that has old home blocks is damaged, you may not be able to recover the volume without erasing the blocks.

You can erase old home blocks manually by using the /HOMEBLOCKS qualifier on the ANALYZE/DISK_STRUCTURE command as follows:

$ ANALYZE/DISK_STRUCTURE/REPAIR/HOMEBLOCKS

Note that this operation can take up to 30 minutes to complete.

ANALYZE/DISK_STRUCTURE Output

By default, the Analyze/Disk_Structure utility directs all output to your terminal. If you prefer, you can use the /LIST qualifier to generate a file containing the following information for each file on the disk:

• File identification (FID)
• File name
• Owner
• Errors associated with the file

To generate a disk usage accounting file, use the /USAGE qualifier. The first record of the file, called the identification record, contains a summary of disk and volume characteristics. The identification record is followed by a series of summary records; one summary record is created for each file on the disk. A summary record contains the owner, size, and name of the file.

For more information about the disk usage accounting file, see Appendix E.

3.1.2 Detecting Shadow Set Errors

When you enter the ANALYZE/DISK_STRUCTURE/SHADOW command, the system checks for shadow set discrepancies---to ensure that every block on the disk is identical. A discrepancy is a block that should be the same on all members but is not. For example, when you enter a WRITE command, it might not be written to all the members when the ANALYZE/DISK/SHADOW processes it.

If a discrepancy is found, a clusterwide WRITE lock is taken on the shadow set, and the questionable blocks are reread. Then either one of two actions occurs:

• If a discrepancy is still present on the second read, the system displays the file name on the screen. The system also dumps the data block containing the discrepancy to the screen or to a file if you specify the /OUTPUT qualifier.
• If no discrepancy is found on the second read, the system considers the error to be a transient one (for example, if a WRITE to that disk block was in progress). The system then logs the transient error in the summary displayed on the user's terminal. However, verification that all members contained the same information is considered a success---in other words, the data on the disk is actually the same, although for a brief period it was not.

In addition, this utility might report explainable discrepancies between the shadow set members if either of the following occurs:

• The shadow set has not undergone a full merge since the shadow set was created. This occurs if the shadow set was created using the DCL command INITIALIZE/SHADOW without the /ERASE qualifier and if the disk devices had different contents.
  It is important to be aware that this is not disk corruption. The blocks that are reported as different have not been written to, but they might contain stale data; the blocks reported as inconsistent might even be allocated to a file because there might be unwritten space between the file's end-of-data location and the end of the allocated space.
• A full merge has not occurred since the shadow set was logically expanded after a new member was added. The following example illustrates this problem:
  • Shadow set DSA1: consists of two members:

    $1$DGA20: (18 GB)
    $1$DGA21: (36 GB)

  • A second 36-GB member, $1$DGA22:, is added to the shadow set with a full copy operation.
  • After the copy completes, $1$DGA20: is removed from the shadow set.
  • At this point, if you enter the SET VOLUME/SIZE DSA1: command, the shadow set virtual unit DSA1: increases to 36 GB. Then, ANALYZE/DISK/SHADOW reports discrepancies because only the first 18 GB of the shadow set contents were copied to $1$DGA22:.
  
  The discrepancies reported by ANALYZE/DISK/SHADOW are harmless because the space in question has not yet been written by applications.

You can eliminate inconsistencies by performing a full merge. To initiate a full merge, enter the DCL command SET SHADOW/DEMAND_MERGE DSAxxx. If the devices are served by controllers that support controller-based minimerge (for example, HSJ50s), enter this command while the shadow set is mounted on only one node within the cluster. Otherwise, a minimerge occurs, and the discrepancy might not be resolved. When you add members to a single member shadow set, a full copy operation also ensures that the disk is consistent both within and outside the file system.

3.2 ANALYZE/DISK_STRUCTURE Usage Summary

The Analyze/Disk_Structure utility checks the readability and validity of Files--11 Structure Levels 1, 2, and 5 disk volumes, and reports errors and inconsistencies. You can detect most classes of errors by invoking the utility once and using its defaults.

Format

ANALYZE/DISK_STRUCTURE device-name:[/qualifier]

Parameter

device-name

Specifies the disk volume or volume set to be verified. If you specify a volume set, all volumes of the volume set must be mounted as Files--11 volumes. For information about the Mount utility, see the MOUNT documentation in this manual.

To terminate an ANALYZE/DISK_STRUCTURE session, press Ctrl/C or Ctrl/Y while the utility executes. You cannot resume a session by using the DCL command CONTINUE.

By default, ANALYZE/DISK_STRUCTURE directs all output to your terminal. Use the /USAGE or /LIST qualifiers to direct output to a file.

To repair a disk effectively, you must have read, write, and delete access to all files on the disk. To effectively scan a disk (/NOREPAIR), you must have read access to all files on the disk. You must also have write access to INDEXF.SYS to force the flushing of the caches for this file. You must also have write access to BITMAP.SYS for the same reason: to force the flushing of the caches for this file. (You need write access to QUOTA.SYS only if the volume is running disk quotas.)

For a complete explanation of file access, see the HP OpenVMS Guide to System Security.

If you specify /REPAIR, the utility uses the ACP control lock volume function to prevent creation, deletion, extension, and truncation activity while the volume is being rebuilt. In this way, the volume is prevented from being modified while the operation is in progress.

If you specify /NOREPAIR, the volume is not locked; the utility does not attempt to write to the disk. However, if users perform file operations while you run the utility, you might receive error messages that incorrectly indicate file damage. To avoid this problem, HP recommends that you run ANALYZE/DISK_STRUCTURE when the disk is in a quiescent state.

3.3 ANALYZE/DISK_STRUCTURE Qualifiers

This section describes and provides examples of each ANALYZE/DISK_STRUCTURE qualifier. The following table summarizes the qualifiers:

Determines whether the Analyze/Disk_Structure utility prompts you to confirm each repair. If you respond with Y or YES, the utility performs the repair. Otherwise, the repair is not performed.

Format

/CONFIRM

/NOCONFIRM

Description

You can use the /CONFIRM qualifier only with the /REPAIR qualifier. The default is /NOCONFIRM.

Example

$ ANALYZE/DISK_STRUCTURE DBA0:/REPAIR/CONFIRM
%VERIFY-I-BACKLINK, incorrect directory back link [SYS0]SYSMAINT.DIR;1
Repair this error? (Y or N): Y
%VERIFY-I-BACKLINK, incorrect directory back link [SYSEXE]SYSBOOT.EXE;1
Repair this error? (Y or N): N
      

The command in this example causes the Analyze/Disk_Structure utility to prompt you for confirmation before performing the indicated repair operation.

Erases home blocks from a volume whose home blocks were not deleted during previous initialization operations.

Format

/HOMEBLOCKS

Description

You can use the /HOMEBLOCKS qualifier only with the /REPAIR qualifier. The operation can take 30 minutes to complete.

Example

$ ANALYZE/DISK_STRUCTURE DBA0:/REPAIR/HOMEBLOCKS
      

The command in this example causes the Analyze/Disk_Structure utility to erase home blocks on DBA0.