Updated: 11 December 1998 |
OpenVMS System Manager's Manual
Previous | Contents | Index |
You can use the following command procedure to interactively back up a disk to a magnetic tape.
To use the procedure, perform the following steps:
$ ! Command procedure DAILYBACK.COM $ ! $ ! Execute this command procedure interactively $ ! by entering the command @[directory]DAILYBACK $ ! at the DCL prompt. $ ! $ ! The BACKUP command in this procedure contains the $ ! output save-set qualifier /REWIND. Therefore, this $ ! command procedure always initializes the output tape. $ ! $ ON ERROR THEN GOTO FAILURE $ INQUIRE DRIVE "Enter the drive name (without a colon)" $ ALLOCATE 'DRIVE' $ INQUIRE SAVESET_SPEC "Enter the save-set specifier" $ INQUIRE LBL "Enter the tape label" $ INQUIRE EXP "Enter the tape expiration date" $ BACKUP/NOASSIST/RECORD/IGNORE=INTERLOCK/SINCE=BACKUP - [...] 'DRIVE':'SAVESET_SPEC'/REWIND/LABEL='LBL'/TAPE_EXPIRATION='EXP' $ DISMOUNT 'DRIVE' $ EXIT $! $FAILURE: $ WRITE SYS$OUTPUT "---> Backup failed" $ WRITE SYS$OUTPUT "" $ DISMOUNT 'DRIVE' $ EXIT |
Volume shadowing maintains multiple copies of the same data on two or more disk volumes. If you use volume shadowing on your system, you can form a shadow set by uniting individual disk volumes (shadow set members). Volume shadowing duplicates data on each member of the shadow set. Per-disk licensing is available for each disk you will be including in a shadow set. This option is effective in a cluster where you intend to shadow only a small number of disks. However, if you have larger systems with many more disks to shadow, traditional capacity (per-CPU) licenses may be more appropriate.
Limits on the numbers of disks allowed in shadow sets are shown in Table 10-7.
Type of Shadow Set | Sets Supported |
---|---|
Single member | Unlimited sets |
Multimember | Total of 400 disks in two- and three-member sets, or both |
These limits apply per cluster. For example, 400 total disks could be configured into 200 two-member shadow sets or into 133 three-member shadow sets per cluster. If single, two-member, and three-member shadow sets are all present on a single cluster, then a maximum of 400 disks may be contained in the two- and three-member shadow sets.
You can use the firmware implementation of RAID level 1 (shadowing) to create shadow sets using the SCSI (Small Computer Systems Interface) disks attached locally to a single SWXCR-xx controller. The StorageWorks RAID Array 210 Subsystem (SWXCR-EA or SWXCR-EB EISA Backplane RAID controllers) and the StorageWorks PCI Backplane RAID controller (SWXCR-PA or SWXCR-PB) have their own firmware implementations of RAID, levels 0, 1, and 5.
SCSI disks connected to these controllers can also be included in shadow sets created using host-based volume shadowing for OpenVMS. For example, with host-based volume shadowing, you can create a RAID1 shadow set containing two like disks, each of which is attached to a separate SWXCR-xx RAID controller located within a cluster. SCSI disks can be configured as shadow sets when attached to systems running volume shadowing for OpenVMS.
For directly connected SCSI devices that have been powered down or do not answer to polling, the elapsed time before a device is removed from a shadow set approaches one minute. In all other situations, the elapsed time closely approximates the number of seconds specified in the SHADOW_MBR_TMO parameter.
Volume shadowing checks for geometries and maximum logical block numbers (LBNs) on devices. This enables devices such as the RZ28 and the RZ28B to operate in the same shadow set. Even though their device IDs differ, their geometries and maximum LBNs will match when configured on like controllers (two HSJ controllers, for example).
When you create a shadow set, individual users access it as a virtual unit. For example, you could create a virtual unit DSA1 that consists of the disks named DUA1:, DUA2:, and DUA3. Users cannot access the individual shadow set members directly, but can perform operations on the virtual unit (DSA1:).
Because of the way volume shadowing duplicates data on each disk in the shadow set, there are special considerations for backing up a shadow set. One strategy for backing up shadow sets involves using the OpenVMS Backup utility.
Do not attempt to back up a shadow set by dismounting an individual shadow set member or by backing up an active shadow set member. You must dismount the entire shadow set and re-create it less one shadow set member. If you do not follow this restriction, the resultant backup copy may contain inconsistent data. |
The proper procedure for using BACKUP to back up a shadow set is described in detail in the Volume Shadowing for OpenVMS manual, and can be summarized as follows.
You cannot perform an incremental backup using this procedure because the backup record date is overwritten when you add the disk volume back into the existing shadow set. |
To mount a disk in the StorageWorks RAID Array 110 Subsystem in a host-based shadow set, you must use the /OVERRIDE=NO_FORCED_ERROR qualifier with the MOUNT command.
The StorageWorks RAID Array 110 Subsystem does not support the
READ/WRITE LONG SCSI commands that are necessary for implementing the
FORCED ERROR function in SCSI. Without FORCED ERROR, you must override
that check by the shadowing driver.
10.15.8.2 Assisted Merging in Mixed-Architecture Clusters
Assisted merging, also known as minimerge, is disabled if shadow sets are mounted on an OpenVMS Alpha node and also on other types of nodes in the same cluster. To reenable assisted merging, apply the CSCPAT (TIMA) kit to all OpenVMS Cluster nodes mounting the shadow set.
With minimerge disabled, shadowing will continue to function normally.
However, a full merge will always be done when a merge operation is
required. A full merge takes considerably longer to complete than a
minimerge operation; Compaq recommends that you install the CSCPAT
(TIMA) kit.
10.16 Restoring User Disks
Occasionally you may want to restore the backup copy of an entire disk. For example, if the disk drive fails, you could restore the backup copy to a working disk. By occasionally saving and restoring an image backup, you can also prevent disk fragmentation.
The way in which you restore a disk depends on whether the most recent
backup was an image (full) or incremental backup. Section 10.16.1
describes the process for restoring a disk when the most recent backup
was an image backup. Section 10.16.2 describes the process for restoring
a disk when one or more incremental backups were performed since the
most recent image backup.
10.16.1 Restoring Image Backups
This section describes how to restore the entire contents of a disk when your most recent backup was an image backup (using the /IMAGE qualifier, as described in Section 10.15.2).
To restore an image backup, use the following procedure.
When you use the /IMAGE qualifier in a restore operation, the disk to which you are restoring the files is initialized. Initializing the disk removes links to the existing files, effectively erasing them. To restore individual files or directories rather than the entire disk, see Section 10.14. |
$ DIRECTORY BACKUP_DISK:[BACKUPS] Directory SYS$SYSDEVICE:[BACKUPS] 19APRIL1998.SAV;1 Total of 1 file. |
$ BACKUP/LIST/REWIND MIA1: Listing of save set(s) Save set: 19APRIL1998.SAV Written by: SYSTEM UIC: [000001,000004] Date: 19-APR-1998 22:03:03.63 . . . |
BACKUP/IMAGE device:save-set-specifier [/SAVE_SET] output-device |
If your backup save set is on a disk or diskette, then you must
also use the /SAVE_SET qualifier immediately after the save-set
specifier (device:save-set-specifier).
The next example shows how to restore an image backup, using the following assumptions:
$ MOUNT/FOREIGN DRA2:(1) %MOUNT-I-MOUNTED, DISK1 mounted on _DRA2: $ BACKUP/IMAGE MIA1:FULL_BACKUP.SAV/REWIND DRA2:(2) $ DISMOUNT/NOUNLOAD DRA2:(3) |
In this example, the individual command lines perform the following actions:
Restoring files after making an image backup and one or more incremental backups is a two-step process. First, restore the most recent image backup. Then, restore each subsequent incremental backup, starting with the most recent.
For the number of directory structure levels you can access see Section 10.14.1.
To restore incremental backups, use the following procedure (note that the first few steps are similar to the procedure for restoring an image backup):
$ DIRECTORY BACKUP_DISK:[BACKUPS] Directory SYS$SYSDEVICE:[BACKUPS] 19APRIL1998.SAV;1 Total of 1 file. |
$ BACKUP/LIST/REWIND MIA0: Listing of save set(s) Save set: 19APRIL1998.SAV Written by: SYSTEM UIC: [000001,000004] Date: 19-APR-1998 22:03:03.63 . . . |
BACKUP/IMAGE device:save-set-specifier[/SAVE_SET] output-specifier |
The /IMAGE qualifier indicates that you are restoring an image
backup. If your backup copy is on a disk or diskette, then you must
also use the /SAVE_SET qualifier immediately after the save-set
specifier (device:save-set-specifier).
MOUNT device-name: volume-label |
The parameter device-name is the name of the drive that
holds the volume you want to mount. The parameter volume-label
is the 1- to 6-character alphanumeric identification you assigned to
the volume with the INITIALIZE command.
BACKUP/INCREMENTAL save-set-specifier[/SAVE_SET] device-specifier |
Remember that you must use the /SAVE_SET qualifier after the
save-set specifier if your backup copies are on a disk or diskette.
Continue restoring the incremental backups, from the most recent to
the oldest, until you have processed all of the incremental backups
since the most recent image backup. If the incremental backups are on
more than one tape, diskette, or disk, then you must load each one
successively when prompted by BACKUP.
When you have processed the
oldest incremental backup, the restore operation is complete.
The next example shows the process of restoring an entire disk after a series of incremental backups, using the following elements and assumptions:
$ MOUNT/FOREIGN DUA2:(1) %MOUNT-I-MOUNTED, WORK_B mounted on _DUA2: $ BACKUP/IMAGE DUA3:WORK_BACKUP.SAV/SAVE_SET DUA2:(2) $ DISMOUNT/NOUNLOAD DUA2:(3) $ MOUNT DUA2: WORK_B (4) %MOUNT-I-MOUNTED, WORK_B mounted on _DUA2: $ BACKUP/INCREMENTAL DUA3:WORK_18_JAN.SAV/SAVE_SET DUA2:(5) $ BACKUP/INCREMENTAL DUA3:WORK_17_JAN.SAV/SAVE_SET DUA2:(6) $ BACKUP/INCREMENTAL DUA3:WORK_16_JAN.SAV/SAVE_SET DUA2:(7) |
In this example, the individual command lines perform the following steps:
BACKUP examines the target disk and the save-set contents to determine which save-set entries to ignore and which target disk entries to delete. If BACKUP encounters a privilege error when attempting to delete directories or other files from the target disk, BACKUP attempts to change the protection of the files so they can be deleted.
BACKUP detects modified directory files and will subsequently save the contents of the directory and its subdirectories to allow proper restoration of renamed directories.
Renaming directories is not recommended. Also, changing security information for a directory changes its modification date. Thus, a directory might appear to be "renamed" and its contents included in incremental save sets if the file protection or security information is changed. The addition of renamed directory contents might increase the size of some incremental save sets. |
BACKUP processes the target disk directory structure by directory levels, in alphabetical order. Thus, circumstances can occur that prevent BACKUP from correctly restoring an incremental save set to a target disk. For example, the target disk does not have sufficient space to hold newly "renamed" directories and their contents prior to deleting the original directories and their contents on the target disk.
If incremental restore fails due to insufficient disk space, a possible solution is to apply the incremental save set a second time (before doing anything else). This causes the first incremental restore to continue and delete directories and their contents, making more space available on the target disk. A second solution is to selectively restore files from the save set.
BACKUP attempts to restore alias or synonym file entries in incremental restore operations that do not specify multiple processing of alias or synonym file entries (/NOALIAS). In cases where the alias entry cannot be restored properly, BACKUP issues an error message indicating the alias file entry, its primary file, and a secondary status of the cause of the failure.
If you specify the /LOG qualifier, then BACKUP issues a message upon successful restoration of alias file entries.
If you specify the /VERIFY qualifier, BACKUP attempts alias entry restoration during the verify pass. Otherwise, alias entry restoration is attempted along with the normal file restoration. The reason for this behavior is that BACKUP attempts to restore all primary files before attempting to restore alias entries that will eventually reference those files.
Previous | Next | Contents | Index |
Copyright © Compaq Computer Corporation 1998. All rights reserved. Legal |
6017PRO_048.HTML
|