system disk backup image

[Boegli]

I've 2 tapes of a system disk backup image.
What the command to restore it with only one tape drive ?

mount/for tape_src:
mount/for dest_disk:

Then ?

[Jan van den Ende]

Boegli,

First:
WELCOME to the VMS Forum!

On the restore of a system image Backup:
>>>
I've 2 tapes of a system disk backup image.
<<<

Do you mean, two different backups, each on one tape, or one backup, that did not fit one tape and had to use a continuation tape?

In the latter case, you need to know WHICH tape is the first.

>>>
mount/for tape_src:
mount/for dest_disk:
<<<

So, I assume you have a running system, and an extra disk that is to become the restored system disk.

$ BACKUP :*.* :
for you fill in the real tape unit ID, for the disk drive ID.

Not knowing anything about the original system, nor about the target, I still think it would be wise to begin with INITialising the dest_disk with some opimization. At least specify /CLUSTER as 4 (smallish disk) or 8 or 16 (modern, large disk).
The default value of 3 (or even smaller) is quite sub-optimal for modern drives.
More info:
$ HELP INIT DEVICE
Then, use
$ BACKUP :*.* : /NOINIT


Remember,
$ HELP is your best friend!

hth

Proost.

Have one on me

jpe

[Robert Gezelter]

Boegli,

First, allow me to second Jan's welcome to the ITRC Forum.

WADU, I will disagree with Jan on one point. From your posting, it would likely that little is known about how your system evolved to its present state. For this reason, I would recommend doing a BACKUP/LIST tape:savesetname/SAVE (please use HELP BACKUP for the precise syntax).

I would ensure that the system disk is re-initialized with the same settings as it had before the need for the restore operation. While changing cluster factors might indeed be a good idea, it is more important to get back to a usable system.

Your posting did not specify what version of OpenVMS you are running, nor did it specify which architecture (VAX, Alpha, or IA64). You may likely need to run the standalone version of BACKUP that came with your distribution. Depending on your architecture, operating system version, and media, the precise procedure for running standalone BACKUP differs.

I hope that the above is helpful.

- Bob Gezelter, http://www.rlgsc.com

[Karl Rohwedder]

You normally would:
$ mount/for 'tape'
$ init 'disk'... ! if you want to change some disk parameters
$ mount/for 'disk'
$ back/IMAGE [/NOINIT] 'tape' 'disk'

Pls. note that the /IMAGE is very important when working with system disks, because of the rooted directory structure, which may otheriwse get compromised.

regards Kalle

[Volker Halle]

Boegli,

let me suggest to use:

$ BACKUP/IMAGE tape_src:/SAVE dest_disk:

An image backup will restore the system disk into the same logical state as at the time of the backup operation. When a continuation tape is needed, backup will dismount/unload the first tape and ask you to load the second tape volume.

Volker.

[Boegli]

Thank you for your answers !

Ok, here are more infos.

I have a working environment running OpenVMS 7.3-1, this is my production machine (cluster of 2 Alpha)

I have a second environment, for the tests, only one machine runing OpenVMS 7.3-1.

The system disk of the test machine is dead, I can read it but I've VBN error on it (don't know what it is) :(

I have a backup of the system disk on 2 tapes.

I've mounted a new disk on the production machine and I want to make a new system disk with my 2 tapes.

I use this command :
backup/image/verify $1$MKA400:TANIC_0915.BCK/SAV $4$DKB500:/log

It works well but the system doesn't ask me to mount the second tape and the backup isn't complete...

Here is the end of the log file :

%BACKUP-S-CREATED, created $4$DKB500:[UNIF_REF.PR]UNIFCO.SNP;1
%BACKUP-S-CREATED, created $4$DKB500:[UNIF_REF.PR]UNIFPR.RDB;1
%BACKUP-S-CREATED, created $4$DKB500:[UNIF_REF]TST_FER.DIR;1
%BACKUP-S-CREATED, created $4$DKB500:[UNIF_REF]UNIF_FORMATION.DIR;1
%BACKUP-S-CREATED, created $4$DKB500:[UNIF_REF.UNIF_FORMATION]DB_2K.DIR;1
%BACKUP-I-STARTVERIFY, starting verification pass
%BACKUP-E-POSITERR, error positioning $1$MKA400:[000000]TANIC_0915.BCK;
-SYSTEM-F-CTRLERR, fatal controller error
%BACKUP-I-OPERSPEC
%BACKUP-I-OPERASSIST, operator assistance has been requested

[Volker Halle]

Boegli,

VBN = Virtual Block Number. This is the relative block number within a file. VBN 1 is the first block of the file. In which context do you see 'VBN errors' ? If you can 'read' your 'dead' system disk (in the other machine, I suppose), than it can't be that dead. Can you run $ ANAL/DISK dcx: on that disk ?

Does the positioning error always happen during the verify path ? If so, try to restore the image abckup without using /VERIFY. You can also do a BACKUP/LIST of the second tape as well, to check whether it is readable.

Volker.

[Louis in Michigan]

If you are mounting your own tapes, then use of the "/NOASSIST" option to the BACKUP command can be quite useful. Without it, the BACKUP program makes an OPCOM system request for the continuation tape.

Example:
BACKUP /NOASSIST /IMAGE -
my_backup_tape:saveset.name/SAVE my_new_disk:

[Steve-Thompson]

Hello Boegli

The fact that the error occurs during the VERIFY stage means the restore operation has actually finished and all data from the specified save-set has been transfered to the disk.
Have you tried booting the disk? It should at least boot this will let you look at the on disk structure.

Jan's first is reply very likely ...
the second tape has a second saveset which may well go onto a diffent disk and will require a second backup operation.

As has been suggested, the best way to check this is doing a
$backup/list=listing.txt mk:*.*/sav

... then look at the file listing.txt with an editor.

Regards
Steven

[Boegli]

backup/list

1st tape :

Save set: TANIC_0915.BCK
Written by: SYSTEM
UIC: [000001,000004]
Date: 15-SEP-2003 11:58:55.67
Command: BACKUP/IMAGE DKC0: MKB400:TANIC_0915.BCK/SAV/IGNORE=(LABEL,NOBACKUP)/REWIN
Operating system: OpenVMS Alpha version V7.3
BACKUP version: AXP73-1
CPU ID register: 80000000
Written on: _MKB400:
Block size: 8192
Group size: 10
Buffer count: 723

2nd tape :

Save set: TANIC_0915.BCK, volume 2
Written by: SYSTEM
UIC: [000001,000004]
Date: 15-SEP-2003 11:58:55.67
Command: BACKUP/IMAGE DKC0: MKB400:TANIC_0915.BCK/SAV/IGNORE=(LABEL,NOBACKUP)/REWIN
Operating system: OpenVMS Alpha version V7.3
BACKUP version: AXP73-1
CPU ID register: 80000000
Written on: _MKB400:
Block size: 8192
Group size: 10
Buffer count: 723


I'm now doing a restore using these options... I will see what happens :

$ backup/image/noverify/noassist $1$MKA400:TANIC_0915.BCK/SAV $4$DKB500:/log

[Volker Halle]

Boegli,

yes, the second tape IS a continuation tape (see: Save set: TANIC_0915.BCK, volume 2), so you need to also restore that 2nd tape in ONE backup operation to except to get a bootable system disk.

Volker.

[Boegli]

Thank you all, now the backup is running.
I will show you what is the message after the end of the first tape.

[Steve-Thompson]

Hi Boegli

At least you've been able to read the two tapes.... that gives a certain amount of confidence in the state of the tapes, given they're "three" years old.

As you've changed the failing disk it should work...

Another thing ... if you get another "failure", clean the heads if the tape drive if you havent already done so, before the next attempt!

Steven

[Boegli]

Here is the message after the first tape :

%BACKUP-S-CREATED, created $4$DKB500:[UNIF_REF]UNIF_FORMATION.DIR;1
%BACKUP-S-CREATED, created $4$DKB500:[UNIF_REF.UNIF_FORMATION]DB_2K.DIR;1
%DISM-W-CANNOTDMT, _$1$MKA400 cannot be dismounted
%BACKUP-I-RESUME, resuming operation on volume 2
%BACKUP-I-WRONGVOL, $1$MKA400:[000000]TANIC_0915.BCK; is not the next volume in the set
%DISM-W-CANNOTDMT, _$1$MKA400 cannot be dismounted


What should I do with the second tape ???

[Volker Halle]

Boegli,

there seems to be something wrong here:

%DISM-W-CANNOTDMT, _$1$MKA400 cannot be dismounted
%BACKUP-I-RESUME, resuming operation on volume 2
%BACKUP-I-WRONGVOL, $1$MKA400:[000000]TANIC_0915.BCK; is not the next volume in the set
%DISM-W-CANNOTDMT, _$1$MKA400 cannot be dismounted

The %DISM-W-CANNOTDMT message should be accompanied by a secondardy status message (according to HELP/MESS CANNOTDMT).

Because the first tape does not get dismounted, backup complains with WRONGVOL, because the first tape is still in the drive.

I've just seen this last week (during a backup to tape test) on a V7.3-2 system, but could not reproduce the circumstances.

Try again without /NOASSIST and don't mount/foreign the tape yourself. Make sure there is a terminal enabled as tape operator ($ RELY/ENABLE=TAPE).

Volker.

[Jan van den Ende]

Boegli,

I am beginning to have serious doubts on your tapedrive, it SHOULD dismount at this point.
At least it explains the positioning error :-(
It will cost quite some time again, and result is not guaranteed, but it MIGHT be worth trying without /NOASSIST; on another terminal $ REPLY/ENABLE [=TAPE], REPLY/TO= CONT after the change of tape, but in view of the previous erros, do not set your hopes too high!

Do you have another tapeunit that you can hook up?
Then THAT would be my choice!.

hth

Proost.

Have one on me.

jpe

[Volker Halle]

Boegli,

when I just checked the logfile of last week's backup tests, I found the %DISM-W-CANNOTDMT message in ONE case only, where we left off the initial MOUNT/FOR/NOUNLOAD tape: command in the DCL procedure.

Did you issue the MOUNT/FOR tape_src: command before your last restore ? If not, you may want to make sure you do. It SHOULD also work, if backup does mount the tape itself, but you never know...

Volker.

[Steve-Thompson]

Boegli

This is odd... While thinking about this I see Volke and Jan have come in with their ideas!

They've seen "%DISM-" problems (albeit spurious) during a backup... I've yet to have the "pleasure", so it surprised me somewhat!

Given your backup is a system disk are you booting from a CD? If not you could try keeping the next attempt as simple as possible.

Boot VMS from a CD and work from the console.
(problems caused by /(no)assist etc disappear)

then .. if you don't mount a disk or tape correctly (foreign/files-11) BACKUP will "fail" your command and tell you what it needs, it's very helpful in that respect.

So ...

$$$ sho dev (make sure you can see required devices)

$$$ mou/fore MK400:
$back/imag/rewi mk: (As above)

then the only thing left to do is the mechanics of changing the tape and "hit return" when BACKUP asks for the second cartridge, nothing else.

By all accounts the second tape DOES belong to the saveset, the timestamp of the savsets confirms this as seen in the listing.

I'm sure you will succeed with perseverance!

Jan is (correctly) concerned with the state of the tapedrive. Positioning errors usually indicate ..
a dodgy tape drive (somethings out of spec.)
dirty tape mechanism
dubious condition of the media.

Good luck
Steven

[GuentherF]

>%DISM-W-CANNOTDMT, _$1$MKA400 cannot be dismounted

I would be surprised if this message is from BACKUP. Even though it is usually not necessary I would DCL-ALLOCATE the tape drive first. Let BACKUP do the MOUNT/FOR and use /ASSIST.

Is this a plain tape drive or some kind of a loader/library?

/Guenther

[Boegli]

Thank you all !
With you I will succeed one day :)

My tape drive on the production machine is :
Media Robot Identifier: DEC TL800 (C) DEC0519

I think it works well because the PROD backup is OK every nights.

Here is exactly what I do (I use eXcursion 7.3.189 to access the machine, user SYSTEM) :

$ mount/for $1$MKA400:
$ mount/for $4$DKB500:
$ SUBMIT /LOG=D00:[EXPLOIT.LOGS] /KEEP /NOPRINT /QUEUE=SYS$BATCH_ONE D$USER_1:[OPER_PASCAL]RESTORE_TANIC_SYS.COM

And the contents of RESTORE_TANIC_SYS.COM :

$ backup/image/noverify/noassist $1$MKA400:TANIC_0915.BCK/SAV $4$DKB500:/log
$ exit

I will try without mounting the tape before or I'll boot on CD and use console.

[Volker Halle · Accepted solution]

Boegli,

you MUST add the MOUNT/FOR commands into the batch command file !!! Specifying /NOASSIST in a batch job does NOT achieve anything - it is ignored.

Please change your .COM file to:

$ mount/for $1$MKA400:
$ mount/for $4$DKB500:
$ backup/image/noverify $1$MKA400:TANIC_0915.BCK/SAV $4$DKB500:
$ exit

The /NOASSIST qualifier would direct BACKUP to ask on the terminal (SYS$COMMAND) for a reply after inserting the 2nd tape. A batch job does not have a 'terminal' as SYS$COMMAND and backup will therefore issue an OPCOM Request, which you need to answer with REPLY/TO=n ...

Make sure, that there is at least one terminal enabled as tape operator terminal.


Guenter,

if you believe it or not, I've seen this error once, when testing on a SSL 1016 tape loader last week:

...
17:13:49$ IF F$GETDVI("TAPE_DEV","DEVNAM") .NES. "_NLA0:"
17:13:49$ THEN
17:13:49$!!! MOUNT/FOR/NOUNLOAD TAPE_DEV:
17:13:49$ ENDIF
...
17:13:49$ BACKUP/IMAGE/IGNORE=(INTER,LABEL)/REWIND DISK$BCK_DISK: -
TAPE_DEV:BCK_DISK1.BCK/SAVE/BLOCK=65535/MEDIA=COMPACTION /NOCRC
%BACKUP-I-LBLOVRWRITE, volume label DSA100 overwritten, new label is BCK_DI
%DISM-W-CANNOTDMT, _$1$MKA500 cannot be dismounted
17:19:05$ GOSUB report_perf
...

For all other tests, I had re-activated the MOUNT/FOR command above and those other tests have NOT failed. Someone else 'may' have been 'playing' with the tape as well, but I don't known.

Volker.

[Boegli]

It's OK :)
I've written all the command lines in a DCL window without using a batch job.
Tomorrow I go on site to test the disk, I hope it will boot.
THANK YOU ALL !

[GuentherF]

Volker,

BACKUP is not doing a dismount of a tape unless it needs to continue on to another tape or, the current tape is rejected.

BACKUP does not dismount the tape when done writing the save set.

So I have no clue why you see a dismount in the first place and then why it failed.

/Guenther