Discussion:
Bound volume sets
(too old to reply)
Antonio Carlini
2020-09-06 22:44:49 UTC
Permalink
I'm recovering some old disks, some of which were members of bound volume sets. This is pretty obvious from some of the volume labels, but is also clear from SH DEV/FULL which lists the relative volume number.

What I can't tell is how many volumes are in the bound volume set. If I mount two members and do a BACKUP/IMAGE/VERIFY and it completes silently, then I suspect I'm OK. I have at least one case (so far) where I have two members (rvn1 and rvn2) and BACKUP starts off by complaining that it cannot get to certain files. Specifically I'm getting:

%BACKUP-E-OPENDIR, error opening directory DISK$VOLUME:[DIR.TREE.PATH]
-SYSTEM-W-NOSUCHFILE, no such file

Now that could be something else entirely, but either way, I'd like to know how to determine whether I have all the volume set members.

For extra credit, in at least one case I have only rvn 2 of a volume set and (since the metadata lives on rvn1 I believe) I can't see any data on it. BACKUP/VOLUME might back that up (or it might not, I'm running V7.2 on VAX) but is it worth doing? Is there any way to get that data back (in a meaningful form)?

Thanks

Antonio Carlini
geze...@rlgsc.com
2020-09-07 11:43:04 UTC
Permalink
Post by Antonio Carlini
I'm recovering some old disks, some of which were members of bound volume sets. This is pretty obvious from some of the volume labels, but is also clear from SH DEV/FULL which lists the relative volume number.
%BACKUP-E-OPENDIR, error opening directory DISK$VOLUME:[DIR.TREE.PATH]
-SYSTEM-W-NOSUCHFILE, no such file
Now that could be something else entirely, but either way, I'd like to know how to determine whether I have all the volume set members.
For extra credit, in at least one case I have only rvn 2 of a volume set and (since the metadata lives on rvn1 I believe) I can't see any data on it. BACKUP/VOLUME might back that up (or it might not, I'm running V7.2 on VAX) but is it worth doing? Is there any way to get that data back (in a meaningful form)?
Thanks
Antonio Carlini
Antonio,

If the the volumes of a volume set are from different times, it would be easy to see how the structures could be inconsistent from volume to volume. "No such file" messages would be expected in such cases.

The space allocation algorithm attempts to keep a file within a single volume, if possible. One can recover files from orphaned members of a volume set, but is admittedly an involved process. I have done it in the past, when the first member of a two volume set suffered a head crash. While we had a backup, there was some important work done since the last backup. I created a virgin lead volume (this was on a VAX-11/780, so I created the "odd-couple", using the console RX01 floppy disk as the head volume).

Once I had a volume set of sorts, I used ANALYZE/DISK_STRUCTURE/LOST to identify files which were not in any directory. Using user interviews, I was able to reconstruct most of the relevant directory structure. It was a tedious process, but ultimately successful enough.

Today, I would probably use a logical disks to create a standin for the missing member(s).

- Bob Gezelter, http://www.rlgsc.com
Hein RMS van den Heuvel
2020-09-07 13:32:38 UTC
Permalink
Post by Antonio Carlini
I'm recovering some old disks, some of which were members of bound volume sets. This is pretty obvious from some of the volume labels, but is also clear from SH DEV/FULL which lists the relative volume number.
What I can't tell is how many volumes are in the bound volume set.
The first relative volume contains the volume set list file ([000000]VOLSET.SYS) which will have volume labels and their relative volume numbers.

You might also want to write a script or program to just scan the MFD (first volume [000000]000000.DIR) recursively down for the file-ID using DCL F$PARSE, and F$FILE ("FID"), reading the directories as files, or parsing the text output from something like:
DIREXX/FILE/NOHEAD/WID=FIL=60 xxx:[000000...]*.*.*

And I think you can mount stray volumes from the set and ask for lost files like Bob suggests.

Good luck,
Hein
Antonio Carlini
2020-09-07 14:09:26 UTC
Permalink
Post by Hein RMS van den Heuvel
The first relative volume contains the volume set list file ([000000]VOLSET.SYS) which will have volume labels and their relative volume numbers.
Thanks, DUMP-ing VOLSET.SYS on the current volume set shows only two members (plus the bound volume set name).
Post by Hein RMS van den Heuvel
DIREXX/FILE/NOHEAD/WID=FIL=60 xxx:[000000...]*.*.*
And I think you can mount stray volumes from the set and ask for lost files like Bob suggests.
Thanks to Bob too.

So far I only have one volume set where I'm missing the first volume, so it may come to that for that one volume.


Antonio
Dave Froble
2020-09-07 17:17:53 UTC
Permalink
Post by Antonio Carlini
I'm recovering some old disks, some of which were members of bound volume sets. This is pretty obvious from some of the volume labels, but is also clear from SH DEV/FULL which lists the relative volume number.
%BACKUP-E-OPENDIR, error opening directory DISK$VOLUME:[DIR.TREE.PATH]
-SYSTEM-W-NOSUCHFILE, no such file
Now that could be something else entirely, but either way, I'd like to know how to determine whether I have all the volume set members.
For extra credit, in at least one case I have only rvn 2 of a volume set and (since the metadata lives on rvn1 I believe) I can't see any data on it. BACKUP/VOLUME might back that up (or it might not, I'm running V7.2 on VAX) but is it worth doing? Is there any way to get that data back (in a meaningful form)?
Thanks
Antonio Carlini
Well now you understand why I've never been keen on volume sets.

Nor RAID5 for that matter ....
--
David Froble Tel: 724-529-0450
Dave Froble Enterprises, Inc. E-Mail: ***@tsoft-inc.com
DFE Ultralights, Inc.
170 Grimplin Road
Vanderbilt, PA 15486
Antonio Carlini
2020-09-11 10:12:13 UTC
Permalink
Post by Dave Froble
Well now you understand why I've never been keen on volume sets.
To be fair to the people who put the system together (in the mid 1990s I guess) the largest disk available in the StorageWorks range was the RZ29B. They did partition the data (docs, user areas, build areas etc) but even those got to be large. So they really didn't have much choice.
Post by Dave Froble
Nor RAID5 for that matter ....
StorageWorks did offer various HW RAID solutions. I guess I'm lucky that they didn't have any of that, otherwise I think I'd have almost no chance of getting any data off!

Antonio
Rich Jordan
2020-09-16 16:47:20 UTC
Permalink
Post by Antonio Carlini
Post by Dave Froble
Well now you understand why I've never been keen on volume sets.
To be fair to the people who put the system together (in the mid 1990s I guess) the largest disk available in the StorageWorks range was the RZ29B. They did partition the data (docs, user areas, build areas etc) but even those got to be large. So they really didn't have much choice.
Post by Dave Froble
Nor RAID5 for that matter ....
StorageWorks did offer various HW RAID solutions. I guess I'm lucky that they didn't have any of that, otherwise I think I'd have almost no chance of getting any data off!
Antonio
Long ago we were tasked to upgrade VMS, MANMAN, and a host of layered products on a customer VAX; they gave me the job (had done VMS but never heard of MANMAN). Talked the customer into sending us backup tapes (which we could read) then proceeded to daisy chain enough small SCSI drives into a bound volume set attached to my VS3100-30 because they had 4GB drives and we had a bunch of RZ56 and RZ25/26 drives.

The bound volume sets let me build a replica of their system to do and test every stage of the upgrade successfully. For that reason I'll always have a soft spot for them.
Loading...