[Info-vax] RAID vs. MOUNT/BIND

[Hein RMS van den Heuvel]

On Nov 17, 7:26 am, hel... at astro.multiCLOTHESvax.de (Phillip Helbig---
undress to reply) wrote:
> In article
> <766c6ea4-cc49-40d4-9548-6bda664a5... at v19g2000yqa.googlegroups.com>, H
>
> Vlems <hvl... at freenet.de> writes:
> > first, I have *very* painful memories with respect to restoring bound volumes.

I remember rumblings about that, but never experienced details hands
on.
Recovering from stripe-sets gone bad has got to be more difficult than
dealing with mis-behaving bind sets as fundamentally they are still
basic ODS disks underneath.
Just some magic with file headers/placement/directories.

> (i.e. back up the bound set and restore to a single disk).

I tried just now, using LD devices, the only gotcha is that you can
not restore/image to a drive (mounted foreign) You have to restore to
[*...] on a mounted device which probably is just fine.
Of course you will lose specific file-id's in that process, because
file-id's are only unique to the underlying disks, not the bind-set.
See example below
(sorry for the file name repeats, was meant to be unique, but this is
cute also).

> > Another thing is you ought to bind shadowsets, don't shadow bindsets.
> That is my feeling as well; is the latter even possible?
Right, not possible.

Depending on exactly why you need the large storage, you may want to
consider SEARCH LISTS as an alternative.  Instead to defining STUFF =
BIG:[TEST] make that be STUFF = SMALL1:[TEST], SMALL2:[test],SMALL3:
[test]. Files on STUFF: will be found anywhere, and new ones created
on the first entry.
- it 'splits' the directory into multiples which can be faster
- new files get create in a smaller directory which may be faster.
- old directories with lots of entries perhaps, may not need updates
anymore.
In all this has great potential for performance, structure, and
reduced backup efforts if there is a archival aspect to the data.
Leave/move the old/stable/readonly files to a non-first spot.
You can put those on a less aggressive backup schedule, or even skip
if really stable.
STUFF = shadowed_ram:[day], shadow:[week], shadow:[month], just_a_disk:
[year], just_a_disk:[archive]

Hein

$ ld create lda10.dsk/size=2000
$ ld create lda11.dsk/size=2000
$ ld create lda12.dsk/size=2000
$ ld create lda13.dsk/size=6000
$ ld connec lda10 lda10.dsk
$ ld connec lda11 lda11.dsk
$ ld connec lda12 lda12.dsk
$ ld connec lda13 lda13.dsk
$ init      lda10 lda10
$ init      lda11 lda11
$ init      lda12 lda12
$ init      lda13 lda13
$ moun /bind=lda_10_11_12 lda10,lda11,lda12 lda10,lda11,lda12
%MOUNT-I-MOUNTED, LDA10 mounted on _EISNER$LDA10:
%MOUNT-I-MOUNTED, LDA11 mounted on _EISNER$LDA11:
%MOUNT-I-MOUNTED, LDA12 mounted on _EISNER$LDA12:
$ show dev disk$lda_10_11_12
EISNER$LDA10:           Mounted alloc        0  LDA10
1968     1   1
EISNER$LDA11:           Mounted alloc        0  LDA11
1969     1   1
EISNER$LDA12:           Mounted alloc        0  LDA12
1969     1   1
$ cre/dir disk$lda_10_11_12:[test]
$ copy/allo=100 nl: disk$lda_10_11_12:[test]10.tmp/vol=3
$ copy/allo=200 nl: disk$lda_10_11_12:[test]10.tmp/vol=2
$ copy/allo=300 nl: disk$lda_10_11_12:[test]10.tmp/vol=1
$ show dev disk$lda_10_11_12
EISNER$LDA10:           Mounted alloc        0  LDA10
1668     1   1
EISNER$LDA11:           Mounted alloc        0  LDA11
1768     1   1
EISNER$LDA12:           Mounted alloc        0  LDA12
1869     1   1
$ dir/file/size disk$lda_10_11_12:[000000]test.dir
TEST.DIR;1           (11,1,2)                      1
$ dir/file/size=all disk$lda_10_11_12:[test]
10.TMP;3             (11,1,1)                      0/300
10.TMP;2             (12,1,2)                      0/200
10.TMP;1             (11,1,3)                      0/100
$ back/image disk$lda_10_11_12: disk$lda_10_11_12.bck/save
$ mount lda13 /for
%MOUNT-I-MOUNTED, LDA13 mounted on _EISNER$LDA13:
$ back/image disk$lda_10_11_12.bck/save  LDA13:
%BACKUP-F-BADSETCNT, incorrect number of devices in output
specification
$ dism lda13
$ mount lda13 lda13
%MOUNT-I-MOUNTED, LDA13 mounted on _EISNER$LDA13:
$ back disk$lda_10_11_12.bck/save  LDA13:[*...]
%BACKUP-E-OPENOUT, error opening LDA13:[000000]SECURITY.SYS;1 as
output
-RMS-E-FEX, file already exists, not superseded
: <3 times total>
$ dir/file/size=all lda13:[*...]
Directory LDA13:[TEST]
10.TMP;3             (18,1,0)                      0/300
10.TMP;2             (19,1,0)                      0/200
10.TMP;1             (20,1,0)                      0/100