14 Feb
2001
14 Feb
'01
2:20 a.m.
Hi,
This was from a private email between Joe and myself - but as he says,
there's plenty of disk / Unix experts on the list that might be able to
answer this one!
We were talking about the possibility of doing a raw backup of the failing
disk in my Tek system and restoring it onto another disk to get it to
work...
What would be ideal would be
if everything's just a linear bunch of blocks
as far as dd and inode tables etc. are concerned - nice theory that I could
use a different drive to replace the failing one, but I'm sure this wouldn't
work! Anyone?
The other thing I mentioned to Joe was the filesystem check - I remember at
one point I could boot the Tek to single-user mode (throwing up a lot of
filesystem errors along the way) and then run fsck - that'd give a lot of
faults, but not correct them; any ideas how to do this? I'm fairly sure the
system wouldn't even go to single-user last time I tried though.
I can
get a 400MB disk dump from the drive by using dd on my linux system - but
is
that any use? Assuming I could even find an
identical drive as a
replacement, even if I did something like catting the raw image to the
new
disk, I'm not sure if that would give me
something that would work (I'd
still have the filesystem corruption too, but that's a separate problem)
Do disks as read by dd appear as a bunch of linear sectors, regardless of
drive parameters (heads, tracks etc.) - if that's the case then maybe I
have
no problem; maybe I can even use any old drive
providing it's the same
size
or larger than the original disk? OK, I won't
be using the full capacity
of
the drive, but I don't really care :) (I have
about 6 SCSI drives
sitting
unused because I don't have space to put them
in my main PC)
I'm just thinking about inode tables and stuff and how they will probably
get screwed up trying to do any raw copying...
From memory I managed to read all but a single block of
data from the drive
in a linux system a couple of years ago (might even still have
that dump
someplace) with a few retries, so if that block wasn't on something
criticial, if I can get the raw data onto another drive somehow and if I can
fix the filesystem I might be able to get a working system again (that's a
few too many ifs though!)
cheers
Jules
--
14 Feb
14 Feb
5:41 a.m.
Julian Richardson skrev:
Hi,
This was from a private email between Joe and myself -
but as he says,
there's plenty of disk / Unix experts on the list that might be able to
answer this one!
We were talking about the possibility of doing a raw
backup of the failing
disk in my Tek system and restoring it onto another disk to get it to
work...
If you don't trust a raw copy, why not just throw in your other (newer) disk
on the SCSI bus, disklabel it, format it, partition it, and do a "copy -R" of
the entire filesystem?
Besides, it's not as though you'd actually destroy your old disk if you made a
copy. Just try the dd route, and if it doesn't work, truy something else. You
really haven't got anything to lose.
--
En ligne avec Thor 2.6.
Kyosuke: Jag heter Kurre, Kurre Carlsson!
Jag: Det heter du inte alls!
4:21 p.m.
At 05:41 PM 2/14/01 +0100, Iggy wrote:
Julian Richardson skrev:
made
a
Hi,
This was from a private email between Joe and
myself - but as he says,
there's plenty of disk / Unix experts on the list that might be able to
answer this one!
We were talking about the possibility of doing a
raw backup of the failing
disk in my Tek system and restoring it onto another disk to get it to
work...
If you don't trust a raw copy, why not just throw in your other (newer) disk
on the SCSI bus, disklabel it, format it, partition it, and do a "copy -R" of
the entire filesystem?
Besides, it's not as though you'd actually destroy your old disk if you copy. Just try the dd route, and if it doesn't
work, truy something else. You
really haven't got anything to lose.
That should work EXCEPT that he can no longer boot the machine bue to
the failed disk drive. He needs to mount the drive in another machine and
duplicate it there. However, it's very unlikely that he'll be able to find
a machine that can read the old drive and correctly interpret the data.
Therefore, he needs to find a way to just read it as raw data and make an
EXACT copy of it on the new drive. The copy will have to be exact since it
will include the i-node tables and all the other things that make up a Unix
file system. That's why I suggested one of the disk duplicating programs
like Ghost.
Joe
15 Feb
15 Feb
12:29 p.m.
Joe skrev:
That should work EXCEPT that he can no longer boot
the machine bue to
the failed disk drive. He needs to mount the drive in another machine and
duplicate it there. However, it's very unlikely that he'll be able to find
a machine that can read the old drive and correctly interpret the data.
Therefore, he needs to find a way to just read it as raw data and make an
EXACT copy of it on the new drive. The copy will have to be exact since it
will include the i-node tables and all the other things that make up a Unix
file system. That's why I suggested one of the disk duplicating programs
like Ghost.
Aha, so that was the missing link. I'm sorry for being so thick.
--
En ligne avec Thor 2.6.
optimus@dec:foo$ make love
make: don't know how to make love. Stop
6:36 p.m.
At 12:29 AM 2/16/01 +0100, you wrote:
Joe skrev:
No promlem. At least you were trying to help.
Joe
That should work EXCEPT that he can no longer
boot the machine bue to
the failed disk drive. He needs to mount the drive in another machine and
duplicate it there. However, it's very unlikely that he'll be able to find
a machine that can read the old drive and correctly interpret the data.
Therefore, he needs to find a way to just read it as raw data and make an
EXACT copy of it on the new drive. The copy will have to be exact since it
will include the i-node tables and all the other things that make up a Unix
file system. That's why I suggested one of the disk duplicating programs
like Ghost.
Aha, so that was the missing link. I'm sorry for being so thick.
8559
days inactive
8560
days old
4 comments
3 participants
participants (3)
-
JRichardson＠softwright.co.uk -
optimus＠canit.se -
rigdonj＠intellistar.net