Several months ago we installed RedHat 9 with software RAID level 1
configured on all partitions using 2 IDE drives.  It ran fine for several
months.  I'm a newbie to linux, so I probably made some mistakes.  I'm
hoping that someone here can help me recover.

One morning, the redhat box seemed completely hung.  The keyboard and mouse
failed to work.  Attempting to shutdown the GUI or switch to a text session
had no effect.  It responded to a ping, but everything else that I tested
seemed to be inoperative.  Finally hit the reset power button.

When I tried to bring it up I discovered that drive(0) has failed.
Unfortunately, I couldn't bring the server back up.  Evidently, I didn't
prepare grub for the loss of the first drive.

I'm reluctant to take any chances of damaging my data that should still be
on the working drive.  Tried to use PowerQuest Drive Image 2002 to back up
the working drive to my Novell server - or even backup to a spare drive.
Can't seem to get Drive Image to work, even with the "Copy All Sectors"
switch.

How can I backup the working drive?
How can I get the new drive configured to boot and re-mirror with the
working drive?

TIA,
....Gregg Nicholas




-----= Posted via webservertalk.com, Uncensored Usenet News =-----
http://www.webservertalk.com - The #1 Newsgroup Service in the World!
-----==  Over 100,000 Newsgroups - 19 Different Servers! =-----

[ Post a follow-up to this message ]

Gregg Nicholas wrote:

> Several months ago we installed RedHat 9 with software RAID level 1
> configured on all partitions using 2 IDE drives.  It ran fine for several
> months.  I'm a newbie to linux, so I probably made some mistakes.  I'm
> hoping that someone here can help me recover.
>
> One morning, the redhat box seemed completely hung.  The keyboard and
> mouse
> failed to work.  Attempting to shutdown the GUI or switch to a text
> session
> had no effect.  It responded to a ping, but everything else that I tested
> seemed to be inoperative.  Finally hit the reset power button.
>
> When I tried to bring it up I discovered that drive(0) has failed.
> Unfortunately, I couldn't bring the server back up.  Evidently, I didn't
> prepare grub for the loss of the first drive.
>
> I'm reluctant to take any chances of damaging my data that should still be
> on the working drive.

Why not simply go to the full system backup you had the foresight to create
within 24 hours of the failure? I won't ask the other, obvious question.

> Tried to use PowerQuest Drive Image 2002 to back up
> the working drive to my Novell server - or even backup to a spare drive.
> Can't seem to get Drive Image to work, even with the "Copy All Sectors"
> switch.
>
> How can I backup the working drive?

There are several options.

1. Install a minimal Linux on drive 0 (/dev/hda), use that to access the
working drive.

2. Get a copy of Knoppix or another CD-bootable distribution, use that to
access the working drive.

3. Exploit the rescue boot floppy you had the foresight to create, use that
to access the working drive.

4. Exploit the ability to boot from e USB memory stick if this ability is
present on your system, use that to access the working drive.

5. And so forth.

In the future, there is a buch better solution: make regular backups. For
the reason, imagine if the second drive had failed instead of the first.

--
Paul Lutus
http://www.arachnoid.com

[ Post a follow-up to this message ]

Paul,
Thanks for responding.

I now realize that I put too much faith in the mirror process.  My previous
experience has been with hardware RAID-5, and a single drive failure isn't a
big deal in that situation.

Following your suggestions, I booted from Knoppix 3.2 (kernel 2.4.20).
I can see /dev/hdb1, hdb2, hdb3, and hdb6

how do I go about getting those partitions to mirror onto hda and get the PC
operational again?


> In the future, there is a buch better solution: make regular backups. For
> the reason, imagine if the second drive had failed instead of the first.

Please forgive my ignorance, but why would it be worse if the second drive
had failed instead?

....Gregg Nicholas




-----= Posted via webservertalk.com, Uncensored Usenet News =-----
http://www.webservertalk.com - The #1 Newsgroup Service in the World!
-----==  Over 100,000 Newsgroups - 19 Different Servers! =-----

[ Post a follow-up to this message ]

Paul Lutus <nospam@nosite.zzz> wrote:
> Gregg Nicholas wrote: 

That doesn't matter. Boot from somewhere else.
[vbcol=seagreen] 

There is no chance. Boot in readonly mode.
[vbcol=seagreen]
> Why not simply go to the full system backup you had the foresight to creat
e
> within 24 hours of the failure? I won't ask the other, obvious question.

There is no need - he has a mirror. All is OK.

But he should know better than to raid1 root or boot. What is the
point? Data on those partitions should almost never change, so a
realtime mirror is just crazy. I could understand it for maybe /home.

> 1. Install a minimal Linux on drive 0 (/dev/hda), use that to access the
> working drive.

No - just boot the thing. He needs a kernel that has md and raid1
support compiled in it.

Though mind you, he could even just forget about it being a raid
partition and boot it as a normal partition. The raid superblock
is at the end of the partition, that's all. Nobody will know. SO
not even a kenrel with raid upport is needed.

Peter

[ Post a follow-up to this message ]

Hi Peter,

I booted with Knoppix 3.2
I think that it has md and raid1 support because the lsraid command seems to
work.  Perhaps that is faulty reasoning.

How do I go about getting this PC working again?

Here is the output from some commands under Knoppix.
---------------------------------------------------------
root@ttyp0[knoppix]# lsraid -R -d /dev/hdb1

lsraid: Unable to find superblock for "/dev/hda1": Invalid argument

# This raidtab was generated by lsraid version 0.7.0.

# It was created from a query on the following devices:

# /dev/hdb1

# md device [dev 9, 0] /dev/md0 queried offline

# Authoritative device is [dev 3, 65] /dev/hdb1

raiddev /dev/md0

raid-level 1

nr-raid-disks 2

nr-spare-disks 0

persistent-superblock 1

chunk-size 64

device (unknown)

raid-disk 0

device /dev/hdb1

raid-disk 1

root@ttyp1[knoppix]# fdisk /dev/hdb

The number of cylinders for this disk is set to 4865.

There is nothing wrong with that, but this is larger than 1024,

and could in certain setups cause problems with:

1) software that runs at boot time (e.g., old versions of LILO)

2) booting and partitioning software from other OSs

(e.g., DOS FDISK, OS/2 FDISK)

Command (m for help): p

Disk /dev/hdb: 40.0 GB, 40020664320 bytes

255 heads, 63 sectors/track, 4865 cylinders

Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System

/dev/hdb1 * 1 33 265041 fd Linux raid autodetect

/dev/hdb2 34 686 5245222+ fd Linux raid autodetect

/dev/hdb3 687 1339 5245222+ fd Linux raid autodetect

/dev/hdb4 1340 4865 28322595 f Win95 Ext'd (LBA)

/dev/hdb5 1340 1372 265041 82 Linux swap

/dev/hdb6 1373 4865 28057491 fd Linux raid autodetect


> No - just boot the thing. He needs a kernel that has md and raid1
> support compiled in it.
>
> Though mind you, he could even just forget about it being a raid
> partition and boot it as a normal partition. The raid superblock
> is at the end of the partition, that's all. Nobody will know. SO
> not even a kenrel with raid upport is needed.
>
> Peter





-----= Posted via webservertalk.com, Uncensored Usenet News =-----
http://www.webservertalk.com - The #1 Newsgroup Service in the World!
-----==  Over 100,000 Newsgroups - 19 Different Servers! =-----

[ Post a follow-up to this message ]

I'm still a bit confused on how I should proceed in recovering my system.
To recap my predicament, I've got a pc that was running Red Hat 9.  It had 2
IDE drives configured with software RAID-1.  The primary drive died.

I've booted with a Knoppix CD and can see the data on the remaining drive.
I've installed a replacement for the failed drive.

What steps do I need to take in order to get my pc booting from the hard
drive and mirroring again?

TIA,
Gregg




-----= Posted via webservertalk.com, Uncensored Usenet News =-----
http://www.webservertalk.com - The #1 Newsgroup Service in the World!
-----==  Over 100,000 Newsgroups - 19 Different Servers! =-----

[ Post a follow-up to this message ]

Gregg Nicholas wrote:

> I'm still a bit confused on how I should proceed in recovering my system.
> To recap my predicament, I've got a pc that was running Red Hat 9.  It had
> 2
> IDE drives configured with software RAID-1.  The primary drive died.
>
> I've booted with a Knoppix CD and can see the data on the remaining drive.
> I've installed a replacement for the failed drive.
>
> What steps do I need to take in order to get my pc booting from the hard
> drive and mirroring again?

1. Use the Knoppix CD to back up your data from the remaining, good, drive.

2. Reinstall the entire system.

3. Restore the data from your backup.

The above default instructions will work for one with less than years of
experience with Linux. If you have more experience, you will know which
steps you can treat editorially.

For example, an experienced Linux user might choose to install a fresh
distribution on the first of the two drives, boot to that drive, and simply
read the data on the second drive. This seems fairly obvious.

BTW the irony of this situation is rather extreme. By choosing RAID-1, and
because you now con't figure out how to proceed, you in essence made your
system half as reliable as it would have been with a ordinary filesystem.

--
Paul Lutus
http://www.arachnoid.com

[ Post a follow-up to this message ]

"Paul Lutus" <nospam@nosite.zzz> wrote in message
news:10jrlf7cadq2hb0@corp.supernews.com...
> BTW the irony of this situation is rather extreme. By choosing RAID-1, and
> because you now con't figure out how to proceed, you in essence made your
> system half as reliable as it would have been with a ordinary filesystem.

Thanks for the advice.
I'd feel more comfortable if I could get an exact image of the drive before
I start messing around with it.  My copy of PowerQuest Drive Image 2002 just
doesn't know what to do with the mirrored partitions.

If I remember correctly, there is a 'dd' command that I might be able to use
from Knoppix to copy the drive.  I'll look into the man pages.

Yes, I can grasp the irony of making my system half as reliable as it
would've been without RAID-1.  At least I still have the data.  I'll put in
a request for hardware raid-5.  In my experience, it doesn't require as much
technical knowledge.

Again, thanks for helping.
....Gregg




-----= Posted via webservertalk.com, Uncensored Usenet News =-----
http://www.webservertalk.com - The #1 Newsgroup Service in the World!
-----==  Over 100,000 Newsgroups - 19 Different Servers! =-----

[ Post a follow-up to this message ]

Gregg Nicholas wrote:

>
> "Paul Lutus" <nospam@nosite.zzz> wrote in message
> news:10jrlf7cadq2hb0@corp.supernews.com... 
>
> Thanks for the advice.
> I'd feel more comfortable if I could get an exact image of the drive
> before
> I start messing around with it.

That's completely understandable. What backup options do you have? Can you
burn a DVD? Do you have a tape drive? Can you make a backup to a second,
spare hard drive, or to a large image file on such a drive? If none of
these options are open to you, you just have to avoid trying to set up RAID
1 on the drive and its new partner, or do this very carefully.

> My copy of PowerQuest Drive Image 2002
> just doesn't know what to do with the mirrored partitions.

You can always mount the partition using the Knoppix CD and then copy the
data as files and directories to a backup destination, say, on a network.
IOW it doesn't have to be backed up as a partition image, and there are
some reasons not to do it that way.

> If I remember correctly, there is a 'dd' command that I might be able to
> use
> from Knoppix to copy the drive.

Yes. Assuming you know which partition contains the data:

# dd if=(source drive and partition) of=(destination drive partition or
file)

This creates a partition image, a block that is the same size as the
original partition. This is not terribly flexible compared to copying the
actual files and directories to a backup location.

Another choice is the use of "tar", which can create a single, compressed
file out of a directory tree. This file can then be used to recreate the
entire original tree at a new location. This also frees you from reliance
on the specific size of the original partition.

> I'll look into the man pages.
>
> Yes, I can grasp the irony of making my system half as reliable as it
> would've been without RAID-1.  At least I still have the data.  I'll put
> in
> a request for hardware raid-5.  In my experience, it doesn't require as
> much technical knowledge.
>
> Again, thanks for helping.

HTH.

--
Paul Lutus
http://www.arachnoid.com

[ Post a follow-up to this message ]

This project turned out to be much easier than I feared.  Made a variety of
backups, at both the partition level and the data level.

Then, because we had RAID-1 on all the partitions, I simply changed the
drive jumpers so that the secondary drive became primary.  Used knoppix to
modify fstab lines from /dev/md? to the appropriate /dev/hda?.  Then
modified grub.

Everything seems to be fine for now.  Have implemented nightly data backups
and we'll leave RAID alone.  Hopefully we'll rebuild this server within a
year - and implement hardware-based RAID-5 at that time.

....Gregg




-----= Posted via webservertalk.com, Uncensored Usenet News =-----
http://www.webservertalk.com - The #1 Newsgroup Service in the World!
-----==  Over 100,000 Newsgroups - 19 Different Servers! =-----

[ Post a follow-up to this message ]