Very Computer

I have been trying to add a second SCSI HD to my current SCSI HD so
that IO might mount its partitions. The both current and second HDs
holds bootable Red Hat operating systems.

I attach the second HD after setting its SCSI ID higher than the
current SCSI ID number. My SCSI adapter sees both disks. The grub
screen reports it is loading the kernel on my current HD, but that's
the same kernel version that's on the second HD, so I don't know which
grub I'm actually seeing.

The boot process reaches the point at which mouting of partitions
starts, and there's a lot of error messages. Since I could not
complete the boot properly, I could not read the log to see what all
the messages are, but toward the end it was something like this:

  chmod failed to get attributes of /var/run/utmp

  /etc/rc.d/rc.sysinit can't find a series or files in /var and its
  subdirectories.

  00a.nocode.ctl: line1: head command not found

  kudzu can't find /usr/sbin/kudzu

  touch can't find /var/lock/subsys to create network file

  starting system logger:

Now there's time out that takes perhaps five menutes, after which the
boot resumes to reach a log in prompt. However, log in fails because
bash can't find /home/user, and it makes itself a user and looks at
the / partition:

        -bash-2.05b$

It seems that I can see the files in /, but other partitions are not
mounted.  

What could explain this?

Am I actually booting the second HD instead of the current HD? How
would I know? If so, how do I prevent that?

Or is the second HD grabbing /dev/sda and pushing my current HD to
/dev/sdb? How would I know? How would I prevent it?

--
      Haines Brown


        www.hartford-hwp.com

On Sat, 26 Apr 2003 20:16:54 GMT, Haines Brown staggered into the Black
Sun and said:

Quote:> I have been trying to add a second SCSI HD to my current SCSI HD so
> that IO might mount its partitions. The both current and second HDs
> holds bootable Red Hat operating systems.

?  Are you trying to do soft-RAID, or did you mean, "I want to attach
another SCSI disk to my system and mount its partitions"?

Quote:> The boot process reaches the point at which mouting of partitions
> starts, and there's a lot of error messages. Since I could not
> complete the boot properly, I could not read the log to see what all
> the messages are, but toward the end it was something like this:
[snip]
> Now there's time out that takes perhaps five menutes, after which the
> boot resumes to reach a log in prompt. However, log in fails because
> bash can't find /home/user

> It seems that I can see the files in /, but other partitions are not
> mounted.  What could explain this?

/etc/fstab is wrong.  Fix it.  IIRC, Redhat writes "filesystem labels"
to the ext or Reiser filesystems it creates, and uses those labels in
fstab instead of device names.  If both these disks have the same or
similar filesystem labels, this might cause a problem.

Quote:> Am I actually booting the second HD instead of the current HD? How
> would I know?

Modify one of the kernels in some way; upgrade to the latest 2.4.21-pre7
patchset or something.  That way you'll know right away which kernel is
being loaded.

Quote:> Or is the second HD grabbing /dev/sda and pushing my current HD to
> /dev/sdb? How would I know?

cat /proc/scsi/scsi , fdisk -l /dev/sd[ab] ?

Quote:> How would I prevent it?

Use devfs, so you can refer to your SCSI devices with names like
/dev/scsi/host0/bus0/target1/lun0/disk , which makes more sense for SCSI
users.  Oops, that's not really an option with Redhat--upgrade to Gentoo
or Mandrake?

--
Matt G|There is no Darkness in Eternity/But only Light too dim for us to see
Brainbench MVP for Linux Admin /
http://www.brainbench.com     /  "He is a rhythmic movement of the
-----------------------------/    penguins, is Tux." --MegaHAL

I had a problem adding a second SCSI HD to my existing bus because
when it came to mount the existing drive's partitions, much was not
were it was supposed to be.

Since I run RedHat 8.0, it was suggested that I replace LABELs in
/etc/fstab with dev interfaces. On the drive, there was only a sda1
for /, sda2 for /boot, and sda3 for swap. The former two had labels,
and I relaced them with / and /boot resp. while running the second
disk by itself.

Unfortunately, the disk was no longer bootable. In fiddling, I only
made things worse, and ended reformatting the disk, and now, of
course, I've no problem at all  (I had nothing on the disk I had to
keep).

Even though I've no longer a problem, I'd still like to know why a
simple replacement of

        LABEL=/       /     ...
        LABEL=/boot   /boot ...
with

        /dev/sda1     /     ...
        /dev/sda2     /boot ...

did not work. Any ideas? It would seem that the only thing to go
wrong would be a swap of the two partitions, but that's not the case,
and I checked carefully for typos.

--
      Haines Brown


        www.hartford-hwp.com