[gentoo-user] udev Migration and SCSI

[gentoo-user] udev Migration and SCSI

[Ian Brandt]

Hi,

I'm trying to remotely upgrade my server from
gentoo-sources-2.4.25_pre7-r2 to gentoo-sources-2.6.13-r9, i.e. from
devfs to udev.  My root partition is on a RAID 1 mirror on an Adaptec
2100S.  My existing fstab is below.  It was summarized to me by the
NOC over the phone, so I don't have the exact text of the startup
error, but it is something to the effect of...

/dev/scsi/host0/bus0/target0/lun0/part3 not found

I'm guessing that udev is tweaking the naming scheme a bit, as warned
under "No Consistent Naming between DevFS and udev" in the Gentoo udev
Guide.

So two questions:

1) How can I tell what the new name is going to be?

2) As I'm doing this upgrade remotely, how can I set up to fail back
to my udev-less 2.4.25 kernel should 2.6.13 still fail to come up?  In
other words, if I change fstab to be udev specific won't that leave me
dead in the water?

Thanks!

Ian

# /etc/fstab: static file system information.
# $Header: /home/cvsroot/gentoo-src/rc-scripts/etc/fstab,v 1.14
2003/10/13 20:03:38 azarah Exp $
#
# noatime turns off atimes for increased performance (atimes normally
aren't
# needed; notail increases performance of ReiserFS (at the expense of
storage
# efficiency).  It's safe to drop the noatime options if you want and to
# switch between notail and tail freely.

# <fs>                  <mountpoint>    <type>          <opts>
          <dump/pass>

# NOTE: If your BOOT partition is ReiserFS, add the notail option to opts.
/dev/scsi/host0/bus0/target0/lun0/part1         /boot           ext3
          noauto,noatime          1 2
/dev/scsi/host0/bus0/target0/lun0/part3         /
reiserfs        noatime                 0 1
/dev/scsi/host0/bus0/target0/lun0/part2         none            swap
          sw                      0 0
/dev/cdroms/cdrom0      /mnt/cdrom      iso9660         noauto,ro
          0 0
#/dev/fd0               /mnt/floppy     auto            noauto
          0 0

# NOTE: The next line is critical for boot!
none                    /proc           proc            defaults
          0 0

# glibc 2.2 and above expects tmpfs to be mounted at /dev/shm for
# POSIX shared memory (shm_open, shm_unlink).
# (tmpfs is a dynamically expandable/shrinkable ramdisk, and will
#  use almost no memory if not populated with files)
# Adding the following line to /etc/fstab should take care of this:

none                    /dev/shm        tmpfs           defaults
          0 0

-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev Migration and SCSI

[Mike Williams]

On Wednesday 19 October 2005 17:50, Ian Brandt wrote:
> 1) How can I tell what the new name is going to be?

I'd imagine /dev/sdXY will exist under both udev and devfs, and be the same, 
they certainly always have done for me.

> 2) As I'm doing this upgrade remotely, how can I set up to fail back
> to my udev-less 2.4.25 kernel should 2.6.13 still fail to come up?  In
> other words, if I change fstab to be udev specific won't that leave me
> dead in the water?

fstab doesn't have to take block devices, it can take labels too, you could 
look into labeling your partitions.

-- 
Mike Williams

-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev Migration and SCSI

[Ian Brandt]

Mike Williams wrote:
>
> I'd imagine /dev/sdXY will exist under both udev and devfs, and be the same, 
> they certainly always have done for me.

For whatever reason I couldn't get /dev/sda3 in fstab to work when I
originally installed Gentoo on this box many moons ago, I had to use
/dev/scsi/host0/bus0/target0/lun0/part3.  I never did figure out why.
 An ls yeilds:

# ls -al /dev/sd*
lr-xr-xr-x  1 root root 33 Oct 18 16:15 /dev/sda ->
scsi/host0/bus0/target0/lun0/disc
lr-xr-xr-x  1 root root 34 Oct 18 16:15 /dev/sda1 ->
scsi/host0/bus0/target0/lun0/part1
lr-xr-xr-x  1 root root 34 Oct 18 16:15 /dev/sda2 ->
scsi/host0/bus0/target0/lun0/part2
lr-xr-xr-x  1 root root 34 Oct 18 16:15 /dev/sda3 ->
scsi/host0/bus0/target0/lun0/part3

Is there some reason the symbolic links wouldn't have worked?

My fear is if I change my root in fstab to /dev/sda3 my 2.4 kernel
won't come back up, and at $125/hr I'm really trying to avoid getting
the NOC involved.

Is there any way to determine/test whether the configuration is going
to work under devfs and/or udev without actually rebooting?  After
editing /etc/fstab to /dev/sdaX I tried:

# mount -fav

And all it said was:

mount: none already mounted on /dev/shm
none on /proc type proc (rw)

But is that a true test of whether it is going to work on reboot?

> fstab doesn't have to take block devices, it can take labels too, you could 
> look into labeling your partitions.


I'll look into that, but same question applies, is it possible to
verify my changes before I reboot?

Thanks Again!

Ian
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev Migration and SCSI

[Mike Williams]

On Wednesday 19 October 2005 22:00, Ian Brandt wrote:
> Is there some reason the symbolic links wouldn't have worked?

None that I'm aware of.

> My fear is if I change my root in fstab to /dev/sda3 my 2.4 kernel
> won't come back up, and at $125/hr I'm really trying to avoid getting
> the NOC involved.

Leave both sets of entries, just comment out the full scsi path ones. Make 
sure a monkey could put it back right.

> Is there any way to determine/test whether the configuration is going
> to work under devfs and/or udev without actually rebooting?  After
> editing /etc/fstab to /dev/sdaX I tried:

Kinda, yes.
Add /dev/sdXY entries, but under someother directory, /mnt/gentoo for example.
i.e.

/dev/sda3 /mnt/gentoo auto noatime 0 1
/dev/sda1 /mnt/gentoo/boot auto ro,noatime 0 0
etc, etc

The mount -a, and see what happens.

> But is that a true test of whether it is going to work on reboot?

It's as good as your going to get, without rebooting :)

> I'll look into that, but same question applies, is it possible to
> verify my changes before I reboot?

Just the mnt/gentoo thing.

-- 
Mike Williams

-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev Migration and SCSI

[Ian Brandt]

Mike Williams wrote:
> Kinda, yes.
> Add /dev/sdXY entries, but under someother directory, /mnt/gentoo for example.
> i.e.
> 
> /dev/sda3 /mnt/gentoo auto noatime 0 1
> /dev/sda1 /mnt/gentoo/boot auto ro,noatime 0 0
> etc, etc
> 
> The mount -a, and see what happens.

Great suggestion.  Trying it I got a rather odd result:

# mount -av
mount: /dev/sda3 already mounted on /mnt/gentoo/
mount: none already mounted on /dev/shm
mount: mount point /mnt/gentoo/boot does not exist

My main curiosity is the first one.  If I check there is no /mnt/gentoo...

# ls -al /mnt/
total 4
drwxr-xr-x   7 root root 192 Oct 20 12:20 .
drwxr-xr-x  20 root root 480 Jan  5  2005 ..
drwxr-xr-x  13 root root 344 Nov  5  2002 .init.d
-rw-r--r--   1 root root   0 Oct 22  2004 .keep
drwx------   2 root root  72 Feb 26  2004 cdrom
drwx------   2 root root  72 Feb 26  2004 floppy
drwxr-xr-x   2 root root  48 Jan  8  2004 lfs

So how could /dev/sda3 already be mounted there?

I tried creating a new path, /mnt/boottest and /mnt/boottest/boot, and
I get the same thing:

mount -av
mount: /dev/sda1 already mounted on /mnt/boottest/boot
mount: /dev/sda3 already mounted on /mnt/boottest/

But again if I look under /mnt/boottest and /mnt/boottest/boot they're
both empty?  And attempts to umount fail:

# umount /mnt/boottest/boot
umount: /dev/sda1: not mounted
umount: /dev/sda1: not mounted

umount /mnt/boottest/
umount: /dev/sda3: not mounted
umount: /dev/sda3: not mounted

Also, with this method of test, can I test mounting swap from
/dev/sda2?  In my existing fstab sda2 is mounted to "none".  Does it
make sense to do the following?...

/dev/sda2    /mnt/gentoo/swap    swap    sw    0 0


Thanks,

Ian
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev Migration and SCSI

[John Jolet]

what does "cat /proc/mounts" say?
On Thursday 20 October 2005 11:53, Ian Brandt wrote:
> Mike Williams wrote:
> > Kinda, yes.
> > Add /dev/sdXY entries, but under someother directory, /mnt/gentoo for
> > example. i.e.
> >
> > /dev/sda3 /mnt/gentoo auto noatime 0 1
> > /dev/sda1 /mnt/gentoo/boot auto ro,noatime 0 0
> > etc, etc
> >
> > The mount -a, and see what happens.
>
> Great suggestion.  Trying it I got a rather odd result:
>
> # mount -av
> mount: /dev/sda3 already mounted on /mnt/gentoo/
> mount: none already mounted on /dev/shm
> mount: mount point /mnt/gentoo/boot does not exist
>
> My main curiosity is the first one.  If I check there is no /mnt/gentoo...
>
> # ls -al /mnt/
> total 4
> drwxr-xr-x   7 root root 192 Oct 20 12:20 .
> drwxr-xr-x  20 root root 480 Jan  5  2005 ..
> drwxr-xr-x  13 root root 344 Nov  5  2002 .init.d
> -rw-r--r--   1 root root   0 Oct 22  2004 .keep
> drwx------   2 root root  72 Feb 26  2004 cdrom
> drwx------   2 root root  72 Feb 26  2004 floppy
> drwxr-xr-x   2 root root  48 Jan  8  2004 lfs
>
> So how could /dev/sda3 already be mounted there?
>
> I tried creating a new path, /mnt/boottest and /mnt/boottest/boot, and
> I get the same thing:
>
> mount -av
> mount: /dev/sda1 already mounted on /mnt/boottest/boot
> mount: /dev/sda3 already mounted on /mnt/boottest/
>
> But again if I look under /mnt/boottest and /mnt/boottest/boot they're
> both empty?  And attempts to umount fail:
>
> # umount /mnt/boottest/boot
> umount: /dev/sda1: not mounted
> umount: /dev/sda1: not mounted
>
> umount /mnt/boottest/
> umount: /dev/sda3: not mounted
> umount: /dev/sda3: not mounted
>
> Also, with this method of test, can I test mounting swap from
> /dev/sda2?  In my existing fstab sda2 is mounted to "none".  Does it
> make sense to do the following?...
>
> /dev/sda2    /mnt/gentoo/swap    swap    sw    0 0
>
>
> Thanks,
>
> Ian

-- 
John Jolet
Your On-Demand IT Department
512-762-0729
www.jolet.net
john@jolet.net
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev Migration and SCSI

[Ian Brandt]

John Jolet wrote:
> what does "cat /proc/mounts" say?

# cat /proc/mounts
rootfs / rootfs rw 0 0
/dev/root / reiserfs rw,noatime 0 0
none /dev devfs rw 0 0
proc /proc proc rw 0 0
none /dev/shm tmpfs rw 0 0
/dev/scsi/host0/bus0/target0/lun0/part1 /boot ext3 rw,noatime 0 0
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev Migration and SCSI

[John Jolet]

okay, and does that agree with /etc/mtab?  
On Thursday 20 October 2005 12:05, Ian Brandt wrote:
> John Jolet wrote:
> > what does "cat /proc/mounts" say?
>
> # cat /proc/mounts
> rootfs / rootfs rw 0 0
> /dev/root / reiserfs rw,noatime 0 0
> none /dev devfs rw 0 0
> proc /proc proc rw 0 0
> none /dev/shm tmpfs rw 0 0
> /dev/scsi/host0/bus0/target0/lun0/part1 /boot ext3 rw,noatime 0 0

-- 
John Jolet
Your On-Demand IT Department
512-762-0729
www.jolet.net
john@jolet.net
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev Migration and SCSI

[Ian Brandt]

John Jolet wrote:
> okay, and does that agree with /etc/mtab?  

Not exactly:

# cat /proc/mounts
rootfs / rootfs rw 0 0
/dev/root / reiserfs rw,noatime 0 0
none /dev devfs rw 0 0
proc /proc proc rw 0 0
none /dev/shm tmpfs rw 0 0
/dev/scsi/host0/bus0/target0/lun0/part1 /boot ext3 rw,noatime 0 0

# cat /etc/mtab
/dev/scsi/host0/bus0/target0/lun0/part3 / reiserfs rw,noatime 0 0
none /dev devfs rw 0 0
proc /proc proc rw 0 0
none /dev/shm tmpfs rw 0 0
/dev/scsi/host0/bus0/target0/lun0/part1 /boot ext3 rw,noatime 0 0
/dev/sda3 /mnt/gentoo/ reiserfs rw,noatime 0 0

Also, just fyi:

# ls -al /dev/root
lr-xr-xr-x  1 root root 34 Oct 18 16:15 /dev/root ->
scsi/host0/bus0/target0/lun0/part3
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev Migration and SCSI

[John Jolet]

mtab has the entry that's blocking you....wonder if you can just 
copy /proc/mounts over /etc/mtab......
On Thursday 20 October 2005 12:51, Ian Brandt wrote:
> John Jolet wrote:
> > okay, and does that agree with /etc/mtab?
>
> Not exactly:
>
> # cat /proc/mounts
> rootfs / rootfs rw 0 0
> /dev/root / reiserfs rw,noatime 0 0
> none /dev devfs rw 0 0
> proc /proc proc rw 0 0
> none /dev/shm tmpfs rw 0 0
> /dev/scsi/host0/bus0/target0/lun0/part1 /boot ext3 rw,noatime 0 0
>
> # cat /etc/mtab
> /dev/scsi/host0/bus0/target0/lun0/part3 / reiserfs rw,noatime 0 0
> none /dev devfs rw 0 0
> proc /proc proc rw 0 0
> none /dev/shm tmpfs rw 0 0
> /dev/scsi/host0/bus0/target0/lun0/part1 /boot ext3 rw,noatime 0 0
> /dev/sda3 /mnt/gentoo/ reiserfs rw,noatime 0 0
>
> Also, just fyi:
>
> # ls -al /dev/root
> lr-xr-xr-x  1 root root 34 Oct 18 16:15 /dev/root ->
> scsi/host0/bus0/target0/lun0/part3

-- 
John Jolet
Your On-Demand IT Department
512-762-0729
www.jolet.net
john@jolet.net
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev Migration and SCSI

[Mike Williams]

On Thursday 20 October 2005 18:57, John Jolet wrote:
> mtab has the entry that's blocking you....wonder if you can just
> copy /proc/mounts over /etc/mtab......

Probably safer to just remove the erroneous sda3 line by hand, and is unlikely 
to cause issues.

-- 
Mike Williams
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev Migration and SCSI

[Mike Williams]

On Thursday 20 October 2005 17:53, Ian Brandt wrote:
> Great suggestion.  Trying it I got a rather odd result:
>
> # mount -av
> mount: /dev/sda3 already mounted on /mnt/gentoo/
> mount: none already mounted on /dev/shm
> mount: mount point /mnt/gentoo/boot does not exist
>
> My main curiosity is the first one.  If I check there is no /mnt/gentoo...

Interesting...
A 'cat /proc/mounts' like John suggest would be helpful, before and after 
attempting to mount stuff, also try the mount manually.
mount /dev/sda3 /mnt/somethingthatexistsbutisntbeingused.

> Also, with this method of test, can I test mounting swap from
> /dev/sda2?  In my existing fstab sda2 is mounted to "none".  Does it
> make sense to do the following?...
>
> /dev/sda2    /mnt/gentoo/swap    swap    sw    0 0

No, as swap isn't "mounted", you can easily live without swap for a while 
though, so just comment it out.

-- 
Mike Williams

-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev Migration and SCSI

[Ian Brandt]

Mike Williams wrote:
> Interesting...
> A 'cat /proc/mounts' like John suggest would be helpful, before and after 
> attempting to mount stuff, also try the mount manually.
> mount /dev/sda3 /mnt/somethingthatexistsbutisntbeingused.

The manual mount worked:

# cat /proc/mounts
rootfs / rootfs rw 0 0
/dev/root / reiserfs rw,noatime 0 0
none /dev devfs rw 0 0
proc /proc proc rw 0 0
none /dev/shm tmpfs rw 0 0
/dev/scsi/host0/bus0/target0/lun0/part1 /boot ext3 rw,noatime 0 0

# mount -v /dev/sda3 /mnt/boottest/
mount: you didn't specify a filesystem type for /dev/sda3
       I will try type reiserfs
/dev/sda3 on /mnt/boottest type reiserfs (rw)

# cat /proc/mounts
rootfs / rootfs rw 0 0
/dev/root / reiserfs rw,noatime 0 0
none /dev devfs rw 0 0
proc /proc proc rw 0 0
none /dev/shm tmpfs rw 0 0
/dev/scsi/host0/bus0/target0/lun0/part1 /boot ext3 rw,noatime 0 0
/dev/sda3 /mnt/boottest reiserfs rw,noatime 0 0


> No, as swap isn't "mounted", you can easily live without swap for a while 
> though, so just comment it out.

Ah yes.
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev Migration and SCSI

[Mike Williams]

On Thursday 20 October 2005 18:51, Ian Brandt wrote:
> The manual mount worked:

OK great, I'd change my fstab, and reboot to 2.4.X/devfs now, but I'm known 
for being a little gungho :)

BTW, what path for root do you pass to grub?

-- 
Mike Williams
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev Migration and SCSI

[Ian Brandt]

Mike Williams wrote:
> OK great, I'd change my fstab, and reboot to 2.4.X/devfs now, but I'm known 
> for being a little gungho :)

Well, guess there's not much more I can do.

Supposing it doesn't come up, would a rescue CD be required to fix it?
 I left a copy of the old /etc/fstab as /etc/fstab.bak, and setup a
login with sudo rights to perform the copy, but if the discs don't
come up, will login be possible?  (I remember the NOC saying there was
a prompt, "enter the root password for maintenance", or something like
that after a previous failure to boot the 2.6 kernel where the issue
was the Adaptec RAID and general kernel I2O drivers clashing.)

> 
> BTW, what path for root do you pass to grub?
> 

I use lilo actually because it does everything I need, and it doesn't
care that the discs are on a raid array.  Grub at least used to
require a special driver.  My lilo.conf is below.  So would I need to
change my boot and root to the sdaX scheme as well?

# cat /etc/lilo.conf

boot=/dev/scsi/host0/bus0/target0/lun0/disc
root=/dev/scsi/host0/bus0/target0/lun0/part3

map=/boot/System.map

prompt
timeout=50

lba32
vga=normal

default=2.4.25

image=/boot/bzImage-2.4.25
      label=2.4.25
      read-only

image=/boot/bzImage-2.4.28-gentoo-r9
      label=2.4.28
      read-only

image=/boot/bzImage-2.6.13-gentoo-r3
      label=2.6.13
      read-only
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev Migration and SCSI -- SOLVED

[Ian Brandt]

Thanks to Mike and John for all their help.  I learned a lot.

I was able to change my lilo.conf and fstab from the
/dev/scsi/hostW/busX/targetY/lunZ/* scheme to /dev/sdaX instead, and
my 2.4 kernel came back up no problem.  (I wish I could remeber what
forced me into using the /dev/scsi scheme in the first place, but oh
well.)

With that change 2.6 came up as well.

Thanks Again!

Ian

Ian Brandt wrote:
> 
> Mike Williams wrote:
> 
>>OK great, I'd change my fstab, and reboot to 2.4.X/devfs now, but I'm known 
>>for being a little gungho :)
> 
> 
> Well, guess there's not much more I can do.
> 
> Supposing it doesn't come up, would a rescue CD be required to fix it?
>  I left a copy of the old /etc/fstab as /etc/fstab.bak, and setup a
> login with sudo rights to perform the copy, but if the discs don't
> come up, will login be possible?  (I remember the NOC saying there was
> a prompt, "enter the root password for maintenance", or something like
> that after a previous failure to boot the 2.6 kernel where the issue
> was the Adaptec RAID and general kernel I2O drivers clashing.)
> 
> 
>>BTW, what path for root do you pass to grub?
>>
> 
> 
> I use lilo actually because it does everything I need, and it doesn't
> care that the discs are on a raid array.  Grub at least used to
> require a special driver.  My lilo.conf is below.  So would I need to
> change my boot and root to the sdaX scheme as well?
> 
> # cat /etc/lilo.conf
> 
> boot=/dev/scsi/host0/bus0/target0/lun0/disc
> root=/dev/scsi/host0/bus0/target0/lun0/part3
> 
> map=/boot/System.map
> 
> prompt
> timeout=50
> 
> lba32
> vga=normal
> 
> default=2.4.25
> 
> image=/boot/bzImage-2.4.25
>       label=2.4.25
>       read-only
> 
> image=/boot/bzImage-2.4.28-gentoo-r9
>       label=2.4.28
>       read-only
> 
> image=/boot/bzImage-2.6.13-gentoo-r3
>       label=2.6.13
>       read-only
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev Migration and SCSI -- SOLVED

[Mike Williams]

On Friday 21 October 2005 23:29, Ian Brandt wrote:
> Thanks to Mike and John for all their help.  I learned a lot.

Well, that's what we're here for :)

> I was able to change my lilo.conf and fstab from the
> /dev/scsi/hostW/busX/targetY/lunZ/* scheme to /dev/sdaX instead, and
> my 2.4 kernel came back up no problem.  (I wish I could remeber what
> forced me into using the /dev/scsi scheme in the first place, but oh
> well.)
>
> With that change 2.6 came up as well.

w00t!

-- 
Mike Williams

-- 
gentoo-user@gentoo.org mailing list