Re: Problem with Ubuntu 22.04 failing to restart from NVME SSD, needs a power cycle

[Glenn Washburn]

On Wed, 24 Aug 2022 10:42:37 +0100
Chris Green <cl@isbd.net> wrote:

> On Tue, Aug 23, 2022 at 09:58:42PM -0500, Glenn Washburn wrote:
> > On Tue, 23 Aug 2022 13:44:27 +0100
> > Chris Green <cl@isbd.net> wrote:
> > 
> > > On Tue, Aug 23, 2022 at 01:39:09AM -0500, Glenn Washburn wrote:
> > > > Hi Chris,
> > > > 
> > > > On Mon, 1 Aug 2022 14:41:14 +0100
> > > > Chris Green <cl@isbd.net> wrote:
> > > > 
> > > > > A year or two ago I got a lot of help from this list when I installed
> > > > > an NVME SSD on my system and, because it wasn't recognised by the BIOS
> > > > > I had to configure a slightly odd boot sequence to get it to work.
> > > > > 
> > > > > It has been working beautifully through two or three versions of
> > > > > [x]ubuntu until I recently upgraded from 21.10 to 22.04.
> > > > > 
> > > > > The system runs OK but it can't be restarted, I have to power down and
> > > > > power up again, then it boots OK.
> > > > > 
> > > > > If I wait for a while after a restart I see:-
> > > > > 
> > > > >     Gave up waiting for suspend/resume device
> > > > >     Common problems:
> > > > >        -Boot arge (cat /proc/cmdline)
> > > > >           - check rootdelay= (did the system wait long enough?)
> > > > >           - missing modules (cat /proc/modules; ls /dev)
> > > > > 
> > > > >     ALERT: UUID=2d6ff70c-1894-4afa-a35c-5abb40549d3b does not exist.
> > > > >     Dropping to a shell!
> > > > 
> > > > This is output most likely from your initrd, and almost certainly not
> > > > from GRUB. I'm wondering what kind of reboot you're doing. Are you sure
> > > > if a soft boot? Do you see any output from GRUB? It occurs to me that
> > > > you might be doing a kexec reboot, in which case GRUB is not even in
> > > > the picture. Regardless, you're either skipping GRUB or GRUB is working
> > > > enough to hand off to the Linux kernel. So this doesn't seem to be a
> > > > GRUB issue.
> > > > 
> > > Thanks for coming back to me, even though this looks like it isn't a
> > > GRUB issue.
> > > 
> > > Yes, I do see GRUB talking to me before the above message, I get the
> > > usual list of things that I can boot with older kernels, memtest, etc.
> > > 
> > > So it looks as if, for some reason, on restart/reboot initrd can't see
> > > the NVME drive. I.e. the kernel hasn't reset/restarted something
> > > necessary.
> > 
> > The fact that you see GRUB says to me that the firmware sees the drive
> > fine, assuming that GRUB is on the NVME drive.
> > 
> No, GRUB isn't on the NVME drive.  That's why this is an unusual
> problem. 
> 
> The relevant bits of /etc/fstab are:-
>     #
>     #
>     # / (/dev/nvme0n1p2) root filesystem, just contains OS
>     #
>     UUID=2d6ff70c-1894-4afa-a35c-5abb40549d3b       /       ext4  
> errors=remount-ro 0       1
>     #
>     #
>     # /boot (/dev/sda1 SATA SSD)
>     #
>     UUID=9369df95-29a8-4cbd-adf0-80dfd765f1ba       /boot   ext4    defaults  
>       0       2
>     #
>     #
>     # swap (/dbev/nvme0n1p1)
>     #
>     UUID=88e3aeed-1d01-4523-a209-b3b4cc027dc7       none    swap    sw        
>       0       0
>     #
>     #
>     # /home (/dev/nvme0n1p3)
>     #
>     UUID=7e8a314e-7eea-4f43-b8ed-9db025e41cbf       /home   ext4    relatime  
>       0       2
> 
> I.e. GRUB is on /dev/sda1 which the BIOS can see, the BIOS is too old to 
> recognise the NVME 
> drive.
> 
> 
> > I assume you're getting a linux shell right after "Dropping to a
> > shell!". In the shell do "lsmod" or "cat /proc/modules" and see if the
> > expected nvme modules are loaded. Look at dmesg for anything strange.
> > Most initrds don't have a decent way to scroll lots of text, so you
> > might try to save dmesg and module listing to a USB stick (if that
> > shows up).
> > 
> > You might try rebooting into a usb live linux distro and verify that
> > the drive doesn't show up and then debug from that environment which
> > will have a lot more tools available.
> > 
> The thing is that this all worked perfectly in several previous versions of 
> [x]ubuntu, 
> back to before version 20.04, it's only on upgrading to 22.04 that I have 
> this problem.

Ok, I believe that. Still I don't see how this has anything to do with
GRUB, which is working, and according to you apparently enough to load
the kernel and initrd. Sounds like this is an issue with Ubuntu 22.04's
initrd. Perhaps its not including a necessary kernel module now (even
if before it was). Or it could be some regression in the linux kernel
that affects your NVME drive. Hard to say until you start diagnosing it
by taking steps like I suggested above. 

Glenn