On 10/11/20 2:04 PM, Marc Shapiro wrote:
On 10/11/20 6:34 AM, Stefan Monnier wrote:
That did it. I am assuming that the system was just in the process of
changing from the initrc to the actual running system? But how do I
get the
boot sequence to activate the LVs automatically each time before
attempting
to mount the / filesystem?
AFAIK you don't need to do anything special for that.
>From what I can tell, the initramfs (tries to) activate the swap and
root LVs, and then the rest of the boot activates them all.
If that doesn't work for you, you're going to have to dig deeper,
Stefan
The swap LV seems to be getting activated, but not the root LV, so it
keeps trying to mount an LV that is not yet activated and eventually
it times out and drops me into a shell. From there, I can manually
activate all LVs and exit the shell and the boot process continues
successfully.
Marc
After more Googling, I found the following page:
https://serverfault.com/questions/199185/logical-volumes-are-inactive-at-boot-time
The last answer on the page suggests a mismatch in the way the LV is
named in:
/usr/share/initramfs-tools/scripts/local-top/lvm2
and
fstab,
namely
'/dev/vgname/lvname' instead of '/dev/mapper/vgname-lvname'.
Both of those files used '/dev/mapper/vgname-lvname', but /etc/lilo.conf
was using '/dev/vgname/lvname'. I changed /etc/lilo.conf to use the
same format as the other two, re-ran lilo, rebooted, and it worked. I
then did the same thing for my Stretch partitions, and the system booted
into Stretch with the LV mounted on / with no issues.
I had not thought that there was any real difference between the two
formats, and this may well be the only instance where there is, but I'm
going to use the '/dev/mapper/vgname-lvname' format in all locations
from here on out.
Marc
