[CentOS] Boot message about "resume device"

[Tom H]

On Sun, Aug 1, 2010 at 8:06 PM, Robert Heller <heller at deepsoft.com> wrote:
> At Sun, 01 Aug 2010 19:10:47 -0400 CentOS mailing list <centos at centos.org> wrote:
>> Tom H wrote:
>> > On Sun, Aug 1, 2010 at 1:28 PM, Edward Diener <eldiener at tropicsoft.com> wrote:
>> >> When I boot CentOS 5.5, I receive the message:
>> >>
>> >> Unable to access resume device ( UUID = some UUID etc. )
>> >>
>> >> How do I find out what actual device to which this UUID refers ? It does
>> >> not appear to be a block device since it does not show when I try 'blkid'.
>> >> To what does "resume device" refer ?
>> >>
>> >> The boot succeeds but I would like to know what this messages means.
>> >
>> > UUID?! "resume" must be set to that UUID in /init in your initrd.
>> > Updating/recreating your initrd should fix this problem.
>>
>> How does one "update/recreate" the initrd image ?
>
> mkinitrd -f /boot/initrd-`uname -r`.img `uname -r`
>>
>> Why would initrd hard-code a partition UUID ? If the UUID changes, which
>> it has in my case when I had to move and reformat the swap partition,
>> then the initrd image is now wrong.
>
> It might also depend on what you have for kernel command line parameters
> and/or what /etc/fstab looks like and/or what your hibernate/resume
> config looks like.  I believe it is possible for these things to use
> more 'symbolic' things (like LABEL= for example [yes, mkswap can label a
> swap partition]).
>
> mkinitrd looks in various places to figure out what the swap partition
> is 'called' and is probably falling back to the UUID as the fallback
> choice.

mkinitrd looks for swap in fstab to get the resume partition and swap
doesn't *_usually_* change, whether defined using LABEL or UUID.