On Tue, 1 May 2001, Russell Cattelan wrote:
> "Martin K. Petersen" wrote:
> > >>>>> "Eric" == Eric Sandeen <sandeen@xxxxxxx> writes:
> > >> The installation went fine after loading the update disk. However,
> > >> upon reboot, raidautorun reports "failed to open /dev/md0" which
> > >> obviously leads to a kernel panic (my / is on md0). Suggestions?
> > Eric> Hm... hopefully Martin or Russell can speak to this - I don't
> > Eric> have any raid machines here, and have not ever tested it...
> > Eric> Could this be a devfs thing? Has /dev/md0 moved to /dev/md/0
> > Eric> perhaps? If devfsd hasn't started, the compatibility links
> > Eric> won't be there if / isn't even mounted.
> > That's probably the case, yes.
> Wouldn't this be before / is mounted? so it probably doesn't matter if
> /dev/ is real or devfs.
> How does the kernel find and mount / if it is an md device? the dev_id
> is usually
> part of the lilo boot process but in the md case?
Do you have a initrdimage for inserting the md modules?
AFAIk they were also modular on redhat kernels.
I managed to make the server I admin 100Km away unbootable after
forgetting to compile raid 1 support in :-/