On Feb 22, 2004, Warren Togami <warren(a)togami.com> wrote:
Any suggestions for improving the procedure
Don't convert the first disk's partition types to raid before you
actually make them raid members. Should they happen to contain raid
control blocks, for any reason (such as having been raid members
before) this might prevent the newly-created raid devices from being
started. Yes, this has happened to me in the past :-)
Create the raid devices with `missing /dev/hdc?', instead of the other
way round, such that /dev/hda? is the master copy. This is
particularly important for the boot partition. Should the boot
partition happen to need a resync due to a crash, you'll be
overwriting the GRUB boot info in /dev/hda with that in /dev/hdc.
This may be bad, and leave you with a non-rebootable system. Yes,
this has happened to me in the past :-)
mdadm manual start on RHEL3 failed would be appreciated.
I wouldn't be surprised if the reason for the mdadm failure turns out
to be
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=107988
--
Alexandre Oliva Enjoy Guarana', see
http://www.ic.unicamp.br/~oliva/
Happy GNU Year! oliva(a){lsd.ic.unicamp.br, gnu.org}
Red Hat GCC Developer aoliva(a){redhat.com, gcc.gnu.org}
Free Software Evangelist Professional serial bug killer