On 19.08.2012 16:39, Andreas Henriksson wrote: > Hello! > > Talked the other day with mjt on irc about mdadm problems I was > having that seems very much like this issue. > We debugged it and mjt found some issues... He probably knows > much more then what I'm about to write here...
I filed a separate bugreport about this, #685161 . But now when I think of it more, it looks like all 3 the most important bugs against mdadm are due to the same root cause: missing raid levels in the mdadm.conf and initramfs script choking on the missing /proc/mdstat as the result. They are: #685161 mdadm initramfs script panics with modern mdadm being unable to load any modules #644389 suspect non-working mdadm.conf created during install #678262 mdadm: fails to honour INITRDSTART='none' in /etc/default/mdadm (This last one, even if titled differently, has the same bug as the root cause - we check for /proc/mdstat BEFORE checking for INITRDSTART=none). And it needs a bit more rewriting -- I'm preparing a modified initramfs hook+script pair to resolve this issue for good. [] > (On a related note, the panic script function looks for a "panic=X" cmdline > argument and tries to reboot after X seconds of showing the panic message > instead of giving a rescue shell.... THIS IS BROKEN. > The reboot command in initramfs does nothing! Everybody gets rescue > shell access!) Yes, but this is initramfs-tools issue, not mdadm issue for sure :) /mjt > -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
