I had installed my laptop with Fedora 9 preview. /boot : plain ext3 / : plain ext3 /home : disk encryption on top of software raid 1 /dev/md0 -> /dev/mapper/luks-md0 tmp and swap also live on crypted partitions, using random keys /etc/crypttab: luks-md1 /dev/md1 none luks-md0 /dev/md0 none tmpcrypt /dev/sda8 /dev/urandom tmp swapcrypt /dev/sda7 /dev/urandom swap I have frequently updated my system using yum. During startup, in the past, as expected, the init scripts asked me to enter my password for the encrypted partition. Today, when I started my computer, I no longer get prompted for the password. The init scripts try multiple times (2 or 3) to start disk encryption, but I always get a red [FAILED]. Is there a way for me to trace where exactly it is failing? Or should I add "echo" statements to /etc/rc.sysinit ? Please let me know if you have ideas how I could help to track this down. Note, I have not made changes to my system scripts, nothing I'm aware of (besides adding a hdparm command to /etc/rc.local ). FWIW, I have changed my /etc/fstab and specified "noauto" for /home. Once the system is up, I am able to manually run "cryptsetup luksOpen" and mount /home fine.
note that picking older kernels for boot does not help. My system is up to date, according to yum.
This seems to be a random behavior. I had repeatedly tried, and always got the reported behavior (yesterday). But today, it works as expected!
Is it just /home that has issues?
Both /home and my separate /extra partition (luks-md0 and luks-md1) have the problem. Either both work, or both don't work. / is a plain partition (no raid, no encryption)
Is the RAID itself being set up properly in this case?
Also, what happens if you add in: 12af846675bf3b89693d81e750a4ad86f1393130 d30e73c36dd761e853ceccc968d85daaaf85f619 and b661d6b395f6a84813f81ae8dae2f200a792d2f5 from http://git.fedorahosted.org/git/initscripts.git?
(In reply to comment #5) > Is the RAID itself being set up properly in this case? Yes. When it drops me to the emergency shell, the md arrays have been started. Also, after I added "noauto" to initscripts, as soon as the system allows me to log in, the RAID arrays are there, too. "cryptsetup luksOpen" and "mount" is all I need to do.
(In reply to comment #6) > Also, what happens if you add in: > 12af846675bf3b89693d81e750a4ad86f1393130 > d30e73c36dd761e853ceccc968d85daaaf85f619 > and > b661d6b395f6a84813f81ae8dae2f200a792d2f5 > > from http://git.fedorahosted.org/git/initscripts.git? I've prepared a rc.sysinit file and saved it locally. As soon as the problem shows up again, I'll test your modified rc.sysinit. Question: Should I have been able to paste your hashes into some search field on that web page? That didn't give me results. I happened to find your 3 commits by mousing over some links and comparing.
No, probably not. Sorry about that, probably should have just generated a patch. I'm wondering if it's a timing issue with the RAID startup, but I'm not sure why that would be the case.
I have never again ran into this issue. Resolving as worksforme.