| Description | I have some systems that have been running CentOS 5.1, but with encrypted filesystems. After 5.2 was released (but without trying to do an explicit upgrade), yum picked up mkinitrd-5.1.19.6-28. During the update process (which was driven via a cron job running "yum -y update"), I noticed that mkinitrd was pegging the cpu and yum appeared stalled. After killing mkinitrd, yum completed but the newly installed kernel (2.6.18-92.1.1.el5) was not reflected in grub.conf.
Running mkinitrd manually showed the following diagnostic just before the cpu gets pegged:
/sbin/mkinitrd: line 368: cd: slaves: No such file or directory
(the line number may be off). This appears to be closely related to https://bugzilla.redhat.com/show_bug.cgi?id=439371 which was a bug fixed in this release of mkinitrd.
One one of the servers, the encrypted filesystem is set up via /etc/crypttab, and resides on a mirrored md device. On another (a laptop), all filesystems except for /boot reside on a single encrypted device (LVM on LUKS), as described at http://www.gno.org/~gdr/sysadmin/centos/5.1/crypto-disk.html
I've attached a patch which solves the issue. This patch comes in three chunks. Only the first chunk is necessary to solve the problem with the "official" crypto setup via /etc/crypttab. The second and third chunks are necessary for the laptop-type config described above. For those interested, the environment variables in the third chunk come from /etc/sysconfig/mkinitrd/crypto-root:
MODULES="$MODULES aes sha256 dm_crypt"
# Do NOT use the /dev/disk/by-* hierarchies here
CRYPTO_ROOT_DEV="/dev/hda2"
CRYPTO_ROOT_MAP="PV0"
I would have reported this upstream, but I don't have access to the RHEL bugzilla.
|
|---|
