Bug#616689: Root-on-LVM setup fails often due to timing issues
My root-on-lvm system has worked reliably for some time, but fails to
boot every time with kernel 3.8.
I suspect the change may be due to the removal of the scsi_wait_scan
module in 3.6. (init-top/udev waits using the module if available.)
Bug 678688 suggests a solution involving repeated calls to a
"local-block" script on availability of a new block device, but it
doesn't seem to have been implemented.
An alternative would be for LVM to provide udev rules to automatically
activate the root vg/lv when the relevant devices are available. I'm not
sure at what point the information needed to make this determination
becomes available.
A third solution would be for the hook scripts to construct a list of
block devices needed to bring up the root vg/lv, and have the lvm
initramfs script wait for these specific devices to be available. This
is more fragile if disks are reconfigured.
Reply to: