On my system, things go better when I (re-)replace the 'udevadm control --exit' in /scripts/init-bottom/udev with 'pkill udevd', as was the case in natty's udev.
Then the timeout-followed-by-readonly-fs on bad boots is instead replaced by a 60 second timeout followed by a seemingly clean boot. kern.log then shows:
Sep 21 12:20:51 o-nolvm kernel: [ 2.450409] init: Handling started event
Sep 21 12:20:51 o-nolvm kernel: [ 2.450572] init: Reconnecting to system bus
Sep 21 12:20:51 o-nolvm kernel: [ 2.451993] init: mountall post-stop process
(336) exited normally
Sep 21 12:20:51 o-nolvm kernel: [ 2.452038] init: mountall state changed from
post-stop to waiting
Sep 21 12:20:51 o-nolvm kernel: [ 2.452383] init: Handling stopped event
Sep 21 12:21:51 o-nolvm kernel: [ 61.925549] init: udevtrigger post-stop proce
ss (301) exited normally
the udevtrigger post-stop just does
post-stop exec udevadm settle
So I'm guessing that in both cases a hung handler or missing event is the problem.
On my system, things go better when I (re-)replace the 'udevadm control --exit' in /scripts/ init-bottom/ udev with 'pkill udevd', as was the case in natty's udev.
Then the timeout- followed- by-readonly- fs on bad boots is instead replaced by a 60 second timeout followed by a seemingly clean boot. kern.log then shows:
Sep 21 12:20:51 o-nolvm kernel: [ 2.450409] init: Handling started event
Sep 21 12:20:51 o-nolvm kernel: [ 2.450572] init: Reconnecting to system bus
Sep 21 12:20:51 o-nolvm kernel: [ 2.451993] init: mountall post-stop process
(336) exited normally
Sep 21 12:20:51 o-nolvm kernel: [ 2.452038] init: mountall state changed from
post-stop to waiting
Sep 21 12:20:51 o-nolvm kernel: [ 2.452383] init: Handling stopped event
Sep 21 12:21:51 o-nolvm kernel: [ 61.925549] init: udevtrigger post-stop proce
ss (301) exited normally
the udevtrigger post-stop just does
post-stop exec udevadm settle
So I'm guessing that in both cases a hung handler or missing event is the problem.