Troubleshooting boot problems

[Patrick Goetz]

Scott James Remnant wrote:
>> When is ufw run vis' a vis the /etc/rc2.d scripts?
>>
> Before, after, during, etc.  There is no fixed relationship between
> these two things, except that /etc/rcS.d (and thus /etc/rc2.d) will not
> be run until the "lo" device is up.
> 

Then this is problematic for my example of the iptables script run from 
rc2.d.  In generally, this sounds like it could cause hard to debug, 
intermittent problems, and I need to just turn ufw off.  Sadly, this is 
installed by default for even the most basic server install?  Lots and 
lots of server users have legacy iptables firewall scripts and are going 
to be blind-sided by this....

 >> OK, well what service emits a runlevel event?
 >>
 > None.  It's emitted by the /sbin/telinit tool.

OK, but this doesn't this assume at least one run level change?  The 
rc2.d scripts are run, so something must be running telinit in order to 
trigger rc.conf -- who does that?


 >>> What emits the net-device-added event?
 >
 > net-device-added is emitted by the Upstart/udev bridge.  Any event of
 > the form ${SUBSYSTEM}-device-{added,changed,removed} is an
 > Upstart copy of a kernel event.
 >
 > You can watch those with, e.g. "udevadm monitor -e"
 >

I like Upstart, but this discussion should make it clear that we might 
be headed for a übergeek train wreck without better documentation and 
perhaps some management tools.  Most sys admins are not going to have 
time/interest to hang out on the ubuntu-devel-discuss list to find out 
why, say, autofs won't start ever since they upgraded their server from 
Lenny to Lucid and discover they have no idea how the system works any 
more.  There's a tipping point for technology adoption that revolves 
around continuity and clarity;  XSLT and XSL-FO are perfect examples of 
great technologies that have failed to be widely adopted because they're 
just too complicated to use, have some poorly thought out default 
behaviors, and there's not quite enough clear documentation explaining 
the idiosyncrasies. (Note that I'm not complaining, just pointing out 
potentially dangerous curves in the road.)