Solved Debian update - keyboard responsive, Lennart Sorrenson not so much

[Lennart Sorensen]

On Wed, Nov 02, 2011 at 07:48:02AM -0400, Russell Reiter wrote:
> That's where I'm stuck on this. Unless you have a solid set of rules,
> devices re polled by udev seem to be reordered in assignment. That's a
> major complaint I see in the various blog's about udev. Udev
> diagnostic tools have improved but there are still some gaps and
> people still encounter problems.

Actually the reordering has always been a problem with linux since
forever.  udev in fact is probably the best thing ever at solving it
since you can tell it what to name thigns based on characteristics of
the device.

It used to be quite commen that if you had two disk controllers, the
load order could be completely random and hence the names could swap.
Go from one kernel version to another and the order PCI devices got
setup changed.  Suddenly eth0 and eth1 swap for no apparent reason.
udev nicely lets you say 'this mac addres is eth0, this other one
is eth1'.

But yes it still needs rules to do it, but at least it can.

> Every time something new, like LSB init is introduced, there is a
> learning curve.

Well insserv (as Debian uses) is just one init system that can use the
dependancy info in the LSB headers.  It is by no means the only one that
could be used and others might work differently.  I don't know of any
that do behave particularly differently though since parallel init based
on dependancies do seem to be the popular style now.

> I see this as a transitional state, not quite right, (according to
> convention) not quite wrong (according to function). I believe there
> is more that one right way to do things in Linux. Whether or not the
> chosen fix conforms with theory is demonstrated over time, assuming
> the machine doesn't grind to a halt when the fix (kludge) is applied.

Well it seems the best fix is to get the system to the same state as a
freshly installed one would be.  In the case of udev that means only one
symlinked script for starting it in rcS.d.  It should not be anywhere
else and with insserv won't do anything anyhow anywhere else.

-- 
Len Sorensen
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists