RFC: Bootscript Changes

Nathan Coulson conathan at conet.dyndns.org
Fri Jan 30 02:21:40 PST 2004


Anyway, I did some work on the bootscripts, and done a bit of modification
[hopefully nothing that'll make people chase me with pitchforks].

I'm not completely sure about sleeping 5 instead of pausing on most errors
[This does not include checkfs].  I just prevented scripts for abnormally
terminating by tinkering with evaluate_retval.  exit 1 still causes it to
stop on errors.  [truth be told, I accidently submitted it].

Changelog:
        * Reorganized the network system to accept multiple services.
        * Ash Compliancy
        * Modified some scripts to follow template
        * Reorganized functions, based upon Ash Compliancy Patch
        * Added a sleep 5 to failure for evaulate_retval, and changed
          evaulate_retval to return true instead of false.  Instead, a
          exit 1 will indicate the script terminated abnormally, causing
          your system to wait until you hit ctrl-J
        * Switched "Press enter to continue" to "Press <ctrl-j> to continue"
        * replaced 3 with $KILLDELAY in sendsignals
        * mountfs does not mount network filesystems [identified by _netdev
          in /etc/fstab]
        * removed unneded touch out of cleanfs

[guess I have to find the coresponding bugs in bugzilla now]

There were a few more changes I was thinking of making:  [the ones I
either dont know about, or the ones people were arguing about earlier].

in the lfs-bootscript bugzilla bug, it mentions adding -d to umount.  I
have never used loopbacks, but it appears that adding it wouldnt hurt.

Also, I found a patch that preserved one level of directories on /var/run
and /var/lock.  should I add that as well?

Also thinking about expanding the setclock script, but NOT symlinking to
rc{0,6}.d/K##.  (That way, the functionality is there if someone wants it,
but not used by default).

---------
After these last two,  I was going to go through the mailing list, and see
what requests people wanted  [Add onto this thread if I forgot you (I got
overwhelmed earlier)]

-from memory

Having an alternative way of starting/stopping daemons [such as MySQL]. 
(Probably export PIDFILE= in the script, and killproc can see if it
exists).

Error Handling [logging?] (I'm not sure how this one will get handled, 
mkfifo sounds bad).  Pausing on critical errors is probably best.



More information about the lfs-dev mailing list