Please test dependency based boot sequencing

[Petter Reinholdtsen]

The last week or so, I have worked on the dependency based boot
sequencing, and I am happy to report that it actually work now.  There
are a few corner cases, but in the common case (as in with the
packages included in a desktop install), the boot ordering is working
as it should.

Now it needs testers, and people to report packages with missing LSB
headers in their init.d scripts, as well as testers to detect bugs in
the dependency information provided by init.d scripts.  Please help me
test this feature to get it ready for Lenny.

This is the boot sequence in my sid chroot, to give you an idea what
dependency based boot sequencing look like at the moment.

/etc/rcS.d/:
README                S06mtab.sh                S12nbd-client
S01glibc.sh           S06udev-mtab              S12networking
S02hostname.sh        S07lvm2                   S13portmap
S02mountkernfs.sh     S08cryptdisks             S14nfs-common
S03udev               S08resize_lvm             S15mountnfs.sh
S04keyboard-setup     S09checkfs.sh             S16mountnfs-bootclean.sh
S04mountdevsubfs.sh   S10mountall.sh            S17bootmisc.sh
S04pcmciautils        S11alsa-utils             S17console-screen.sh
S04procps             S11discover               S17console-setup
S05bootlogd           S11fuse                   S17irqbalance
S05checkroot.sh       S11hwclock.sh             S17libpam-foreground
S05hwclockfirst.sh    S11ifupdown               S17nviboot
S05keymap.sh          S11mountall-bootclean.sh  S17x11-common
S06cryptdisks-early   S11urandom                S18stop-bootlogd-single
S06ifupdown-clean     S12libdevmapper1.02.1
S06module-init-tools  S12mountoverflowtmp

/etc/rc2.d:
README            S14klogd              S17open-backdoor
S01halt           S14nfs-common         S17openbsd-inetd
S01uml-utilities  S14nviboot            S17pcscd
S02kexec          S14rsync              S17preload
S03umountroot     S14slapd              S17rc.local
S04pcmciautils    S15avahi-daemon       S17report-reboot
S04umountfs       S15nfs-kernel-server  S17sitesummary-client
S11kexec-load     S15umountnfs.sh       S17spamassassin
S13nbd-client     S16sendsigs           S17ssh
S13portmap        S17atd                S17start-wlan
S13sysklogd       S17boot_xconf         S17update-hostname
S13yaws           S17cfengine2          S18bootchart
S14atftpd         S17console-screen.sh  S18exim4
S14cron           S17enable-nat         S18stop-bootlogd
S14cupsys         S17fetch-ldap-cert    S19rmnologin
S14dbus           S17hotkey-setup       S19usplash
S14fam            S17irqbalance
S14gpsd           S17nbd-server

/etc/rc6.d/:
K01rmnologin           K03ssh                K07yaws
K01usplash             K03start-wlan         K08mountoverflowtmp
K02exim4               K03update-hostname    K08networking
K03atd                 K04sendsigs           K09alsa-utils
K03boot_xconf          K05avahi-daemon       K09fuse
K03cfengine2           K05nfs-kernel-server  K09hwclock.sh
K03console-screen.sh   K05umountnfs.sh       K09ifupdown
K03console-setup       K06atftpd             K09kexec-load
K03enable-nat          K06cron               K09urandom
K03hotkey-setup        K06cupsys             K12cryptdisks
K03irqbalance          K06dbus               K13lvm2
K03nbd-server          K06fam                K14cryptdisks-early
K03nviboot             K06gpsd               K16keyboard-setup
K03open-backdoor       K06klogd              K16pcmciautils
K03openbsd-inetd       K06nfs-common         K16umountfs
K03pcscd               K06rsync              K17umountroot
K03preload             K06slapd              K18kexec
K03report-reboot       K07nbd-client         K19reboot
K03sitesummary-client  K07portmap            K19uml-utilities
K03spamassassin        K07sysklogd           README

The bug with lvm2 being ordered before umountfs during shutdown has
been fixed in initscripts version 2.86.ds1-47.  See
<URL:http://wiki.debian.org/LSBInitScripts/DependencyBasedBoot> for
instructions and known issues with the dependency headers in unstable.

Several more scripts to depend on $remote_fs for the shutdown order,
to make sure they get a chance to kill their process before it is done
by sendsigs (or they should just just remove the stop symlinks in
runlevel 0 and 6, if that make sense for them).

Happy hacking,
-- 
Petter Reinholdtsen