Re: The debian boot dependency graph image

2006-09-10 Thread Petter Reinholdtsen

[Nathanael Nerode]
 syslog-ng also lacks information.

If you find such packages, please report bugs to bts about it.  Most
of the base system and selected packages I use often have such bugs
reported against them, but there are lots of packages left to fix.

 The udev dependency information is not really accurate: a lot of
 things depend upon udev running first, and don't say so.  This is
 partly because the long-term plan is to run udev in the chroot, but
 I think for now probably the dependency should be specified.

Actually, it is more accurate now.  mountdevsubfs.sh now depend on
udev if it exist, and thus udev is executed where it should be,
between mountkernfs and mountdevsubfs.  So now insserv will generate
this boot sequence for the base system+udev based on existing LSB
headers and override files:

  /etc/rcS.d/
  README   S05checkroot.sh   S09mountall-bootclean.sh
  S01glibc.sh  S06ifupdown-clean S09urandom
  S01hostname.sh   S06module-init-tools  S10networking
  S01mountkernfs.shS06mtab.shS11mountnfs.sh
  S02procps.sh S06udev-mtab  S12mountnfs-bootclean.sh
  S02udev  S07checkfs.sh S13bootmisc.sh
  S03mountdevsubfs.sh  S07libdevmapper1.02   S13hwclock.sh
  S04bootlogd  S08mountall.shS13stop-bootlogd-single
  S04hwclockfirst.sh   S09ifupdown

  /etc/rc2.d/
  README  S13openbsd-inetd  S14sysklogd  S15klogd  S17rmnologin
  S09makedev  S13rc.local   S15cron  S16stop-bootlogd

Recently I have started checking the shutdown sequence too, and it is
mostly correct, though there are some issues left to fix before it is
properly handled:

  /etc/rc6.d/
  K06cron  K08openbsd-inetd  K25hwclock.shK35networking  K60umountroot
  K06klogd K20makedevK30urandom   K36ifupdownK90reboot
  K07sysklogd  K20sendsigs   K31umountnfs.sh  K40umountfsREADME

Friendly,
-- 
Petter Reinholdtsen


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Re: The debian boot dependency graph image

2006-09-09 Thread Nathanael Nerode
Petter Reinholdtsen wrote:

 The scripts listed in the upper right corner are all those scripts
 without dependency information available.  This is the complete list
 for my installation:
 
   hwclockfirst.sh ifupdown-clean modutils hwclock.sh libdevmapper1.01
   libdevmapper1.02 lvm hibernate ifupdown nviboot xserver-xorg vbesave
   sysklogd klogd acct acpid apmd apt-index-watcher atftpd cupsys
   dbus-1 nullmailer openbsd-inetd rsync snmpd ssh uml-utilities
   snmptrapfmt anacron binfmt-support acpi-support libnss-ldap

syslog-ng also lacks information.  (We should be getting rid of sysklogd and
klogd in favor of syslog-ng or metalog; I don't know why we haven't yet.)

The udev dependency information is not really accurate:
a lot of things depend upon udev running first, and don't say so.  This is
partly because the long-term plan is to run udev in the chroot, but I think
for now probably the dependency should be specified.

-- 
Nathanael Nerode  [EMAIL PROTECTED]

Bush admitted to violating FISA and said he was proud of it.
So why isn't he in prison yet?...


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Re: The debian boot dependency graph image

2006-09-08 Thread Petter Reinholdtsen

[Francesco Pedrini]
 I've tried, but it doesn't work:

You need to use insserv version 1.08.0-3 or later.  Sorry for
forgetting to mention that.

 [EMAIL PROTECTED]:~$ sudo /usr/share/insserv/check-initd-order -g -o 
lsb-graph.dot

There is no need to run it as root, I believe.  It only need to read
the init.d scripts, and those are readable by all.

 Unknown option: o
 Unable to properly handle multiple provides: mountdevsubfs mountvirtfs
 LSB header missing in /etc/rcS.d/S25libdevmapper1.02
 Use of uninitialized value in split 
 at /usr/share/insserv/check-initd-order line 69.
 Use of uninitialized value in concatenation (.) or string 
 at /usr/share/insserv/check-initd-order line 84.
 Use of uninitialized value in concatenation (.) or string 
 at /usr/share/insserv/check-initd-order line 93.
 LSB header missing in /etc/rcS.d/S70screen-cleanup
 LSB header missing in /etc/rcS.d/S75schroot
 LSB header missing in /etc/rcS.d/S80installation-report
 LSB header missing in /etc/rc2.d/S19spamassassin
 LSB header missing in /etc/rc2.d/S20apt-index-watcher
 Use of uninitialized value in split 
 at /usr/share/insserv/check-initd-order line 69.
 Use of uninitialized value in concatenation (.) or string 
 at /usr/share/insserv/check-initd-order line 84.
 Use of uninitialized value in concatenation (.) or string 
 at /usr/share/insserv/check-initd-order line 84.
 Use of uninitialized value in concatenation (.) or string 
 at /usr/share/insserv/check-initd-order line 93.
 Use of uninitialized value in split 
 at /usr/share/insserv/check-initd-order line 69.
 Use of uninitialized value in concatenation (.) or string 
 at /usr/share/insserv/check-initd-order line 84.
 Use of uninitialized value in concatenation (.) or string 
 at /usr/share/insserv/check-initd-order line 93.
 LSB header missing in /etc/rc2.d/S20ddclient

The warnings are mostly fixed in the later version of the script.  The
missing LSB headers are real problems with the packages including the
scripts.  This is the output when I run the svn version on my sid chroot:

  % ./debian/check-initd-order -o -g  /dev/null
  LSB header missing in /etc/rcS.d/S18hwclockfirst.sh
  LSB header missing in /etc/rcS.d/S18ifupdown-clean
  LSB header missing in /etc/rcS.d/S20modutils
  LSB header missing in /etc/rcS.d/S22hwclock.sh
  LSB header missing in /etc/rcS.d/S25libdevmapper1.01
  LSB header missing in /etc/rcS.d/S25libdevmapper1.02
  LSB header missing in /etc/rcS.d/S26lvm
  LSB header missing in /etc/rcS.d/S31hibernate
  error: File ../init.d/pppd-dns is missing the provides header
  LSB header missing in /etc/rcS.d/S39ifupdown
  LSB header missing in /etc/rcS.d/S70nviboot
  LSB header missing in /etc/rcS.d/S71xserver-xorg
  LSB header missing in /etc/rc2.d/S01acct
  LSB header missing in /etc/rc2.d/S01atftpd
  LSB header missing in /etc/rc2.d/S01binfmt-support
  LSB header missing in /etc/rc2.d/S01openbsd-inetd
  LSB header missing in /etc/rc2.d/S01snmpd
  LSB header missing in /etc/rc2.d/S01uml-utilities
  LSB header missing in /etc/rc2.d/S01vbesave
  LSB header missing in /etc/rc2.d/S02hwclock.sh
  LSB header missing in /etc/rc2.d/S02nviboot
  LSB header missing in /etc/rc2.d/S02snmptrapfmt
  LSB header missing in /etc/rc2.d/S02xserver-xorg
  LSB header missing in /etc/rc2.d/S03lvm
  LSB header missing in /etc/rc2.d/S04acpi-support
  LSB header missing in /etc/rc2.d/S04libnss-ldap
  LSB header missing in /etc/rc2.d/S05modutils
  LSB header missing in /etc/rc2.d/S06libdevmapper1.01
  LSB header missing in /etc/rc2.d/S08sysklogd
  LSB header missing in /etc/rc2.d/S09acpid
  LSB header missing in /etc/rc2.d/S09anacron
  LSB header missing in /etc/rc2.d/S09apmd
  LSB header missing in /etc/rc2.d/S09cupsys
  LSB header missing in /etc/rc2.d/S09dbus-1
  Unable to properly handle multiple provides: exim4 mail-transport-agent
  LSB header missing in /etc/rc2.d/S09klogd
  LSB header missing in /etc/rc2.d/S09nullmailer
  LSB header missing in /etc/rc2.d/S09rsync
  LSB header missing in /etc/rc2.d/S09ssh
  LSB header missing in /etc/rc2.d/S10apt-index-watcher
  Unable to properly handle multiple provides: x-display-manager kdm
  %

Still some work left before the dependencies are documented and
correct. :)

 Unable to read /etc/rc2.d/S20inetd 
 at /usr/share/insserv/check-initd-order line 180.

You seem to have a dangling pointer, /etc/rc2.d/S20inetd.  The script
can't handle those.  I'll modify it to make sure that isn't a fatal
problem in the next release of insserv.

 i get the same error even without -o option...

 i've tested it on PPC, the insserv version is 1.08.0-1

Thanks for testing.

Friendly,
-- 
Petter Reinholdtsen


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Re: The debian boot dependency graph image

2006-09-08 Thread Wouter Verhelst
On Thu, Sep 07, 2006 at 05:09:49PM +0200, Marco d'Itri wrote:
 On Sep 07, Wouter Verhelst [EMAIL PROTECTED] wrote:
 
  On Thu, Sep 07, 2006 at 03:11:02PM +0200, Marco d'Itri wrote:
   Now, try thinking about how many of the blocks which are not listed
   as depending on udev actually do.
  None, because udev isn't actually a hard dependency for any of those
  scripts, so listing it as such is wrong.
 If it's in the picture then it's relevante.

My point being that you don't want to list them as hard dependencies
unless you really need udev and can't work with static devices for some
other reason.

Moreover, some dependencies may depend and may have to be rearranged
depending on configuration. I don't think it's possible to express that
with LSB-style initscripts, unfortunately.

-- 
Lo-lan-do Home is where you have to wash the dishes.
  -- #debian-devel, Freenode, 2004-09-22


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Re: The debian boot dependency graph image

2006-09-08 Thread Petter Reinholdtsen

[Wouter Verhelst]
 My point being that you don't want to list them as hard dependencies
 unless you really need udev and can't work with static devices for
 some other reason.

I suspect very few packages should depend on udev.  Most packages
depend on local and remote file systems being mounted, and it is up to
the boot system to get to that stage.

The next upload of mountdevsub.sh will have a should-start dependency
on udev, to make sure the proper order is generated.

 Moreover, some dependencies may depend and may have to be rearranged
 depending on configuration. I don't think it's possible to express
 that with LSB-style initscripts, unfortunately.

Actually, I've implemented the concept of override files in inssert, a
program reading these headers, and the local admin can add a file in
/etc/ to modify the configuration.

Friendly,
-- 
Petter Reinholdtsen


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



The debian boot dependency graph image

2006-09-07 Thread Petter Reinholdtsen

For some time now, the runtime dependencies of init.d scripts have
been documented in the scripts, using the LSB convention.  Now, enough
scripts have this information present to make a useful graph of the
dependencies in the debian boot.  The current state of affairs in my
sid chroot look like this:

  URL: http://user.skolelinux.no/~pere/debian/lsb-info-20060907.png 

The scripts listed in the upper right corner are all those scripts
without dependency information available.  This is the complete list
for my installation:

  hwclockfirst.sh ifupdown-clean modutils hwclock.sh libdevmapper1.01
  libdevmapper1.02 lvm hibernate ifupdown nviboot xserver-xorg vbesave
  sysklogd klogd acct acpid apmd apt-index-watcher atftpd cupsys
  dbus-1 nullmailer openbsd-inetd rsync snmpd ssh uml-utilities
  snmptrapfmt anacron binfmt-support acpi-support libnss-ldap

A status summary for the packages used in a desktop installation is
available from
URL:http://initscripts-ng.alioth.debian.org/soc2006-bootsystem/lsblist.html.

The dotty file used to generate this graph is available from

  URL: http://user.skolelinux.no/~pere/debian/lsb-info-20060907.dot 

To generate your own graph, install the 'insserv' package, and run
these commands to get  your own

  /usr/share/insserv/check-initd-order -g -o  lsb-graph.dot
  dot -Tpng -o lsb-graph.png lsb-graph.dot

If you want to make a graph using the dependency information provided
in the insserv package for the scripts without such info, remove the
-o flag.

Friendly,
-- 
Petter Reinholdtsen


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Re: The debian boot dependency graph image

2006-09-07 Thread Marco d'Itri
On Sep 07, Petter Reinholdtsen [EMAIL PROTECTED] wrote:

 For some time now, the runtime dependencies of init.d scripts have
 been documented in the scripts, using the LSB convention.  Now, enough
 scripts have this information present to make a useful graph of the
 dependencies in the debian boot.  The current state of affairs in my
 sid chroot look like this:
 
   URL: http://user.skolelinux.no/~pere/debian/lsb-info-20060907.png 
Now, try thinking about how many of the blocks which are not listed
as depending on udev actually do.

-- 
ciao,
Marco


signature.asc
Description: Digital signature


Re: The debian boot dependency graph image

2006-09-07 Thread Pierre Habouzit
Le jeu 7 septembre 2006 15:11, Marco d'Itri a écrit :
 On Sep 07, Petter Reinholdtsen [EMAIL PROTECTED] wrote:
  For some time now, the runtime dependencies of init.d scripts have
  been documented in the scripts, using the LSB convention.  Now,
  enough scripts have this information present to make a useful graph
  of the dependencies in the debian boot.  The current state of
  affairs in my sid chroot look like this:
 
URL:
  http://user.skolelinux.no/~pere/debian/lsb-info-20060907.png 

 Now, try thinking about how many of the blocks which are not listed
 as depending on udev actually do.

your point beeing ?
-- 
·O·  Pierre Habouzit
··O[EMAIL PROTECTED]
OOOhttp://www.madism.org


pgpI8Xm2X0357.pgp
Description: PGP signature


Re: The debian boot dependency graph image

2006-09-07 Thread Wouter Verhelst
On Thu, Sep 07, 2006 at 03:11:02PM +0200, Marco d'Itri wrote:
 Now, try thinking about how many of the blocks which are not listed
 as depending on udev actually do.

None, because udev isn't actually a hard dependency for any of those
scripts, so listing it as such is wrong.

-- 
Lo-lan-do Home is where you have to wash the dishes.
  -- #debian-devel, Freenode, 2004-09-22


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Re: The debian boot dependency graph image

2006-09-07 Thread Marco d'Itri
On Sep 07, Pierre Habouzit [EMAIL PROTECTED] wrote:

   http://user.skolelinux.no/~pere/debian/lsb-info-20060907.png 
  Now, try thinking about how many of the blocks which are not listed
  as depending on udev actually do.
 your point beeing ?
That if you want a really dependency-based boot process more work will
be needed.

-- 
ciao,
Marco


signature.asc
Description: Digital signature


Re: The debian boot dependency graph image

2006-09-07 Thread Marco d'Itri
On Sep 07, Wouter Verhelst [EMAIL PROTECTED] wrote:

 On Thu, Sep 07, 2006 at 03:11:02PM +0200, Marco d'Itri wrote:
  Now, try thinking about how many of the blocks which are not listed
  as depending on udev actually do.
 None, because udev isn't actually a hard dependency for any of those
 scripts, so listing it as such is wrong.
If it's in the picture then it's relevante.

-- 
ciao,
Marco


signature.asc
Description: Digital signature


Re: The debian boot dependency graph image

2006-09-07 Thread Martin Wuertele
* Marco d'Itri [EMAIL PROTECTED] [2006-09-07 15:11]:

 On Sep 07, Petter Reinholdtsen [EMAIL PROTECTED] wrote:
 
  For some time now, the runtime dependencies of init.d scripts have
  been documented in the scripts, using the LSB convention.  Now, enough
  scripts have this information present to make a useful graph of the
  dependencies in the debian boot.  The current state of affairs in my
  sid chroot look like this:
  
URL: http://user.skolelinux.no/~pere/debian/lsb-info-20060907.png 
 Now, try thinking about how many of the blocks which are not listed
 as depending on udev actually do.
 
I can't think of a single one that would not work with good old static
dev tough udev might be the recommended way.

yours Martin
-- 
[EMAIL PROTECTED]  Debian GNU/Linux - The Universal Operating System
fabbione I AM ON TV!
fabbione CIAO MAMMA!


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Re: The debian boot dependency graph image

2006-09-07 Thread Josselin Mouette
Le jeudi 07 septembre 2006 à 15:18 +0200, Martin Wuertele a écrit :
 URL: http://user.skolelinux.no/~pere/debian/lsb-info-20060907.png 
  Now, try thinking about how many of the blocks which are not listed
  as depending on udev actually do.
  
 I can't think of a single one that would not work with good old static
 dev tough udev might be the recommended way.

But nevertheless, they won't work if udev is installed, until the device
appears. The current boot scheme is unsuitable for such cases.
-- 
 .''`.   Josselin Mouette/\./\
: :' :   [EMAIL PROTECTED]
`. `'[EMAIL PROTECTED]
   `-  Debian GNU/Linux -- The power of freedom



Re: The debian boot dependency graph image

2006-09-07 Thread Stephen Gran
This one time, at band camp, Josselin Mouette said:
 Le jeudi 07 septembre 2006 à 15:18 +0200, Martin Wuertele a écrit :
   
  I can't think of a single one that would not work with good old static
  dev tough udev might be the recommended way.
 
 But nevertheless, they won't work if udev is installed, until the device
 appears. The current boot scheme is unsuitable for such cases.

This is what the Should-Start filed is for, AIUI.
-- 
 -
|   ,''`.Stephen Gran |
|  : :' :[EMAIL PROTECTED] |
|  `. `'Debian user, admin, and developer |
|`- http://www.debian.org |
 -


signature.asc
Description: Digital signature


Re: The debian boot dependency graph image

2006-09-07 Thread Martin Wuertele
* Marco d'Itri [EMAIL PROTECTED] [2006-09-07 17:11]:

 On Sep 07, Wouter Verhelst [EMAIL PROTECTED] wrote:
 
  On Thu, Sep 07, 2006 at 03:11:02PM +0200, Marco d'Itri wrote:
   Now, try thinking about how many of the blocks which are not listed
   as depending on udev actually do.
  None, because udev isn't actually a hard dependency for any of those
  scripts, so listing it as such is wrong.
 If it's in the picture then it's relevante.
 
Then the picture/scripts should be fixed.

yours Martin
-- 
[EMAIL PROTECTED]  Debian GNU/Linux - The Universal Operating System
NForcer NP: Imperanon - Shadowsouls
alphascorpii NE: Schokomüsli! :)



Re: The debian boot dependency graph image

2006-09-07 Thread Francesco Pedrini
On Thursday 07 September 2006 12:03, Petter Reinholdtsen wrote:
 For some time now, the runtime dependencies of init.d scripts have
 been documented in the scripts, using the LSB convention.  Now,
 enough scripts have this information present to make a useful graph
 of the dependencies in the debian boot.  The current state of affairs
 in my sid chroot look like this:

   URL: http://user.skolelinux.no/~pere/debian/lsb-info-20060907.png
[SNIP]

 If you want to make a graph using the dependency information provided
 in the insserv package for the scripts without such info, remove the
 -o flag.

I've tried, but it doesn't work:

[EMAIL PROTECTED]:~$ sudo /usr/share/insserv/check-initd-order -g -o 
lsb-graph.dot
Unknown option: o
Unable to properly handle multiple provides: mountdevsubfs mountvirtfs
LSB header missing in /etc/rcS.d/S25libdevmapper1.02
Use of uninitialized value in split 
at /usr/share/insserv/check-initd-order line 69.
Use of uninitialized value in concatenation (.) or string 
at /usr/share/insserv/check-initd-order line 84.
Use of uninitialized value in concatenation (.) or string 
at /usr/share/insserv/check-initd-order line 93.
LSB header missing in /etc/rcS.d/S70screen-cleanup
LSB header missing in /etc/rcS.d/S75schroot
LSB header missing in /etc/rcS.d/S80installation-report
LSB header missing in /etc/rc2.d/S19spamassassin
LSB header missing in /etc/rc2.d/S20apt-index-watcher
Use of uninitialized value in split 
at /usr/share/insserv/check-initd-order line 69.
Use of uninitialized value in concatenation (.) or string 
at /usr/share/insserv/check-initd-order line 84.
Use of uninitialized value in concatenation (.) or string 
at /usr/share/insserv/check-initd-order line 84.
Use of uninitialized value in concatenation (.) or string 
at /usr/share/insserv/check-initd-order line 93.
Use of uninitialized value in split 
at /usr/share/insserv/check-initd-order line 69.
Use of uninitialized value in concatenation (.) or string 
at /usr/share/insserv/check-initd-order line 84.
Use of uninitialized value in concatenation (.) or string 
at /usr/share/insserv/check-initd-order line 93.
LSB header missing in /etc/rc2.d/S20ddclient
Unable to read /etc/rc2.d/S20inetd 
at /usr/share/insserv/check-initd-order line 180.

i get the same error even without -o option...

i've tested it on PPC, the insserv version is 1.08.0-1


HTH

Francesco

-- 
:wq


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]