Hi Jörg! On Wed, 20 Aug 2008 12:45:11 +0200, Jörg Sommer wrote: > Luca Capello schrieb am Tue 19. Aug, 22:53 (+0200): >> According to [1], I'd say that bootchart is correct, since it should be >> started after rmnologin: >> >> # Required-Start: $remote_fs rmnologin > > I don't think so.
I think you have missed my point, which is not at all related to bootchart. - with the current situation (i.e. a default Debian installation and no insserv), rmnologin is started just before stop-bootlogd: ===== (sid-amd64)[EMAIL PROTECTED]:/# ls -1 /etc/rc2.d/ README S99rc.local S99rmnologin S99stop-bootlogd (sid-amd64)[EMAIL PROTECTED]:/# ===== - as soon as insserv is enabled, rmnologin becomes the last script: ===== (sid-amd64)[EMAIL PROTECTED]:/# apt-get install insserv [...] (sid-amd64)[EMAIL PROTECTED]:/# update-bootsystem-insserv [...] (sid-amd64)[EMAIL PROTECTED]:/# ls -1 /etc/rc2.d/ README S01rc.local S02stop-bootlogd S03rmnologin (sid-amd64)[EMAIL PROTECTED]:/# ===== This is plainly wrong and has nothing to do with bootchart, since stop-bootlogd should start *after* rmnologin, according to the LSB headers: ===== (sid-amd64)[EMAIL PROTECTED]:/# grep Required-Start /etc/init.d/rmnologin # Required-Start: $remote_fs $all (sid-amd64)[EMAIL PROTECTED]:/# grep Required-Start /etc/init.d/stop-bootlogd # Required-Start: $remote_fs rmnologin (sid-amd64)[EMAIL PROTECTED]:/# ===== Now let bootchart enters the game! - with no inserv: ===== (sid-amd64)[EMAIL PROTECTED]:/# ls -1 /etc/rc2.d/ README S99bootchart S99rc.local S99rmnologin S99stop-bootlogd (sid-amd64)[EMAIL PROTECTED]:/# ===== And I've just discovered that this situation is wrong as well: S99rc.local could cointain other processes which would take time and would not be taken into account by bootchart. Thus, we need something like S99z_bootchart. Should I file a new bug? - with insserv: ===== (sid-amd64)[EMAIL PROTECTED]:/# ls -1 /etc/rc2.d/ README S01rc.local S02bootchart S02stop-bootlogd S03rmnologin (sid-amd64)[EMAIL PROTECTED]:/# ===== Again, this is wrong, since according to the LSB headers: ===== (sid-amd64)[EMAIL PROTECTED]:/# grep Required-Start /etc/init.d/bootchart # Required-Start: $remote_fs rmnologin (sid-amd64)[EMAIL PROTECTED]:/# ===== > It's valid to place the rmnologin script earlier in the boot sequence, > if you use a graphical login. Then bootchart would stop to early. While rmnologin can be started before, this is not the case in a plain Debian installation. However, from my points above, we can assume that the last script is stop-bootlogd and not rmnologin, thus I'd say that bootchart should be started *after* stop-bootlogd. > On <http://wiki.debian.org/LSBInitScripts>, it's said about $all: > ‘facility supported by insserv to start a script after all the other > scripts, at the end of the boot sequence.’ This is exactly what I want > ‘at the end of the boot sequence.’ Therefore, I think $all is correct > and enough. Again, given the points above, the correct line should be: # Required-Start: $remote_fs stop-bootlogd This, however, generates another problem: stop-bootlogd isn't started at init 1, while bootchart yes. > If insserv doesn't comply the spec, it's a bug in insserv. If the spec > isn't grantable, the spec should be reworked. This is exactly my point: it seems that insserv doesn't completely respect the Required-Start header. Thx, bye, Gismo / Luca
pgpX2KiavlrBX.pgp
Description: PGP signature
