[Merge] lp:~ubuntu-core-dev/ubuntu/utopic/sysvinit/unreviewed into lp:ubuntu/sysvinit

Martin Pitt martin.pitt at ubuntu.com
Thu May 22 11:18:45 UTC 2014


Review: Approve

This just merely changes the order and is noise (in debian/initscripts.preinst):

180 	+ eliminate_conffile "/etc/init.d/mountoverflowtmp"
181	eliminate_conffile "/etc/init.d/rmnologin"
182	eliminate_conffile "/etc/init.d/mtab.sh"
183	- eliminate_conffile "/etc/init.d/mountoverflowtmp"

invoke-rc.d:

1365	if which initctl >/dev/null && initctl version 2>/dev/null | grep -q upstart \
1366	- && [ -e "$UPSTARTDIR/${INITSCRIPTID}.conf" ]
1367	+ && initctl status ${INITSCRIPTID} 1>/dev/null 2>/dev/null

This changes behaviour in chroots: If the host system is running upstart, the chroot has /etc/init/$INITSCRIPTID.conf, but the host doesn't, then in the current code this would still try and use upstart. With this MP this will then be false as the host upstart doesn't know about a job named ${INITSCRIPTID}. There is a fallback later in the code, we just need to be extra-sure that this is really the behaviour that we want. It might still be safer to just explicitly enumerate all possible locations for an upstart job? (NB, I haven't really understood yet how upstart's chroot support works).

The rest looks fine to me (thanks for your further explanations about the .legacy-bootordering check etc.). Thanks for working on this!
-- 
https://code.launchpad.net/~ubuntu-core-dev/ubuntu/utopic/sysvinit/unreviewed/+merge/219999
Your team Ubuntu branches is subscribed to branch lp:ubuntu/sysvinit.



More information about the Ubuntu-reviews mailing list