Bug#401258: Reclose bug
Peter == Peter Palfrader [EMAIL PROTECTED] writes: Peter Now hprop gets disabled again in upgrades from sarge to Peter etch, and when the package is removed - not purged - the Peter services still are enabled. Peter Both of these issues would have been fixed with my version Peter of the patch. Unfortunately, there are a number of tradeoffs. I really don't like the idea of dpkg messing around with config files for every upgrade. An administrator may just have used update-inetd to disable the entry, and doesn't want it to be re-enabled. I did try to raise the issue at: http://lists.debian.org/debian-devel/2006/12/msg00197.html but unfortunately, I only got one response that agrees the solution I implemented. If you still disagree, then please raise the issue on debian-devel, I will follow what ever action is decided on debian-devel. (we really do need some consistent method of implementing this inetd stuff in all packages). -- Brian May [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#401258: Reclose bug
On Thu, 11 Jan 2007, Brian May wrote: Peter == Peter Palfrader [EMAIL PROTECTED] writes: Peter Now hprop gets disabled again in upgrades from sarge to Peter etch, and when the package is removed - not purged - the Peter services still are enabled. Peter Both of these issues would have been fixed with my version Peter of the patch. Unfortunately, there are a number of tradeoffs. I really don't like the idea of dpkg messing around with config files for every upgrade. An administrator may just have used update-inetd to disable the entry, and doesn't want it to be re-enabled. That's the same issue with people who disable services using update-rc.d remove basename - they are using the wrong tool for the job. I did try to raise the issue at: http://lists.debian.org/debian-devel/2006/12/msg00197.html but unfortunately, I only got one response that agrees the solution I implemented. I read that, but at the time I thought explaining it all in the bug report was sufficient. If you still disagree, then please raise the issue on debian-devel, I will follow what ever action is decided on debian-devel. I felt very, very strong about the issue and contemplated going to the tech ctte over your fix. In the end I decided to let it rest - I had upgraded my kerberos servers and if the maintainer doesn't care about breakage then I won't either. I would still welcome a proper fix, but it's your call. (we really do need some consistent method of implementing this inetd stuff in all packages). Yes. -- | .''`. ** Debian GNU/Linux ** Peter Palfrader | : :' : The universal http://www.palfrader.org/ | `. `' Operating System | `-http://www.debian.org/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#401258: Reclose bug
On Wed, 13 Dec 2006, Brian May wrote: Peter == Peter Palfrader [EMAIL PROTECTED] writes: Peter Actually, this is really bad. Peter It means the lines do _not_ get added on iditial Peter installations: Yes, as I worked out, too. Thanks for letting me know; I have now replaced the -n with a -z in version -8. Now hprop gets disabled again in upgrades from sarge to etch, and when the package is removed - not purged - the services still are enabled. Both of these issues would have been fixed with my version of the patch. Peter -- | .''`. ** Debian GNU/Linux ** Peter Palfrader | : :' : The universal http://www.palfrader.org/ | `. `' Operating System | `-http://www.debian.org/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#401258: Reclose bug
On Wed, 13 Dec 2006, Brian May wrote: I consider this cleaner then Peter's solution which always enables hprop on upgrades even if you don't want it. It does not always enable hprop. It only enables it if was previously enabled. It has always been my intention that hprop should be disabled by default. And it is, with my patch: [EMAIL PROTECTED]:/tmp/buildd# tail -n 2 /etc/inetd.conf kerberos-admstream tcp nowait root/usr/sbin/tcpd /usr/lib/heimdal-servers/kadmind #krb_prop stream tcp nowait root/usr/sbin/tcpd /usr/sbin/hpropd [EMAIL PROTECTED]:/tmp/buildd# dpkg -s heimdal-kdc | grep Version Version: 0.7.2.dfsg.1-6 [EMAIL PROTECTED]:/tmp/buildd# dpkg -i heimdal-kdc_0.7.2.dfsg.1-6+weasel_i386.deb (Reading database ... 18095 files and directories currently installed.) Preparing to replace heimdal-kdc 0.7.2.dfsg.1-6 (using heimdal-kdc_0.7.2.dfsg.1-6+weasel_i386.deb) ... Unpacking replacement heimdal-kdc ... Setting up heimdal-kdc (0.7.2.dfsg.1-6+weasel) ... [EMAIL PROTECTED]:/tmp/buildd# tail -n 2 /etc/inetd.conf kerberos-admstream tcp nowait root/usr/sbin/tcpd /usr/lib/heimdal-servers/kadmind #krb_prop stream tcp nowait root/usr/sbin/tcpd /usr/sbin/hpropd It also keeps the proper state when it is enabled: [EMAIL PROTECTED]:/tmp/buildd# tail -n 2 /etc/inetd.conf kerberos-admstream tcp nowait root/usr/sbin/tcpd /usr/lib/heimdal-servers/kadmind krb_propstream tcp nowait root/usr/sbin/tcpd /usr/sbin/hpropd [EMAIL PROTECTED]:/tmp/buildd# dpkg -s heimdal-kdc | grep Version Version: 0.7.2.dfsg.1-6 [EMAIL PROTECTED]:/tmp/buildd# dpkg -i heimdal-kdc_0.7.2.dfsg.1-6+weasel_i386.deb (Reading database ... 18095 files and directories currently installed.) Preparing to replace heimdal-kdc 0.7.2.dfsg.1-6 (using heimdal-kdc_0.7.2.dfsg.1-6+weasel_i386.deb) ... Unpacking replacement heimdal-kdc ... Setting up heimdal-kdc (0.7.2.dfsg.1-6+weasel) ... [EMAIL PROTECTED]:/tmp/buildd# tail -n 2 /etc/inetd.conf kerberos-admstream tcp nowait root/usr/sbin/tcpd /usr/lib/heimdal-servers/kadmind krb_propstream tcp nowait root/usr/sbin/tcpd /usr/sbin/hpropd What happens is that --disable does not touch commented lines, as my short changelog entry explained. So from kerberos-admstream tcp nowait root/usr/sbin/tcpd /usr/lib/heimdal-servers/kadmind krb_propstream tcp nowait root/usr/sbin/tcpd /usr/sbin/hpropd the prerm makes #off# kerberos-admstream tcp nowait root/usr/sbin/tcpd /usr/lib/heimdal-servers/kadmind #off# krb_propstream tcp nowait root/usr/sbin/tcpd /usr/sbin/hpropd and my postinst enables it back to kerberos-admstream tcp nowait root/usr/sbin/tcpd /usr/lib/heimdal-servers/kadmind krb_propstream tcp nowait root/usr/sbin/tcpd /usr/sbin/hpropd And from kerberos-admstream tcp nowait root/usr/sbin/tcpd /usr/lib/heimdal-servers/kadmind #krb_propstream tcp nowait root/usr/sbin/tcpd /usr/sbin/hpropd we end up at #off# kerberos-admstream tcp nowait root/usr/sbin/tcpd /usr/lib/heimdal-servers/kadmind #krb_prop stream tcp nowait root/usr/sbin/tcpd /usr/sbin/hpropd and back at kerberos-admstream tcp nowait root/usr/sbin/tcpd /usr/lib/heimdal-servers/kadmind #krb_propstream tcp nowait root/usr/sbin/tcpd /usr/sbin/hpropd after the postinst. Peter is mistaken, add_servers is only called if $2 is empty - i.e. on an initial installation. Am not: [EMAIL PROTECTED]:~# tail -n 2 /etc/inetd.conf kerberos-admstream tcp nowait root/usr/sbin/tcpd /usr/lib/heimdal-servers/kadmind krb_propstream tcp nowait root/usr/sbin/tcpd /usr/sbin/hpropd [EMAIL PROTECTED]:~# dpkg -s heimdal-kdc | grep Version Version: 0.7.2.dfsg.1-6 [EMAIL PROTECTED]:~# dpkg -i heimdal-kdc_0.7.2.dfsg.1-7_i386.deb # with set -x in its postinst (Reading database ... 18095 files and directories currently installed.) Preparing to replace heimdal-kdc 0.7.2.dfsg.1-6 (using heimdal-kdc_0.7.2.dfsg.1-7_i386.deb) ... Unpacking replacement heimdal-kdc ... Setting up heimdal-kdc (0.7.2.dfsg.1-7) ... + . /usr/share/debconf/confmodule ++ '[' '!' '' ']' ++ PERL_DL_NONLAZY=1 ++ export PERL_DL_NONLAZY ++ '[' '' ']' ++ exec /usr/share/debconf/frontend /var/lib/dpkg/info/heimdal-kdc.postinst configure 0.7.2.dfsg.1-6 + . /usr/share/debconf/confmodule ++ '[' '!' 1
Bug#401258: Reclose bug
On Wed, 13 Dec 2006, Peter Palfrader wrote: Peter is mistaken, add_servers is only called if $2 is empty - i.e. on an initial installation. | if [ -n $2 ] | then | add_servers Actually, this is really bad. It means the lines do _not_ get added on iditial installations: [EMAIL PROTECTED]:/# dpkg -l heimdal-kdc Desired=Unknown/Install/Remove/Purge/Hold | Status=Not/Installed/Config-files/Unpacked/Failed-config/Half-installed |/ Err?=(none)/Hold/Reinst-required/X=both-problems (Status,Err: uppercase=bad) ||/ Name VersionDescription +++-==-==-=== pn heimdal-kdcnone (no description available) [EMAIL PROTECTED]:/# dpkg -i heimdal-kdc_0.7.2.dfsg.1-7_i386.deb Selecting previously deselected package heimdal-kdc. (Reading database ... 12797 files and directories currently installed.) Unpacking heimdal-kdc (from heimdal-kdc_0.7.2.dfsg.1-7_i386.deb) ... Setting up heimdal-kdc (0.7.2.dfsg.1-7) ... [..] + cp -a /usr/share/doc/heimdal-kdc/examples/kadmind.acl /etc/heimdal-kdc/kadmind.acl + kstash --master-key-fd=0 kstash: writing key to `/var/lib/heimdal-kdc/m-key' + echo -e '\n\n' + kadmin -l init CAME.SBG.AC.AT Realm max ticket life [unlimited]:Realm max renewable ticket life [unlimited]:+ touch /etc/heimdal-kdc/.configured + case $1 in + '[' -n '' ']' + dpkg --compare-versions '' le 0.7.2.dfsg.1-6 + enable_servers + update-inetd --pattern '[ \t]/usr/lib/heimdal-servers/kadmind' --enable kerberos-adm + '[' -x /etc/init.d/heimdal-kdc ']' + update-rc.d heimdal-kdc defaults ++ which invoke-rc.d + '[' -x /usr/sbin/invoke-rc.d ']' + invoke-rc.d heimdal-kdc start [EMAIL PROTECTED]:/# grep heimdal /etc/inetd.conf [EMAIL PROTECTED]:/# -- | .''`. ** Debian GNU/Linux ** Peter Palfrader | : :' : The universal http://www.palfrader.org/ | `. `' Operating System | `-http://www.debian.org/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#401258: Reclose bug
Peter == Peter Palfrader [EMAIL PROTECTED] writes: Peter Actually, this is really bad. Peter It means the lines do _not_ get added on iditial Peter installations: Yes, as I worked out, too. Thanks for letting me know; I have now replaced the -n with a -z in version -8. -- Brian May [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]