rwp | cousin_luigi, $all was basically an easy way to avoid declaring dependies and to just say it depended upon $all of everything else. | 05:44 |
---|---|---|
rwp | The effect is that init scripts depending upon $all are run last in the order. | 05:44 |
rwp | It's frowned upon because it avoids saying what is actually needed. And also if there are a dozen scripts that all depend upon $all then what will be the order between them? | 05:44 |
rwp | Your package is a router/firewall related thing? In which case it really should run during the S single user state which is passed through before runlevel 2 the multiuser state. | 05:45 |
rwp | It should run early rather than late. So depending upon $all really seems like the wrong direction. | 05:45 |
rwp | Note that I have not looked at the recent state of your package after you have done all of the cleanup to it that you have been doing. | 05:45 |
cousin_luigi | rwp: https://gitlab.com/lbaldoni/voyage-util/-/blob/debhelper/debian/voyage-util.init?ref_type=heads#L71 this is one of the doubts I have. What's that for? | 06:30 |
cousin_luigi | (from the script that used to be run after $all) | 06:31 |
rwp | I agree cousin_luigi that line looks pretty crazy there! | 06:45 |
cousin_luigi | I think it might have been related to the first install of voyage linux, which was somewhat convoluted. | 06:45 |
rwp | Seeing that start_leds routine... Is this for a Raspberry Pi? | 06:46 |
rwp | Do you understand the nologin flag? When "shutdown -r +5" for example is called then the system will shutdown in 5 minutes, will "wall" write-all logged in users every minute nagging them that the system is rebooting in a moment, and will touch /run/nologin as a flag file. | 06:48 |
rwp | If {/var,}/run/nologin exists then no one can log into the system during that time when the system is waiting to reboot. Enforced by PAM (plugable auth modules) IIRC. | 06:49 |
rwp | The /etc/init.d/rmnologin start removes that file for the configuration that /run is mounted on the file system. These days /run is a tmpfs and any files there will evaporate at reboot. But before it was a tmpfs then rmnologin would clear it at boot time. | 06:50 |
cousin_luigi | rwp: No, it's for PCEngines SBCs | 06:51 |
rwp | Honestly this package you are re-working started out with some pretty questionable things in it. | 06:51 |
cousin_luigi | rwp: Oh that. Makes sense. | 06:51 |
cousin_luigi | rwp: I agree. It was messy. | 06:51 |
rwp | I don't know why voyage-util would need to call /etc/init.d/rmnologin start itself as it should have been called already. | 06:52 |
cousin_luigi | One would have to look at the original custom system as it was, I suppose. I don't think the package has everything. | 06:53 |
rwp | Plus normally packaged scripts like postinst and logrotate and such should call init scripts using the invode-rc.d utility, which respects the policy-rc.d configuration present, required for when everything is running in a chroot. You don't want to have daemons started in the chroots unless explicitly desired. | 06:53 |
cousin_luigi | rwp: Sorry, what chroot? | 06:54 |
rwp | If you install packages into a chroot and then upgrade then by default package postinst scripts will restart daemons, and possibly start daemons, even in a chroot, which is not desired. | 06:55 |
rwp | For example if you debootstrap a chroot and then upgrade the chroot you would find that cron is running in the chroot, perhaps surprisingly. | 06:56 |
cousin_luigi | I see. | 06:56 |
rwp | So there is the policy-rc.d layer to tell invode-rc.d not to run init scripts. | 06:56 |
rwp | See: /usr/share/doc/init-system-helpers/README.policy-rc.d.gz | 06:57 |
rwp | But here one init script is calling another init script and so it is really throwing all previous strategy out the window at that time. I think it needs to not do it. | 06:57 |
rwp | And I must relocate. As they say I don't need to go home but I can't stay here. BBIAB maybe. | 06:58 |
cousin_luigi | Bye and thanks. | 06:58 |
cousin_luigi | https://gitlab.com/lbaldoni/voyage-util/-/blob/main/debian/rules?ref_type=heads#L15 <- I don't understand this. What do the last commands inside the double quotes do, exactly? | 15:41 |
bb|hcb | cousin_luigi: man dh_installinit | 16:01 |
bb|hcb | Those are parameters for update-rc.d | 16:01 |
cousin_luigi | I assumed as much, but I couldn't reconcile them with what update-rc.d takes as input | 16:02 |
cousin_luigi | will read more about it | 16:02 |
DelTomix | in man update-rc.d you will notice "Older versions of update-rc.d also supported start and stop options. These options are no longer supported, and are now equivalent to the defaults option." | 16:03 |
cousin_luigi | oh | 16:04 |
cousin_luigi | silly me | 16:04 |
bb|hcb | I have no idea what this package is about, but if you set the proper LSB headers, then removing the override will make what you want | 16:05 |
cousin_luigi | bb|hcb: The package is about making me want to hit the wall with my forehead. | 16:06 |
cousin_luigi | Should rename it selfharmd. | 16:07 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!