LeePen | Morning. Can anybody tell me if and how /etc/machine-id is created on Devuan? | 11:14 |
---|---|---|
hightower2 | LeePen, I am on Devuan and at least in my installation, /etc/machine-id is not there. (Package lsb-base is installed as well, if it for your question matters) | 11:15 |
hightower2 | if it matters for your question* | 11:16 |
KatolaZ | LeePen: machine-id is created by a systemd package | 11:17 |
KatolaZ | IIRC | 11:17 |
KatolaZ | LeePen: why do you need that? | 11:17 |
LeePen | I am trying to sort out why some of the elogind tests fail. I wold like to enable them for build | 11:23 |
LeePen | They rely on /etc/machine-id. I think we should use /var/lib/dbus/machine-id instead? | 11:23 |
LeePen | s/wold/sould/ | 11:24 |
LeePen | s/sould/would/ !!!! | 11:24 |
LeePen | hightower2: thanks, yes I have it on some of my Devuan boxes but not others, so they could be remnants of systemd infecton before I migrated away from debian. | 11:52 |
KatolaZ | LeePen: I guess we shouldn't use a machine-id at all, if possible | 12:17 |
LeePen | KatolaZ: Yes but elogind needs it and depends on dbus, so /var/lib/dbus/machine-id should be OK for this case? | 12:19 |
KatolaZ | what generates it? | 12:20 |
KatolaZ | and why is it needed? | 12:20 |
KatolaZ | ok it's generated by systemd-machine-id-setup | 12:22 |
KatolaZ | now, why is it needed at all? | 12:22 |
LeePen | dbus generates it on install and from the initscript with dbus-uuidgen | 12:23 |
* kilobyte can think of only downsides of a persistent unique machine id, unless explicitely configured by the admin as such (like a hostname). | 12:23 | |
LeePen | That is /var/lib/dbus/machine-id. I agree we can do without /etc/machine-id | 12:24 |
kilobyte | so putting a machine-id in run and generating on boot would be good enough for systemd compat | 12:24 |
kilobyte | s,run,/run, | 12:24 |
KatolaZ | LeePen: does dbus need it at all? | 12:26 |
KatolaZ | I would definitely prefer to have no unique, persistent machine-id in Devuan | 12:26 |
KatolaZ | of any kind | 12:26 |
LeePen | KatolaZ: I think so, but just its own version in /var/lib/dbus | 12:27 |
KatolaZ | so I agree that if it is needed, we should re-geerate it at every boot | 12:27 |
KatolaZ | (to say the least) | 12:27 |
KatolaZ | we have forked dbus | 12:27 |
KatolaZ | so we can act there as well | 12:27 |
onefang | Or, I could pull my finger out and write dCar, a lighter and faster dBus. | 12:28 |
LeePen | KatolaZ: That is fine. At least in terms of being able to build the | 12:29 |
LeePen | elogind tests and have them pass in a chroot we will need to ensure | 12:29 |
LeePen | dbus will create the one-time /run/machine-id in a chroot too. | 12:30 |
KatolaZ | LeePen: dbus actually does that when installed | 12:33 |
KatolaZ | in dbus.init | 12:33 |
KatolaZ | it created a machine-id under /var/lib/dbus | 12:34 |
LeePen | Yes, via dbus-uuidgen. I think I said that ;) | 12:34 |
KatolaZ | no in dbus.init | 12:35 |
KatolaZ | yeah in the end is dbus-uuidgen | 12:35 |
LeePen | So all we need for it to be non-persistent is delete it at dbus.init stop and it will be recreated next time. | 12:37 |
KatolaZ | maybe | 12:40 |
KatolaZ | it looks like it is re-generated anyway during dbus startup | 12:41 |
KatolaZ | LeePen: at least in Devuan | 12:42 |
LeePen | Yep. Thanks all -- helpful as always. | 12:42 |
KatolaZ | but it looks like 1.12.10 from debian has exactly the same | 12:43 |
LeePen | I have bugged elogind upstream to use /var/lib/dbus/machine-id rather than the systemd file in /etc | 12:43 |
KatolaZ | so if elogind needs it it could just put a symlink to /var/lib/dbus | 12:43 |
KatolaZ | ok | 12:43 |
KatolaZ | even better | 12:44 |
KatolaZ | much better | 12:44 |
LeePen | I notice on my vairous systems, that some have /etc/machine-id. I imagine it is left over from a debian installer systemd boot. We might want to clear those out in Devuan? | 12:45 |
KatolaZ | well, I guess so | 12:47 |
KatolaZ | but I am not sure which package should be in charge of the removal, TBH | 12:47 |
LeePen | Yes, it is one of those weird unowned files. | 12:48 |
drawkula | systemd.postrm? .prerm? | 13:01 |
drawkula | systemd.postinst creates /etc/machine-id | 13:01 |
drawkula | so removing systemd should clean ist up... but who would want to remove systemd? | 13:02 |
drawkula | mwhuahaahahahaaa | 13:02 |
LeePen | elogind upstream have accepted the patch already! :) | 13:37 |
KatolaZ | great LeePen :) | 13:40 |
KatolaZ | OmegaPhil: are you working on udisks2? | 14:54 |
OmegaPhil | KatolaZ: No, sorry, I've had a singificant change to what I do in life | 19:50 |
OmegaPhil | So effectively not active with computing stuff | 19:50 |
KatolaZ | ok OmegaPhil | 19:51 |
KatolaZ | thanks | 19:51 |
OmegaPhil | Took me by surprise | 19:51 |
KatolaZ | di you mind if I have a go at the package then? | 19:51 |
OmegaPhil | sure | 19:51 |
KatolaZ | so you do mind? | 19:51 |
OmegaPhil | sorry, go ahead | 19:51 |
KatolaZ | ahahahah | 19:51 |
KatolaZ | OK | 19:52 |
KatolaZ | thanks | 19:52 |
KatolaZ | I will leave you as maintainer for now :) | 19:52 |
OmegaPhil | Maybe my life will turn around again | 19:53 |
Centurion_Dan | o/ | 20:56 |
KatolaZ | Centurion_Dan: o/ | 20:59 |
KatolaZ | will be afk for a few hours | 20:59 |
KatolaZ | built several packages | 20:59 |
KatolaZ | see devuan-dev | 20:59 |
KatolaZ | syl | 20:59 |
KatolaZ | o/ | 20:59 |
Centurion_Dan | I see but you killed the ppc64el for sysvinit? | 20:59 |
Centurion_Dan | was it offline? | 21:00 |
Centurion_Dan | \o | 21:00 |
gnu_srs2 | Centurion_Dan: Where is consolekit2, experimental? | 21:00 |
gnu_srs2 | source package preferrably | 21:01 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!