=== robins is now known as robinsmidsrod [14:32] any core upstart developer here? [16:42] wraiden: you should just ask your question.. somebody smart enough to answer should be on sometime today. :) [17:01] i'll write an email to the list and are allready up to implement the things i wanted to get opinions on... [17:02] i need to set the signal that is uses to stop the exec job ... [17:02] for slapd (openldap) [17:03] it needs sigint and will eat the directory-database if it gets stopped by sigterm or sigkill... [17:03] i'll implement a stopsignal stanza ... [17:04] default value for it is term [17:05] but can get be set int, usr1 or usr2 also [17:05] wraiden: right, in that case, pre-stop is needed [17:05] pre-stop ? [17:05] no need for a new stanza [17:06] send your sigint in pre-stop [17:06] and wait for slapd to shutdown, then TERM will never be sent [17:06] ah [17:07] and upstart will supervise the slapd pid and will not respawn it? [17:07] There is an open bug against upstart to provide custom actions, so that reload could be changed from HUP to USR1 or something like that [17:07] jepp [17:07] yes, at that point, the job's goal will be "stop" [17:07] so no respawn takes case [17:07] i plan to implement that also [17:07] takes place [17:08] but that needs a change in initctl to hand over the actual signal sending to the job to upstart itself [17:08] initctl doesn't parse the job configs so there is no way to change the behavior without that... [17:09] I'm not sure the best way to spin and wait for it to die.. its not pretty but just watching status would work I *think* [17:09] mhm [17:11] i'll go and test that ... [17:11] btw: the stopsignal stanza is a rally low hanging fruit [17:12] *really [17:12] ot: is the copyright / legal stuff still needed to contribute to upstart? [17:32] wraiden: not OT at all, but its not clear either. ;) [17:33] wraiden: stopsignal seems rather.. coarse. I'm thinking Keybuk will prefer something a bit more elegant. Like 'action stop signal INT' rather than 'stopsignal' [17:37] i've implemented "oom score value" support (patch is on mailinglist) so thats no problem as im familar with the stanza parsing now ... :) [17:39] an rfc mail for scott to read is also out to the mailinglist ... i'll wait until he has time to respond to it... [17:39] *reguarding the stopsignal stuff [17:40] anyway, time to go home for me now... bye [17:41] and thanks for your hint with the pre-stop hack for slapd :) [18:54] hello again... [19:20] wraiden: hi [19:31] wraiden: I think Keybuk doesn't require copyright/legal stuff since yesterday anymore, but best he answers that himself ;) [19:31] @ keybuk: Mailinglist... [19:31] ah, you asked there too [19:32] wraiden: you have a question? [19:32] in the mail ... [19:32] sorry, I don't have lots of this channel, so am missing context [19:32] * wraiden == Marc [19:32] ah, hi! [19:32] * Keybuk = Scott [19:32] :) [19:33] I did just reply to your mail [19:34] wraiden: did you want to follow up on the copyright assignment bit? it seems there is a confusion there from JanC's comment [19:38] mhm, if it's not needed anymore... [19:38] so let me clarify [19:38] JanC was in the channel before i made my way back home from work and i asked it here ... [19:38] Canonical Ltd require Copyright Assignment for projects under their stewardship [19:39] so i think JanC yust wanted to let me know [19:39] since I was employed by Canonical Ltd for the past seven years, that policy applied to Upstart too [19:40] however I left Canonical in January, and now work for Google Inc, who do not require copyright assignment [19:40] so the 1.0, 1.1 and 1.2 releases have contained patches for which copyright was not assigned by the authors [19:40] including myself, and google [19:40] ah, ok, they allowed you to take the project with you?! [19:40] they released the project under the GPL [19:40] german law is crazy about such things... [19:41] jeah [19:41] but the copyright... [19:41] they don't need to have "allowed" me [19:41] ok [19:41] the GPL is clear on that [19:42] ok [19:42] so I continue to maintain Upstart [19:42] for the past six weeks, I was in discussions with Canonical about an appropriate way forward [19:43] but we had agreed a time limit (end of April) for those, and that was reached without an agreement [19:43] so things revert back to Upstart continues as a GPL project without any form of assignment [19:43] (one of the options was a relicense to Apache 2.0 so Canonical could continue to release private versions) [19:44] good to hear that [19:46] so yeah, don't worry about it ;-) [19:47] some of Lennarts FUD blogpost doesn't state the right information than... [19:47] *shock* [19:53] pretty onesided function mentioning in his post as he doesn't looks out of his sandbox and just looks what of systemds functions were implemented in other inits... [19:55] *writing kill signal code* [19:56] it's Lennart, I would expect nothing less [20:08] yeah [20:08] got that impression the day he introduced his audio stuff to gnome... [20:19] pulseaudio is mostly useful (now) ☺ [20:27] yeah, now... [20:32] first hack send to list now [20:33] got rid of the signal == 0 then send sigkill as its clearer to just send SIGKILL as param in system_send_signal... [20:34] *reading on xen-devel list* [21:05] testsuite errors on my system on this: [21:05] ...with error after modification of job [21:05] BAD: wrong value for file, expected (nil) got 0xeb2a10 [21:05] at tests/test_conf.c:709 (test_source_reload_job_dir). [21:05] FAIL: test_conf [21:06] expected? [21:07] it's totaly unrelated to my changes as i tested an upstream version and get the error there also... [22:50] * wraiden goes and gets som sleep... [22:50] ... updated patch is out to the list now...