| === apachetransit is now known as udslogger | ||
| ion | o hai | 15:34 |
|---|---|---|
| === halflin__ is now known as halfline | ||
| === Stevee_ is now known as Stevee | ||
| Keybuk | hai | 15:54 |
| Stevee | hello scott | 16:02 |
| Keybuk | hello | 16:05 |
| Stevee | yes, I know I'm nerving you, but I have to ask once again: Is there any kind of release date of the next major version of upstart, or some kind of possible periode ( for example this year, or spring next year ... ) | 16:09 |
| Keybuk | it's a "when it's ready" | 16:14 |
| Keybuk | the new release changes many fundamental things in the core, in an attempt to correct all of the problems I've had (and others have reported) | 16:14 |
| Keybuk | and the goal is that it'll stablise to be 1.0 | 16:14 |
| Keybuk | obviously I don't want to change the core again after this | 16:15 |
| Keybuk | so I'd rather release when ready than release on a given time, and then upset the apple cart all over again | 16:15 |
| jds2001 | so i have a usecase with sysvinit that i'd like to duplicate with upstart | 16:16 |
| Keybuk | sure | 16:16 |
| Keybuk | would love to hear it | 16:16 |
| jds2001 | it's the concept of a "firstboot" runlevel. | 16:16 |
| Keybuk | does sysvinit have such a thing? | 16:16 |
| jds2001 | we install an inittab with runlevel 9 as initdefault | 16:16 |
| jds2001 | runlevel 9 does stuff, and cleans up after itself. | 16:17 |
| Keybuk | ah, I see | 16:17 |
| jds2001 | I've defined a runlevel 9 in /etc/init, but when I boot, it just says "invalid runlevel" | 16:17 |
| Keybuk | I do something quite similar with Upstart for boot testing | 16:17 |
| Keybuk | an /etc/init/foo.conf runs, and as a final step, deletes itself | 16:17 |
| Keybuk | the invalid runlevel is a whole Debian/RH thing - Debian's sysvinit never supported 7,8,9,a,b,c runlevels afaik | 16:18 |
| Keybuk | that was a quirk of RH's | 16:18 |
| jds2001 | here's the thing - i dont want anything except rc-sysinit to run, then run this, then clean up and reboot | 16:18 |
| Keybuk | (Debian instead gives you 3,4,5 to play with for your own, as 2 is the default) | 16:18 |
| Keybuk | anything? | 16:18 |
| Keybuk | you don't even want filesystems mounted? | 16:19 |
| Keybuk | or udev started? | 16:19 |
| jds2001 | rc-sysinit takes care of that, right? | 16:19 |
| jds2001 | this is rhel6, btw | 16:19 |
| Stevee | thanks for the reply Keybuk, I'm just wondering last rel of a version i quite a half year ago, after that some patches and it's hard / impossible to follow the current development process ... | 16:19 |
| Keybuk | jds2001: I dont know about RHEL, I'm afraid | 16:22 |
| Keybuk | are they still using Upstart? I'd heard they went to systemd | 16:22 |
| Stevee | I love upstart it's a relly good init system but the current version looks like unfinished, especially if you aren't using the compatiblity mode | 16:22 |
| Keybuk | but in Ubuntu, Web OS and Chrome OS (the three major Upstart distros) - everything is done as Upstart jobs, rc-sysinit is only there for backwards compatibility with older packers and LSB things dropped in | 16:22 |
| Keybuk | Stevee: there isn't a compatibility mode? | 16:23 |
| Keybuk | Ubuntu has had an entirely native boot for three releases now | 16:23 |
| jds2001 | Keybuk: yeah, it's not native. in rcS.conf it runs the old rc.sysinit | 16:24 |
| jds2001 | and has a 'start on startup' in it. | 16:24 |
| Stevee | yes i know, but i mean, without any runlevels - full event based ... | 16:24 |
| Stevee | .. thats somethimes really difficult to handle | 16:24 |
| Stevee | fedora scheduled to swith to systemd with fc14 but dropped that and now will do that with fc15 | 16:26 |
| Stevee | so they are still using upstart | 16:28 |
| jds2001 | and rhel6 will use it for it's lifetime as well. | 16:31 |
| jds2001 | sadmac: you around by chance? | 16:40 |
| sadmac | jds2001: hey | 16:40 |
| jds2001 | ltnt :) | 16:40 |
| jds2001 | figured you'd be a good point for my rhel6 query :) | 16:40 |
| jds2001 | i need something to execute after rc.sysinit, prior to anything else, which will do stuff, clean up after itself, and reboot. | 16:41 |
| jds2001 | sort of a "firstboot" type thing. | 16:41 |
| jds2001 | and I'm having a heck of a time figuring out how to do it. In rhel5, we just used an initdefualt to runlevel 9, whcih seems not to work in upstart :( | 16:42 |
| sadmac | jds2001: yeah I don't think we retained the 789 quirk as Keybuk says. I never saw a patch to add that back | 16:42 |
| jds2001 | yeah, any other way to accomplish similar goals? | 16:43 |
| jds2001 | i.e. what event happens after rcS.conf that I can 'start on'? | 16:43 |
| jds2001 | i see that as being a task, and at the end it would delete itself and reboot. | 16:43 |
| jds2001 | or am i completely on crack? | 16:44 |
| sadmac | jds2001: you could make it start on starting rc | 16:44 |
| sadmac | jds2001: which would block the normal runlevel from getting started and run your task instead | 16:45 |
| sadmac | jds2001: but it'd resume booting afterward, so you'd have to reboot at the end of your task | 16:46 |
| jds2001 | sure, we do that anyway | 16:46 |
| sadmac | jds2001: also it'd do this every time regardless of runlevel, so you'd have to have it detect if it was supposed to be running and quit if it wasn't. | 16:47 |
| jds2001 | i thinik if it's there, it should run. | 16:48 |
| jds2001 | we do that today - if it doesnt get cleaned up, the initdefault is still 9 | 16:48 |
| sadmac | jds2001: that works too | 16:48 |
| jds2001 | that way if it gets b0rked, we know :) | 16:49 |
| jds2001 | i'll play around and let you know if I have problems.... | 16:50 |
| sadmac | k | 16:50 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!