[00:02] erm, I get 'debugshell just booted' but I have no keyboard input.. [00:04] yeah, debugshell looks broken to me.. "sh: can't access tty; job control turned off" [00:08] cjwatson: what would be the preferred way to run a command inside /target? do I need to chroot /target; command(s) or...? [00:10] that's not broken [00:10] the can't access tty message is just a random warning, ignore it [00:10] well, ok, no keyboard input is broken :) [00:10] that might be due to using a USB keyboard [00:10] I'd like to ignore it if my keyboard input would work [00:11] if so, plug in a PS/2 keyboard if you can [00:11] i doubt I have one [00:11] generally use 'chroot /target command' (NO SEMICOLON) unless the command uses debconf; if the latter is the case then talk to me [00:11] well, it's possible to build a new initrd but that's a whole lot more work [00:12] the reason I ask about the chroot is it would likely be simple to just run ufw commands directly to allow for the services in kickstart [00:12] if its not possible I'd have to do a bunch of sed nonsense.. [00:32] cjwatson: ok, I've updated firewall.sh to use "chroot /target ufw allow ". I notice you have --high, --medium, --dhcp listed as well but I don't see those on the kickstart documentation as valid? [00:40] ubiquity: cjwatson * r2679 ubiquity/ (bin/ubiquity-dm debian/changelog): [00:40] ubiquity: Add ubiquity-dm implementation for matchbox-window-manager (thanks, [00:40] ubiquity: Patrick Hétu; see LP #159363). [01:01] cjwatson: I'm going to call it a day. I sent you my changes to firewall.sh for testing. [01:08] thanks [01:08] you can't run ufw in the handler directly, though you can register a post_firewall function [01:09] (because the handlers themselves are run long before /target exists) [01:29] debian-installer-utils: cjwatson * r656 ubuntu/debian/changelog: releasing version 1.57ubuntu1 [02:13] cjwatson: that makes sense. likely based on my limited knowlege of how it all falls together [08:05] evand, when --automatic is invoked, ok_handler doesn't get called at all right?, just the run(self,priority,question) of each page === ebel_ is now known as ebel [13:00] evand, we have collected quite a stack of fix committed, can push fixes into the releases? [13:01] plan being that we release first fixes external to wubi, then we test wubi on new ISOs, then we get approval for new Wubi [13:01] here is a list of relevant non-wubi fixes http://paste.ubuntu.com/10907/ [13:04] the ones left out are 225593 and 222546 [13:04] to be addressed by myself/seb128 and lamont respectively [13:56] cjwatson: ok, got your email and I'm revisiting my obvious mistake. [14:14] superm1: correct [14:14] xivulon: ack'ed [14:15] thx [15:42] hi gang - i'm writing a package. ideally, i'd like the postinst to create/populate a mysql database. this deb works fine, *except* when it gets run via my preseed.cfg file. any guidelines/clues/hints on how to handle this? [15:59] cjwatson: I notice in auth.sh you're rewriting a conf file which is put in place in the POSTPOOL. Considering I'm trying to run commands what is your suggestion on handling it? [16:00] POSTSPOOL is just for the early handler to leave notes for the post-handler later on [16:00] you can run whatever you like in the post-handler [16:03] I guess my difficulty is not seeing how/where those tie together. [16:11] Zelut: basically the early handlers run before just about anything else in the installer; the post-handlers run from finish-install right at the end [16:13] cjwatson: I'm using auth.sh as an example, you create $POSTPOOL/auth.handler and then populate the files with the variables and then dump those into /target in auth_post() [16:15] cjwatson: assuming I'm reading it right that is.. If I were to follow a similar setup is it as simple as doing a test? if [ -f $POSTPOOL/firewall.handler/ssh ]; then run ufw command? [16:16] cjwatson: basically create the dir and files in the case statement, as is done in auth.sh, and then in the _post() check for those files and run the commands as needed? [16:18] sure, that sort of approach would be fine [16:18] you'll need to spell $POSTSPOOL correctly though ;-) [16:19] also it's a good habit to always quote shell variable expansions unless you have a good reason not to, so [ -f "$POSTSPOOL/firewall.handler/ssh" ] [16:19] of course. I take many code-shortcuts in IRC [16:20] I just say 'cos it's a common thing I end up correcting in patches, that's all :) [16:20] no offence intended [16:21] ok, I think I have the hang of it.. let me give it another try. [16:30] cjwatson: best way to handle errors? ie; ufw enable || echo "something broke". send via logger? [16:34] feeding errors to the warn function is about the best you have :-/ [17:04] cjwatson: ok, just about done.. do I still need to run commands in chroot /target? [17:05] yeah [17:25] cjwatson: ok, let me send you another revision. [17:49] any help on the mysql at install time question? [17:49] i've googled myself to death on this. [17:50] HopsNBarley: (I have to go in a minute, but) posting /var/log/syslog from the installation, along with your preseed file minus any passwords, would help [17:50] /var/log/installer/syslog if you have a completed installation post-reboot [17:53] is there a preferred pastebin for this room? [17:53] cjwatson, thanks. [17:54] HopsNBarley: Nope, any pastebin will do. [17:56] I normally use paste.ubuntu.com [17:57] great. pastbin.ca just killed my firefox! [17:57] * HopsNBarley starts over. [17:58] here's the error message in the syslog: in-target: : Can't connect to local MySQL server through socket '/var/run/mysqld/mysqld.sock' (2) [17:58] just prior to this, the postinst ran: invoke-rc.d mysql start [17:59] my intuition is that the chrooted environment isn't yet completely made enough to run mysql. [17:59] missing /proc, for example? I don't know yet. [17:59] Again, this deb works fine "normally" outside of the installer. [18:01] oh, right, I understand better now [18:01] the installer deliberately disables starting up services [18:02] okay, that seems a reasonable thing for it to do... so what's the workaround? surely some other packages have this issue. [18:02] since normally it isn't appropriate [18:02] actually, I've not encountered this before ... [18:03] are you just feeding this into pkgsel/include? [18:03] what if it moved installation of the package to the late_script? [18:03] late_script? [18:03] HopsNBarley: I've done a hack before where I've added a script to be called in rc.local so it finished on reboot.. [18:03] likely not the best solution however [18:04] so, if you did it in preseed/late_command, then that would work provided that (a) you don't need to interact with debconf, (b) you start up the mysql server manually and tear it down again at the end, and (c) you use chroot /target rather than in-target [18:04] cjwatson, the package is added like so: d-i pkgsel/include string debconf-utils openssh-server apache2-mpm-prefork myapp [18:04] otherwise Zelut's solution is actually not all that unreasonable (though certainly not elegant :-) ) [18:06] cjwatson, could you briefly explain (a) ?? [18:06] if it needs explaining, it doesn't apply to you [18:06] (-; [18:06] lol [18:06] got to go [18:07] hey, THANKS! [18:07] one more, while ya'll are on the line: d-i base-installer/excludes doesn't seem to work. i saw a bug on this. [18:08] this is gutsy. [18:12] thanks also Zelut. [19:38] cjwatson: do you have a list of completed kickstart compatible items? I'm thinking about seeing what else I can knock out for Intrepid. [21:56] evand, xivulon: hi, I am done with bug #136682. [21:58] hi tormod, yes I have already told evand early on today :) [21:58] Zelut: ISTR I put a list of incomplete bits in the installation guide [21:58] it's in the list of the fixes that require approval and hopefully will be in new ISOs soon [21:59] thanks again, that was really helpful [21:59] HopsNBarley: err, there's no such question ... [22:02] HopsNBarley: as in, doesn't work because it doesn't exist :-) What documentation recommended that?