[12:56] <lamont> on my machine, the hfs+ filesystem is autodetected and mounted.  On my daughter's machine, it isn't.  But I can manually mount it.  Why is that, huh?
[12:56] <lamont> 2.6.12-10-amd64-generic vs 2.6.12-10-686
[12:57] <lamont> her's a upgraded-over-time-from warty machine, mine is a virgin breezy install.
[12:57] <mjg59> Christ knows
[12:57] <mjg59> Almost certainly not my fault :)
[12:57] <lamont> mjg59: do you know the official sequence?
[12:57] <mjg59> Nope
[12:57] <lamont> mjg59: true, but ENOPITTI
[12:58] <lamont> and you know all, so, um, ....
[12:58] <Burgwork> mjg59, should my laptop be automounting  my NTFS partition?
[12:58] <mjg59> Are you in the same groups?
[12:58] <lamont> well, I'm in the same groups on both machines, and it doesn't mount for me there either
[12:59] <lamont> I freely admit that I did jack around a few groups on her machine at one point... that could be what I'm actually fighting
[01:02] <lamont> hrm.. no seb128 either
[01:03] <Burgwork> lamont, they are all on vacation
[01:04] <zul> lamont: you're screwed ahahah...er...
[01:04] <lamont> I suppose I could boot a livecd and see if that works...
[01:14] <ds> does anyone know why cups needs it's own authentication domain?
[01:46] <hunger> HurricaneJ: Install debfoster and do sudo debfoster.
[01:47] <hunger> HurricaneJ: That will ask about all installed apps and remove anything not needed anymore as well.
[02:48] <robertj> wow...xchat-gnome is...different
[02:49] <apokryphos> oh?
[02:49] <Amaranth> xchat-gnome > *
[02:49] <Amaranth> takes a day or so to get used to, but i don't want to go back
[02:50] <robertj> there are two features I am missing and I think either way I would need to write my own plugin
[02:50] <robertj> first is a channel grep and the second is a last x lines from nick
[02:52] <robertj> much changed since .8?
[02:52] <Amaranth> dunno, think the last i used was .7
[02:52] <Amaranth> i'm stuck on OS X right now
[02:52] <robertj> Amaranth: I get stuck there alot too
[02:52] <FireRabbit> anyone around farmiliar with setting up APT repositories?
[02:52] <Amaranth> robertj: haha, not like that
[02:53] <robertj> what is sad is that I often end up running ubuntu in VirtualPC
[02:53] <Amaranth> robertj: Airport Extreme
[02:53] <robertj> oh
[02:53] <Amaranth> and the open source driver still sucks
[02:53] <robertj> I have to baby sit some OS X apps at work occasionally
[02:53] <robertj> VirtualPC aint bad if you are patient, playing minesweeper, and have a dual machine with 2 gigs of ram
[02:53] <Amaranth> haha
[02:54] <Amaranth> G4 1.42Ghz 512MB
[02:54] <robertj> Forget about it
[02:54] <robertj> painful
[02:54] <robertj> took me 2 days to install Windows and Office + patches
[02:57] <ds> er, why would a package depend on libc6-amd64?  (on i386)
[02:58] <crimsun> I only know of alsa-driver b-ding on it?
[03:00] <FireRabbit> ah nevermind, figured it out
[03:04] <robertj>  Aramanth: does Ubuntu have any sort of policy on "Special" folders. I've noticed that gnome-screensaver looks in ~/Pictures by default and DCC downloads default to a Downloads folder in xchat-gnome
[03:06] <robertj> I guess if you were going to be really all "integrated" and stuff it would be nice to have a gnome-chat option "Add to Address Book" and "Show joins/parts for only users in my address book in channels with more than X people."
[03:10] <Amaranth> robertj: Yeah, let's bloat the preferences with things most people don't care about (join/part thing). ;)
[03:11] <robertj> for the most part, I care about join/parts on 10 user channels but not on 100
[03:11] <robertj> but if they were in my address book I would always want to see them
[03:12] <robertj> but you don't need an option for that last part, it would just be a "don't show parts for strangers on channels with less than X people"
[03:13] <lifeless> Amaranth: bah, thats not a reason not to do it. 
[03:14] <lifeless> Amaranth: thats the 'if enough people dont want it, dont to it' meme, which is nonsense.
[03:14] <robertj> Effects gets my vote for the preference to vote off the island ;)
[03:14] <lifeless> Amaranth: ask instead 'how can we add this without harming new users'.
[03:14] <Amaranth> lifeless: No X users would be good.
[03:15] <Amaranth> lifeless: I can see "Show join/part messages for people in my address book."
[03:15] <robertj> Amaranth: I think its kinda stupid
[03:15] <robertj> of course you want to see those
[03:15] <robertj> and if you don't, too bad control freak
[03:15] <lifeless> Amaranth: what about just a right mouse click on a channel, then click 'off' on show join/part
[03:15] <lifeless> per channel, saved in the background, not part of massive preference chaos.
[03:15] <Amaranth> lifeless: But should it turn them all off or only unknown people?
[03:16] <robertj> that'd be fine too, do you still show the people known in your address book without asking?
[03:16] <lifeless> Amaranth: I'm not a ui designer, but thats a good question.
[03:16] <robertj> what about people on ldap sources...etc
[03:16] <lifeless> Amaranth: I'm simply saying 'not implementing because its hard is nuts'
[03:16] <lifeless> not implementing because its a 'bad idea' is ok.
[03:17] <Amaranth> lifeless: If only 3 people want it is it worth implementing, no matter how hard it is?
[03:17] <Amaranth> I could see hiding it in gconf if one of those 3 submitted a patch.
[03:17] <lifeless> Amaranth: if one of them does it, or is sponsoring it, and it wont harm anyone else, sure.
[03:17] <lifeless> hiding it in gconf makes it non discoverable, which just means more education and rants.
[03:19] <Amaranth> lifeless: If it's only in gconf people who want it know where to find it and people who don't don't have clutter in the preferences dialog.
[03:19] <marcin> Amaranth: what are you guys talking about?
[03:19] <Amaranth> marcin: Imaginary program, imaginary feature.
[03:20] <marcin> Amaranth: eog?
[03:20] <Amaranth> I wish colloquy and iTunes would stop spiking to 40% CPU each
[03:20] <robertj> no real program imaginary features
[03:20] <Amaranth> marcin: No...
[03:21] <Amaranth> why the fsck does an irc client need 10-40% CPU constantly?
[03:22] <marcin> Amaranth: switch to erc ;)
[03:22] <Amaranth> marcin: I already have a buggy OS, I don't need two.
[03:22] <marcin> erc is irc client in emacs
[03:23] <Amaranth> I know.
[03:23] <Amaranth> emacs is the second buggy OS
[03:23] <marcin> Amaranth: it needs maybe 1% CPU
[03:23] <marcin> Amaranth: I agree that it's buggy ;)
[03:24] <marcin> Amaranth: but it's nice that you call emacs - OS :)
[03:25] <Amaranth> not really
[03:25] <robertj> the notification on message in xchat-gnome is great too, but it doesn't have an option to show you the message as well
[03:25] <Amaranth> it's a polite way of saying it's bloated
[03:26] <robertj> I recommend XChat Aqua
[03:26] <marcin> Amaranth: :)
[03:26] <marcin> Amaranth: night
[03:41] <lifeless> Amaranth: what are you smoking ? gconf is -not- a place that people -know to look- to find features. 
[03:42] <lifeless> Amaranth: if gconf was so useful, no program would need preference dialogs at all.
[03:43] <jdub> gconf is just scaffolding
[03:43] <jdub> it is rarely the answer to anything usability or user interface design related
[03:44] <robertj> lifeless: gconf is ok for "I want this feature, and somewhere there may be one other person who does too" type stuff
[03:44] <psusi> gconf is like windows' ini file apis or registry... it provides a nice way for a program to save settings without having to reinvent the wheel
[03:44] <jdub> robertj: no, gconf is where (almost) *all* gnome configuration information is stored
[03:45] <robertj> jdub: I meant as a primary method for setting and unsetting options
[03:45] <jdub> robertj: it's not that, either
[03:47] <robertj> jdub: fine, gconf-editor is a suitable interface for setting some very eccentric settings intended exclusively for a small number of highly-technical individuals?
[03:47] <lifeless> robertj: I disagree
[03:47] <psusi> like windows' registry editor, it is not meant to be a primary method for users to change configuration options... the tool is only there to provide A way to do so... for power users and system administrators
[03:48] <lifeless> robertj: gconf-editor is a debugging tool for when preference setting mechanisms go wrong, it should never be the 'way' to enable or disable something.
[03:49] <robertj> Iifeless: some features are so unusual that either they don't go in or they are exposed exclusively through gconf because implementation time for a better gui exceeds the utility of actually putting them in
[03:50] <robertj> not to mention that user time is a resource that gets consumed looking through options
[03:51] <psusi> aye.... but that should only be for the most esoteric and technical options
[03:51] <robertj> psusi: which is what I just said no?
[03:51] <psusi> yes... I'm agreeing...
[03:52] <robertj> but really there are alot of features noone needs that we want to still provide access to because it doesn't really hurt anyone to expose them through the editor
[03:52] <robertj> "and that's about all I got to say about that"
[03:54] <psusi> is there anyone else I could have a talk with about gnome-volume-manager and pmount besides pitti? I've not seen him around lately
[03:54] <jdub> psusi: he's having a break - will be back soon enough
[03:55] <psusi> ahh... when will that be, and I guess that means the answer to my first question is no? ;)
[03:55] <zul> oh hey jdub 
[03:55] <jdub> psusi: not sure; you can always ask your question though
[03:57] <psusi> well... I'm trying to modify the hal information so that g-v-m and pmount will auto mount a read/write udf formatted cdrw in read/write mode via the packet writing interface... I have managed to get that to happen, only it doesn't show up on the desktop
[03:57] <psusi> it does get auto mounted correctly in /media though, and I can access it from the command line
[03:57] <psusi> so why would it not show up on the desktop?
[03:57] <tseng> the desktop only shows things that are user mountable
[03:58] <tseng> assuming you have show_volumes
[03:59] <psusi> it is user mountable... I modified the hal policy rules to change the block.device from /dev/hda to /dev/pktcdvd/pktcdvd0 so when I insert the disk, it does get auto mounted correctly... just doesn't show up
[03:59] <robertj> psusi: does it have the .created_by_pmount file?
[03:59] <psusi> in /media/cdrecorder?
[04:00] <robertj> yeah
[04:00] <psusi> when I umount it, yes
[04:02] <psusi> the really strange thing is... if I mount it by hand read only, it shows up
[04:04] <psusi> mount it read/write via the packet writing device, and it doesn't show up
[04:05] <stub> Launchpad going down in 15 mins for the regular updates, which will put the wikis into read only mode. Downtime estimated at 40 mins.
[04:05] <psusi> where's the code that notices stuff in /media and shows it on the desktop?  It looks like it doesn't like devices that aren't in /dev ( and not a subdir of it )
[04:11] <psusi> weird... moved /dev/pktcdvd/pktcdvd0 to /dev and now when I mount it, it shows up on the desktop... but under the name "cdrecorder" instead of the volume label of LinuxUDF... and the icon does not go away when I umount it
[04:12] <psusi> mount it again with /dev/hda insteaed of /dev/pktcdvd0 and LinuxUDF shows up on the desktop... so there's two icons for the same thing... only with different labels
[04:16] <zul> hey jeff
[04:16] <jbailey> Zooool.
[04:16] <jbailey> I'm playing with my new openwrtized linksys.
[04:16] <tseng> there is only zul
[04:16] <jbailey> Fitting Ubuntu on this is going to be a bitch.
[04:17] <tseng> jbailey: good luck with the kernel
[04:17] <jbailey> tseng: Yeah. =(
[04:17] <jbailey> tseng: Although ISTR someone was reverse engineering the wireless broadcom driver.
[04:17] <jbailey> So perhaps it will actually be possible soon.
[04:18] <jdub> jbailey: netboot, netboot ;)
[04:18] <SEJeff> tseng, Get with the times. There is only xul :)
[04:19] <jdub> totally acceptable configuration for a firewall! ;-)
[04:19] <jbailey> jdub: That's actually half the reason I wanted the openwrt.  My old gnet was interfering with netbooting my ltsp box. =)
[04:19] <jbailey> That and I want pure IPv6 on my local lan, but hey... =)
[04:20] <SEJeff> And I thought I had a lot of machines on my lan
[04:21] <SEJeff> How likely would vSecurity have of making it into universe (possibly main) for dapper?
[04:23] <jbailey> SEJeff: Things can still make it into universe no problem for another couple of weeks
[04:24] <jbailey> SEJeff: It needs that first, then a main inclusion report done.
[04:24] <SEJeff> jbailey, Would it be possible to make it into main though? Theoraticlly?
[04:24] <SEJeff> This is from ubuntu-hardened project
[04:24] <jbailey> SEJeff: Sure, no reason why note.  Things can be promoted to main even after version freeze.
[04:25] <SEJeff> jbailey, Excellent
[04:25] <SEJeff> I *really* want Ubuntu to have something like vSecurity and thats why I've been helping trulux test vSecurity
[04:25] <jbailey> I don't know anything about it.
[04:26] <SEJeff> jbailey, Do you want a quick rundown of what it can do?
[04:26] <jbailey> SEJeff: I think I'm a bit sleepy right now.  I might ask you tomorrow though. =)
[04:27] <SEJeff> jbailey: Please do. It's quite impressive once you see it work and nothing breaks
[04:27] <SEJeff> jbailey: I've been trying to break it the past few days without much luck
[04:27] <jbailey> Nothing breaks is always good advertising. =)
[04:28] <jbailey> I tried looking the other day to see if there was someway of setting capabilities from the filesystem rather than just using SUID/SGID, but it looks like it's still not possible.
[04:28] <SEJeff> jbailey, vsecurity includes cap_over
[04:28] <jbailey> psusi: Why is main important for that?
[04:28] <SEJeff> A way to add something like NET_RAW to /bin/ping
[04:28] <SEJeff> That way, you can remove suid root from ping and it still runs perfectly
[04:28] <psusi> jbailey, isn't there a wrapper program like sudo that you can use?
[04:28] <jbailey> SEJeff: Ah nice.  Does that just use extended attributes in ext3?
[04:29] <psusi> jbailey, well, main AND added to a seed like ubuntu-base... right now it's just in universe
[04:29] <SEJeff> jbailey, Extended attributes won't support that
[04:29] <jbailey> psusi: I can do it with a wrapper, but why?  CAP_NET_RAW for /bin/ping is an excellent example.
[04:29] <SEJeff> http://www.randombit.net/projects/cap_over/ This is the code it uses
[04:29] <jbailey> There's no real reason why you need a wrapper program that's SUID just to drop privs for ping.  It would be a bit excessive.
[04:30] <psusi> true
[04:30] <SEJeff> we work with the developer of cap_over
[04:30] <SEJeff> And the trusted path execution code written by IBM also works flawlessly
[04:30] <SEJeff> That would out of the box kill thinks like the lapper worm
[04:30] <jbailey> Ah, I hadn't seen cap_over before.
[04:30] <psusi> jbailey, untill it is in ubuntu-base or otherwise installed by default, hardware fakeraid won't be supported out of the box
[04:31] <psusi> and I'll have to debootstrap my system from a livecd again ;)
[04:31] <jbailey> Is that just using the same mechanisms that selinux uses?  I remember they integrated a bunch of them.
[04:31] <SEJeff> jbailey: Yes, lsm
[04:31] <SEJeff> vsecurity uses lsm extensively
[04:31] <jbailey> Cool.  Looks like something to add to my reading list.
[04:31] <robertj> ooh yummy nautilus patch for D&D on the places bar
[04:31] <SEJeff> there will be a new release out soon and trulux has been going crazy lately on polishing it up / fixing some things
[04:41] <psusi> does adam conrad hang out in here?
[04:42] <Amaranth> i'd need an irc nick ;)
[04:42] <Amaranth> sounds familiar...
[04:43] <psusi> Amaranth, you're Adam Conrad?
[04:43] <Amaranth> no
[04:43] <Amaranth> <--Travis Watkins
[04:43] <psusi> ohh... I don't know his nick.. that's why I'm asking ;)
[04:44] <dilinger> psusi: infinity
[04:45] <Amaranth> *facepalm*
[04:45] <zul> i think he is on vacation as well
[04:45] <zul> or asleep
[04:45] <jbailey> SEJeff: My only problem with these policy files is that it means that the metadata is separate from the file.  So I can't tar it up and untar it elsewhere and have it Just Work.
[04:45] <psusi> doh... that would explain why he hasn't responded to my dmraid bug
[04:45] <jbailey> SEJeff: Or mv the file or whatnot.  Do you know if someone is addressing that at all?
[04:46] <psusi> does tar understand extended attributes?
[04:46] <jbailey> Yes.
[04:46] <psusi> cool
[04:47] <jbailey> It can tar them up using the posix interfaces.
[04:47] <SEJeff> jbailey, I thought gnu tar didn't do extended attributes and that is what star was written for
[04:48] <SEJeff> jbailey, Think about how much of a PITA having to look at every file for attributes is
[04:48] <Amaranth> perhaps they merged?
[04:48] <SEJeff> vs using grep -r though /etc/cap.d/ or something like that
[04:48] <SEJeff> not sure
[04:48] <jbailey> SEJeff: Right, but star doesn't do anything that tar doesn't do anymore.
[04:49] <jbailey> SEJeff: tar was inactive for a long time.  It's now actively maintained.
[04:49] <SEJeff> jbailey, News to me
[04:49] <SEJeff> Very good to hear. I learn something new every day
[04:49] <psusi> I like tar's new incremental backup feature with a snapshot file... you can build tiered dumps with that
[04:50] <jbailey> SEJeff: I'm theoretically one of gnutar's co maintainers. =)
[04:50] <psusi> so you can do things like a full backup only once a month, then do smaller weekly backups, and even smaller daily backups... but to restore you only need to untar 3 files... the last monthly, the last weekly, the last daily... not every single daily since the last full backup
[04:51] <SEJeff> Good to know
[04:51] <psusi> and it supposedly notices things like deleted files in newer backups and won't restore them from older backups
[04:55] <zul> right night..
[04:55] <SEJeff> night
[04:55] <SEJeff> It's that time for me also. Another day converting hp-ux servers --> linux servers woohoo
[04:56] <jdub> SEJeff: a righteous mission
[04:57] <jbailey> SEJeff: On ia64 or hppa? =)
[04:57] <whiprush> howdy jeff.
[04:57] <whiprush> and jeff.
[04:57] <jbailey> whiprush: =)
[04:57] <jdub> name stealing bastardos!
[04:57] <SEJeff> jeff is a good name
[04:57] <zul> jbailey: troll
[04:57] <SEJeff> I hear that people named jeff are born brilliant
[04:58] <SEJeff> It's just what I hear
[04:58] <jbailey> zul: For which comment? =)
[04:58] <zul> either or
[04:58] <SEJeff> jbailey, PA-Risc, hppa stuff
[04:58] <SEJeff> Night guys
[04:58] <zul> 3/exit
[04:58] <jbailey> g'n SEJeff, Chuck.
[04:59] <whiprush> jdub: I got a cool email from someone that's totally jdub-esque. The process is still in its early inception, but I just have to send it to you because it rocks.
[05:00] <jdub> whiprush: none of what you said follows, but it sounds cool anyway... ;)
[05:01] <whiprush> jeff.waugh@ubuntu.com? (sorry, I am away from my normal machine)
[05:01] <jdub> yeah
[05:02] <whiprush> you've got mail!
[05:03] <jdub> yay mail
[05:05] <whiprush> jdub: Somehow I kind of feel that we should let you guys know some LoCo team stuff like that and get an "attaboy". Otherwise we end up just going to bars and drinking instead of doing real advocacy work. :)
[05:05] <jdub> haha
[05:06] <jdub> haven't got it yet
[05:06] <jdub> turning off sender address verification has resulted in 400 mails in the last 10 minutes
[05:06] <psusi> rofl
[05:06] <whiprush> So a few months back newforge ran this story about a jesuit school in detroit running ubuntu and ltsp.
[05:07] <whiprush> The mail I sent you is from a dude that works for Ford and wants to form a non-profit building ubuntu computers for disadvantaged youths and stuff.
[05:07] <jdub> awesome
[05:08] <whiprush> apparently he's meeting with the COO of Ford and working on getting sponsorship.
[05:08] <whiprush> so he goes onto the ubuntu wiki, and finds us local people.
[05:09] <whiprush> jdub: So this ties into "Getting jdub to come to the midwest US." plan ...
[05:09] <whiprush> so we're gonna help the guy out, etc. etc.
[05:11] <jdub> whiprush: did you see the latest poll?
[05:11] <psusi> do any of you guys have a microsoft optical intellimouse?  I can't believe I've only seen one other person besides me complain that xorg 7 has reassigned buttons 6 and 7 to 8 and 9 ( wtf? it's only a 7 button mouse! ) so back and forward don't work in firefox anymore
[05:11] <whiprush> no, I've been tied up in family business until today.
[05:11] <whiprush> need to catch up on fridge businesss.
[05:12] <psusi> isn't Ford broke right now? ;)
[05:12] <whiprush> no, that's GM.
[05:12] <whiprush> well, they're not as broke as GM anyway. :)
[05:12] <psusi> hehe
[05:12] <whiprush> jdub: ugh, north america isn't doing so hot in that poll.
[05:13] <jdub> whiprush: given that you've already had a go... ;-)
[05:13] <jdub> whiprush: mind you, look at my page on the wiki
[05:13] <whiprush> heh
[05:13] <whiprush> yeah, but you end up at the summit and 2 linuxworlds /anyway/ ...
[05:14] <whiprush> so maybe we can sneak in. :)
[05:17] <whiprush> jdub: Had you come to detroit, no one would have stolen your laptops. They would have just taken your wallet. :p
[05:18] <jdub> haha
[05:18] <jdub> whiprush: you should definitely look into the freegeek project in portland (and new ones that have popped up elsewhere in the USA)
[05:18] <jdub> whiprush: they're *extremely* well organised
[05:19] <jdub> whiprush: sounds like a natural fit for this idea in some ways
[05:19] <jdub> (just reading the email now)
[05:19] <whiprush> ah, mako was into that wasn't he?
[05:19] <whiprush> I recall the idea
[05:20] <jdub> don't think so
[05:20] <jdub> freegeek is a co-op refurbishing group
[05:20] <whiprush> I will look into that
[05:22] <whiprush> I'll send them a mail. At the least they probably know best-practices or something, which we know 0 about.
[05:24] <psusi> anyone here read "Understanding the Linux kernel" published by Oreilly?  picked it up the other day and it seems nice
[05:25] <dilinger> psusi: it's decent for writing drivers
[05:25] <dilinger> psusi: not really all that useful for core stuff.  robert love's book is much more thorough (well, the old edition; haven't read hte new one)
[05:26] <psusi> hrm... what's the title?
[05:26] <whiprush> rml's book is funny to boot. :)
[05:26] <psusi> this book seems to talk about the whole thing rather than just how to write drivers to me
[05:34] <mako> whiprush, jdub: i've spent some time at freegeek.. i know a few of the people there
[05:34] <whiprush> mako: do they have chapters outside of portland?
[05:35] <whiprush> I'm kind of in that situation, but it's probably best if I mail you about it when I get better details.
[05:36] <mako> whiprush: yes
[05:36] <mako> whiprush: one in PA and one in WI IIRC
[05:37] <whiprush> mako: cool and the gang. Someone locally wants to do ubuntu prebuilds. I'll mail you when I get more info.
[05:37] <mako> whiprush: please do, that sounds great
[05:37] <whiprush> mako: I'll send you the note the guy mailed me now.
[05:48] <jdub> whiprush: i visited there when i was in portland - it's really amazingly good. best model to use for something like this.
[05:48] <whiprush> this sounds excellent.
[05:51] <jdub> whiprush: (to the point where, with funding, it would be worth going there or shipping out a couple of the principals as a learning expedition)
[05:52] <whiprush> jdub: much to early to be thinking that. But yeah, long term that would be great.
[05:57] <shadeofgrey> hi everybody
[05:57] <shadeofgrey> im here because i dfound a very blatant bug in dapper
[05:58] <shadeofgrey> the option to make your mouse left handed - the checkbox will fill in properly and the icon will change and indicate that the buttons have been properly swapped, but it doesnt actually change the button assignments
[06:00] <jdub> elmo, Znarl: ping
[06:02] <seth_k|lappy> shadeofgrey, http://bugzilla.ubuntu.com
[06:09] <Amaranth> wow, people actually use left-handed mice?
[06:09] <seth_k|lappy> dunno, my brother is left-handed and doesn't
[06:10] <Amaranth> i'm left-handed and i don't
[06:12] <shadeofgrey> what do i assign as the "component" for the above bug?
[06:20] <nofear> hey gotta stupid question..
[06:21] <nofear> I am on a satilite connection and i got a idiot for a brother who downloads alot.. and I want to set up a box using ubuntu to shair internet connections but I want to cap downloads to a certain speed. 
[06:21] <nofear> is that Possible to do in ubuntu?
[06:23] <Amaranth> yeah, i think iptables can do that
[06:23] <Amaranth> it can do everything else :P
[06:29] <shadeofgrey> would oneof you kind people care to take a minute and file a bug report on my behalf if im very specific about the symtops/problems etc.? the interface online for filing bugs is very complicated and i dont know the answers to most of the fill in the blank spaces
[06:32] <crimsun> what prevents you from filing it?
[06:34] <shadeofgrey> a lot of fill in the blanks that i dont know the answer to like components and such
[06:34] <shadeofgrey> its not a bug with a specific program. its a big problem with the mouse applet under system --> preferences
[06:35] <crimsun> just fill in as much as you can
[06:35] <shadeofgrey> well
[06:36] <shadeofgrey> thats proving to be a problem as well...  because i cant make the buttons on my mouse left handed I cant even navigate the form well.  im very physically handicapped
[06:36] <shadeofgrey> i tried using the tabb key on my keyboard but the experimental version of firefox on dapper races through them when u use keyboard to skip forward and back in forms
[06:36] <crimsun> so is the Mouse preference dialog itself limiting you? If so, in what way?
[06:37] <shadeofgrey> the mouse pref3erence dialog is accepting the checkbox "make left handed mouse" which is immediately supposed to swap the buttons - but when you click okay and the applet closes the changes are never applied
[06:38] <crimsun> are you running current Dapper? I had that problem as well up until last week.
[06:38] <shadeofgrey> im using flight 2
[06:38] <shadeofgrey> thats the newest is it not?
[06:38] <crimsun> it's the latest snapshot, but your issue has been resolved in current Dapper
[06:38] <shadeofgrey> should i be downloading the latest nightly?
[06:39] <jdub> shadeofgrey: those are CD releases - latest dapper is always in the archive
[06:39] <crimsun> you just need to update and dist-upgrade
[06:39] <shadeofgrey> okay.. lemme boot into dapper and ill try
[06:40] <shadeofgrey> brb 5 mins
[06:40] <crimsun> either via Applications> Accessories> Terminal, or System> Administration> Synaptic Manager (choose Smart Upgrade)
[08:01] <jdub> slomo_: ping
[08:02] <jdub> slomo_: n/m
[08:38] <zakame> elmo: Please sync gpdf, libcurses-perl, libevocosm, libfwbuilder, libgnome2-wnck-perl, libpanelappletmm2.6, and nufw from Debian Sid, ok to override Ubuntu changes.  Thanks :-)
[12:42] <hunger> mojo: XP can do bridging just fine.
[12:48] <Mez> hunger: depends what you call fine
[12:50] <hunger> Mez: So "fine" that it used to do bridging by default, causing me trouble without end when I booted win up and suddenly I was bridging some WLAN into the management network at a customer's site:-(
[12:51] <hunger> Damn! I am on the wrong channel *AGAIN*. Sorry.
[12:52] <hunger> Someday I will figure out how this client displays channels.
[03:12] <zul> morning
[03:40] <zakame> wb Seveas 
[04:05] <zwnj> apt-get doesn't send NO-CACHE in HTTP request.  it causes that i have to force our cache to get Release and Release.gpg files with "wget --no-cache".  for security updates, it's a big mistake.
[04:07] <Treenaks> zwnj: you want apt to cache packages
[04:07] <Robot101> Treenaks: yes, but not package files
[04:07] <Treenaks> Robot101: those too
[04:07] <Treenaks> Robot101:  the caches should at LEAST do an If-Modified-Since: / ETag match on the originating server
[04:07] <Robot101> no, you really do want the proxy to talk to the server and find what the latest file is
[04:08] <Robot101> doing a HEAD with no-proxy is fine
[04:08] <Robot101> er, no-cache
[04:08] <zwnj> Robot101: yes
[04:08] <mojo_> gosh, are there any work around to fix locales problem? I want to fix the locales now so I can go on with other testing, ( I recieve err perl: warning: Setting locale failed.
[04:08] <mojo_> perl: warning: Please check that your locale settings:
[04:08] <mojo_>         LANGUAGE = "C",
[04:08] <mojo_>         LC_ALL = (unset),
[04:08] <mojo_>         LANG = "C.UTF-8"
[04:08] <mojo_> )
[04:12] <mvo> zwnj, Robot101: have you tried runing with "apt-get update -o Acquire::http::No-Cache=true" ?
[04:13] <Treenaks> mvo: that's not default?!
[04:13] <zwnj> mvo: no, i didn't know about this option at all!
[04:14] <zwnj> btw, i should try it later, when some Release files get update
[04:15] <mvo> Treenaks: no, default is sending a I-M-S 
[04:15] <Treenaks> mvo: sounds reasonable.. so the proxy/cache is broken then?
[04:16] <Treenaks> mvo: or the remote server
[04:16] <mvo> Treenaks: I don't know about cache used, if zwnj send me the output of "apt-get update -o Debug::Acquire::http=true 2> debug.log" I can have a closer look
[04:17] <mvo> it seems that quite a few cache are problematic :/
[04:17] <zwnj> mvo: sure, but it works fine now
[04:17] <mvo> zwnj: with No-Cache=true?
[04:18] <zwnj> mvo: i did wget --no-cache for all Release files with a bash script, so there's no problem now
[04:19] <zwnj> mvo: but before that, i had Release file for about two weeks ago
[04:19] <mvo> zwnj: what proxy/cache do you use?
[04:19] <mvo> or your provider?
[04:19] <zwnj> mvo: we have a global web cache in university
[04:20] <zwnj> we had the same problem with yup last year, and svidal patched yum to solve this
[04:20] <zwnj> ^yup^yum
[04:21] <mvo> zwnj: I would be interessted if the No-Cache=true switch helps. you can just make it default then. I still would be interessted in looking at the ouput of Debug::Acquire::http=true when you have the problem the next time
[04:22] <zwnj> mvo: sure, thanks btw
[04:22] <mvo> cheers
[04:22] <zwnj> mvo: but how i can set it as default just for Release/Release.gpg files?
[04:23] <mvo> *ick* right. that's a problem :/
[04:44] <verwilst> hellow!
[04:44] <Pygi> ho
[04:44] <verwilst> i seem to have a problem on a few servers running dapper
[04:44] <verwilst> with locales_2.3.7-2_all.deb
[04:44] <verwilst> that's being silly ;)
[04:44] <Robot101> verwilst: don't run your servers on dapper, you utter crack fiend
[04:44] <Robot101> verwilst: and, /topic
[04:44] <sivang> hey Pygi 
[04:44] <Pygi> I suggest you head over to #ubuntu-server
[04:44] <Pygi> hello sivang
[04:44] <Pygi> what's up? how holidays passed? :)
[04:45] <verwilst> oh
[04:45] <verwilst> :d
[04:45] <sivang> verwilst: you need to touch a dir, and they will set up cleanly. It's all due to locales transitiont as noted in Martin Pitt's email
[04:45] <verwilst> aah yes, Martin Pitt's email, how could i forget.. :d
[04:46] <verwilst> who's martin? ;) and on ubuntu-devel list i presume?
[04:46] <sivang> verwilst: well, dapper is under heavy development
[04:46] <verwilst> sivang: yeah i know :)
[04:46] <sivang> verwilst: if you want to use it, you would most probably need to read ubuntu-devel mailing list to be able to keep up
[04:46] <Pygi> verwilst: and still not ready for prime time :)
[04:47] <verwilst> Pygi: oh, but those servers will be in testing for some weeks ;)
[04:47] <verwilst> it'll probably be settled down a bit by then ;)
[04:47] <pitti> verwilst: what is 'a problem'?
[04:47] <pitti> sivang: the directory existence issue was fixed in -2 yesterday
[04:47] <hunger> Hey pitti.
[04:47] <pitti> :)
[04:47] <pitti> hi hunger
[04:48] <Pygi> hi pitti
[04:48] <hunger> pitti: Yeah, I'd duck and cover, too if I were in your place;-)
[04:48] <zakame> heya pitti 
[04:50] <hunger> pitti: I assigned another wishlist bug about mount.crypt to you yesterday. I hope that is OK... it is a oneliner and you need to touch that file anyway when taking care of the other one.
[04:51] <ogra_ibook> hey pitti 
[04:51] <sivang> pitti: Cool :) Howdy ? 
[04:52] <ogra_ibook> pitti, there were locale problems in dapper, did you already hear about them ? :P
[04:52] <sivang> ogra_ibook: hehe
[04:52] <hunger> strange... one problem I managed not to run into;-)
[04:53] <verwilst> pitti: hm, it's -2, lemme check the exact error
[04:53] <hunger> ls -alF
[04:54] <verwilst> Preparing to replace locales 2.3.6-3 (using .../locales_2.3.7-2_all.deb) ...
[04:54] <verwilst> dpkg: error processing /var/cache/apt/archives/locales_2.3.7-2_all.deb (--unpack):
[04:54] <verwilst> is all it says
[05:02] <zwnj> mvo: Acquire::http::No-Cache=true works, thanks
[05:03] <psusi> pitti: are you around?  I have some questions about pmount, g-v-m, and gnome-vfs if you have a second
[05:03] <pitti> psusi: at the phone now, will ping you back
[05:03] <psusi> sweet
[05:23] <verwilst> oh well,i filed a bugreport
[05:39] <pitti> psusi: sorry, I got disconnected; what's up?
[05:40] <psusi> pitti: ok... first a bit of background info... I've been messing with the udftools package to set up a cdrw for full read/write mounting
[05:40] <psusi> I got it working and it is nice, only when you insert the disk, it is auto mounted read only... so I decided to modify the hal policies to fix that
[05:41] <psusi> the problem is that to mount it read/write you have to mount it via /dev/pktcdvd/pktcdvd0 instead of /dev/hda
[05:41] <pitti> right
[05:41] <psusi> so I fixed the hal policy xml files to override block.device for cdrw media in that drive that has a udf filesystem on it
[05:41] <psusi> so now it does correctly get automounted read/write... only it does not show up on the desktop... I'm thinking there's something wrong with gnome-vfs?
[05:42] <pitti> how did you get pmount to mount /dev/pktcdvd/pktcdvd0 ?
[05:42] <psusi> I used the hal policy xml files to change block.device on that hal object for the media to /dev/pktcdvd/pktcdvd0 instead of /dev/hda
[05:43] <pitti> ah, heh, I see
[05:44] <psusi> if I manually mount /dev/hda in /media/cdrecorder, it shows up on the desktop using the voluem lame as the name of the icon "LinuxUDF"
[05:44] <psusi> if I use /dev/pktcdvd/pktcdvd0 instead, then it won't show up
[05:45] <psusi> then I think before I went to bed last night, I found that if I manually mounted it with a certain combination of mount options via the packet interface, it DID show up on the desktop, but as "cdrecorder" instead of "LinuxUDF" and the icon would not go away when I manually unmounted the volume
[05:46] <psusi> so it looks like something wonkey is going on in gnome-vfs right?  if so... can you sugest what code I should take a look at?
[05:46] <pitti> hm, not off the top of my head
[05:46] <pitti> I'd have to look into the drives exported by hal
[05:47] <pitti> the pktcdvd devices appear automatically?
[05:47] <psusi> I installed udftools and had to dpkg-reconfigure udftools and it set up /dev/pktcdvd/0 to connect to /dev/hda
[05:48] <pitti> ah, I see
[05:48] <pitti> so this mapping needs to be taken into account
[05:48] <psusi> and it seems that udev creates /dev/pktcdvd/pktcdvd0 with cdrom as the owning gid
[05:48] <pitti> that sounds fine
[05:48] <psusi> taken into account in what way that I did not already?
[05:49] <pitti> I think the main reason for not showing up is that hal does not have support for /dev/pktcdvd devices
[05:49] <pitti> i. e. it does not export a 'block' device for them
[05:50] <pitti> doing that would be the cleanest way IMHO
[05:50] <pitti> there is certainly a way (ioctl) to find out to which actual device a pktcdvd device is attached to
[05:51] <psusi> that sounds like it would be cleaner... but shouldn't my method work too?
[05:51] <pitti> g-vfs relies on hal
[05:52] <pitti> so, no
[05:52] <psusi> hrm....
[05:52] <pitti> instead of changing the /dev/hda entry, your fdi should create a new device
[05:52] <pitti> not sure whether that is possible, though
[05:53] <pitti> but figuring out the mapping is probably impossible in an fdi
[05:53] <pitti> so hal needs proper support for them
[05:53] <psusi> well, figuring out the mapping is easy, if a bit nieve...
[05:54] <psusi> hal detects /dev/hda... and so it tries to mount it... the only thing that needs to change is when mounting, it needs to use the other block device in place of hda
[05:54] <jbailey> SEJeff: ping
[05:55] <pitti> psusi: I'm still surprised that your hack works at all - I didn't expect pmount to mount pktcdvd devices
[05:55] <psusi> you don't really want to have a whole new hal device for the packet device
[05:55] <psusi> pitti: ohh, I also had to add it to /etc/pmount.allow I think
[05:56] <pitti> aah
[05:56] <pitti> that makes sense then :)
[05:57] <psusi> so it seems simple enough to me that the configure process for udftools could add the device to /etc/pmount.allow, and fix up the hal fdi to override block.device... and this does seem to correctly auto mount... except that gnome-vfs doesn't like it
[06:00] <psusi> does this seem like a simple and sane way to go about modifying the udftools package to support automount?
[06:04] <pitti> psusi: I'm a bit hesitant about changing /etc/pmount.allow automatically
[06:05] <pitti> psusi: I'd rather add proper support for packet writing to pmount
[06:05] <psusi> hrm... I can't remember now why that file even needed to be changed...
[06:06] <psusi> ohh, I think it is because pmount could not find the sysfs node for pktcdvd0
[06:06] <psusi> how do you think that pmount should go about properly supporting packet writing?
[06:07] <pitti> allow to mount pktcdvd without a pmount.allow entry
[06:07] <psusi> hrm....
[06:08] <psusi> well, I guess yuo could make pmount somehow understand that when it is asked to mount /dev/hda, that it should instead use /dev/pktcdvd/pktcdvd0, is that what you mean?  rather than change hal so g-v-m tells pmount to mount /dev/pktcdvd/pktcdvd0?
[06:09] <pitti> well, initially I just thougt about allowing to mount /dev/pktcdvd/
[06:09] <psusi> you mean hard coding /dev/pktcdvd/* in the whitelist?
[06:09] <pitti> not automatically figuring out the reverse mapping
[06:09] <pitti> something like that
[06:09] <pitti> these devices need to have some properties that make then unique
[06:10] <pitti> s/then/them/
[06:10] <psusi> what do you mean?
[06:12] <ogra_ibook> and you'll need a very good worked out policy for hal to handle the dual usage of the device as well i guess
[06:13] <ogra_ibook> cdrecord wont write an audiocd with a dev like /dev/pktcdvd/pktcdvd0
[06:13] <pitti> for g-v-m you mean
[06:13] <pitti> yep
[06:13] <ogra_ibook> nope, for hal itself ...to report both usages ...
[06:13] <pitti> this pktcdvd thingy is pretty hackish...
[06:13] <ogra_ibook> yup
[06:13] <pitti> yes, that's why I would prefer two devices
[06:13] <pitti> in hal
[06:14] <ogra_ibook> yup, sounds better ....
[06:14] <ogra_ibook> but anyway, you need to do some decision what to do with an empty cd/dvd
[06:14] <pitti> right
[06:15] <pitti> hm, this is so ugly, don't expect it to happen in the next two weeks anyway :)
[06:15] <ogra_ibook> heh
[06:16] <psusi> hrm... but two devices in hal doesn't make sense because there aren't really two devices... there is only one... you don't want to try to mount both
[06:16] <pitti> hm, but they are two different device nodes
[06:16] <ogra_ibook> from a system POV they are different
[06:16] <psusi> right... but one is purely virtual
[06:20] <psusi> and should only be used when mounting read/write... entirely in place of the real device
[06:55] <Lathiat> anywhere here actually use EVMS?
[06:55] <Treenaks> Lathiat: I would, if I knew how to actually manage it without the ncurses tool puking at me
[06:55] <erich^^> Can I setup a sarge pbuilder on an ubuntu system easily?
[06:56] <Lathiat> Treenaks: well, the gtk gui is useless
[06:56] <Lathiat> but i mean im about to setup a server im gonna be doing raid1 + lvm
[06:56] <Lathiat> traditonaly i would do it myself but im wondering if EVMS is worth touching
[06:56] <Lathiat> or if i want to avoid it :)
[06:57] <Lathiat> erich^^: sure
[06:57] <Lathiat> erich^^: just point it at the appropriate mirrors etc
[06:57] <erich^^> okay. I wasn't expecting much trouble, and I would just have copied over a base.tgz from a sarge box otherwise. thanks.
[06:58] <Lathiat> also does anyone have a nice solution for booting off a sw-raid1 where if the first disk dies you can still boot, short of rsyncing a /boot partition and keeping separate grub partitions on each one, like can i give grub a list of devices to try or something?
[06:58] <erich^^> Lathiat: if grub will actually be able to detect the failed disc properly...
[06:59] <tseng> a general rule of thumb is "dont put crazy software dependent layers on top of /boot"
[06:59] <Lathiat> brb
[06:59] <erich^^> Lathiat: I'd put two entries in the menu, and in the case of a failure have someone choose the other one...
[06:59] <pitti> Lathiat: do you depend on grub?
[06:59] <pitti> Lathiat: on my server with raid-1 I use lilo which does the job perfectly
[07:00] <erich^^> One feature I'm really missing in grub though is a try-once option. Lilo can do that, try once to boot a certain kernel, then go back to the old default.
[07:00] <pitti> Lathiat: I boot from the RAID, and keep both discs perfectly symmetrical
[07:00] <pitti> Lathiat: i. e. I don't have a separate /boot
[07:00] <erich^^> You could probably use that - make the secondary disk the default; boot once from the primary disk and make it set the primary as try-one on each boot.
[07:01] <pitti> Lathiat: but even if you want to keep apart /boot, why not put it onto a raid partition as well?
[07:01] <erich^^> pitti: he has it in raid-1
[07:01] <pitti> oh, so where's the problem?
[07:01] <erich^^> pitti: the question is how to make grub actually access the second disk.
[07:01] <pitti> ah, I see
[07:01] <erich^^> because your config will probably list (hd0,x), too...
[07:02] <erich^^> But given the nature of bad disks, I don't think grub would be able to detect a bad disk automatically anyway...
[07:02] <pitti> no idea then, I don't want to use grub on my server, and that's the only raid-1 machine I have
[07:04] <erich^^> pitti: what are you using instead? did you try unplugging your primary disk to see if the system comes up okay?
[07:04] <pitti> erich^^: I use lilo because of lilo -R
[07:05] <pitti> erich^^: hda died twice on that system; I asked the guys in the DC to put hdc as hda, and insert a new hdc, that worked fine
[07:05] <pitti> since /etc/lilo.conf and fstab only mention /dev/mdN, the discs are symmetic
[07:06] <pitti> erich^^: I didn't try to *boot* from /dev/hdc; I don't have physical access to my server, it probably requires a BIOS setting
[07:07] <erich^^> pitti: yeah, that is about as far as I got, too. that removing the bad drive and making the good one hda will work.
[07:07] <pitti> booting from hdc is certainly a BIOS issue?
[07:07] <pitti> I don't see what grub could do about it
[07:09] <Lathiat> the bios could fall back
[07:09] <Lathiat> the problem si if you just have a standard grub config pointing at sda
[07:09] <Lathiat> it wont boot sd that doesnt exist
[07:10] <pitti> well, lilo/grub should always be pointed to raid devices
[07:10] <Lathiat> hrm ok i thought the way it worked
[07:10] <Lathiat> it just read the 'raid1' by reading 1 disk
[07:11] <Lathiat> hence not being able to read raid0 due to striping
[07:11] <Lathiat> pitti: so, if you unplug your first disk, and the bios boots from it
[07:11] <Lathiat> pitti: does lilo boot from it?
[07:11] <pitti> Lathiat: I hope, but I don't know; as I said, this server is 800 km away from me in a DC
[07:11] <pitti> so I can't try
[07:11] <Lathiat> right
[07:11] <Lathiat> well, mines sitting next to me, so i might try that with lilo & grub
[07:11] <pitti> given that the BIOS boots from hdc, it should just work
[07:12] <Lathiat> pitti: assuming lilo actually looks at hdc and not hda for the kernel image
[07:13] <pitti> ideally it should look at all physical devices of the raid
[07:13] <pitti> and maybe try all in succession
[07:13] <Lathiat> ideally
[07:13] <pitti> but I don't know that
[07:13] <Lathiat> i'll try and let you know
[07:13] <pitti> would be interesting, thank you
[07:13] <Lathiat> atm on a box we have an ugly hack of a /boot being individual partitions on both drives
[07:13] <Lathiat> and we rsync the kernel images accross and have different lilo configs on each
[07:13] <pitti> that sucks, right
[07:13] <Lathiat> (i didnt set that up)
[07:27] <erich^^> pitti: I don't think grub can actually read raid. But it will be able to ignore the raid superblock and just access it as if it were a regular disk, I think. In raid1 that is.
[07:28] <erich^^> But basically when you have a hardware failure and need to reboot, just replace the broken disk and swap them... it's not much more work...
[07:29] <sivang> rehi all
[07:29] <sivang> pitti: still here ? :)
[07:31] <pitti> erich^^: no raid support in grub? that's a severe shortcoming, Id' say
[07:32] <Lathiat> erich^^: its mroe work when you have to physically get to the machine
[07:32] <Lathiat> erich^^: and yeh thats what i thought
[07:32] <erich^^> pitti: well, given that you have 512 bytes...
[07:32] <pitti> erich^^: grub has way more :)
[07:33] <Lathiat> pitti: which.. it loads off the disk? :)
[07:33] <erich^^> pitti: these 512 bytes need to contain enough code to find the rest of grub - if you put that data on a non-linear storage, good luck
[07:33] <Lathiat> which... would need raid? :)
[07:33] <pitti> Lathiat: *brown paperbag*
[07:33] <Lathiat> on my head or yours? ;p
[07:35] <pitti> mine of course :)
[07:35] <Lathiat> but yeh, i guess thats why hardware raid is still usefull :)
[07:35] <erich^^> grub could probably have raid support to load kernels and such, as long as you give it some extra k of storage to find its data... thats why it's not uncommon to have a non-raid partition on the disks then.
[07:35] <erich^^> Lathiat: well, just having a 16k solid-state-disk (like the flash chip of a hardware raid card) would probably do the trick just fine...
[07:37] <psusi> yea... you should just mirror the entire drive... normally the bios and grub will only use the first one, but that's ok... if the first one fails, unplug it and the second one should boot fine
[07:38] <Lathiat> psusi: youd have to put the second one onto the same slot as the first, tho
[07:38] <erich^^> But using raid-1 for /boot isn't really a problem, is it?
[07:38] <erich^^> you could probably modify grub to fail back to the secondary drive...
[07:38] <Lathiat> erich^^: its a problem in that, if the first disk dies, it wont boot because the grub/lilo config points at the first disk being hd(0,0) nto hd(0,1), which isnt there 
[07:39] <erich^^> but it won't help when the primary drive is completely dead unless your bios is smart enough...
[07:39] <Lathiat> yeh i suspect you may be able to squeeze that into the first part of grub... maybe.
[07:39] <Lathiat> erich^^: yeh
[07:39] <psusi> grub will access whichever disk it is on... if it is on both because they are mirroed, then it's up to the bios which disk it boots
[07:39] <Q-FUNK> elmo: could you sync rus-ispell with unstable, please? thanks!
[07:39] <erich^^> psusi: yeah, but what if the first sector of the first drive is fine, the rest is trashed?
[07:40] <psusi> then you would need to yank it out :)
[07:40] <psusi> or configure the bios to boot from the other drive
[07:40] <Lathiat> what drive you boot from is not the problem
[07:40] <erich^^> yeah, or have a grub in the first sector that will fail over.
[07:41] <erich^^> still, hard to do in 512 bytes I guess.
[07:41] <erich^^> But this basically applies to any stage of grub... stage_1 can be readable when stage_2 is broken...
[07:41] <erich^^> but it gets less helpful with each...
[07:41] <psusi> I'd say impossible... you don't even have 512 bytes for code... some of that is the partition table...
[07:41] <psusi> I've modified boot sector code before... it is TIGHT
[07:41] <erich^^> I'm not sure it's worth the hassle.
[07:42] <psusi> basically if parts of the disk that contain any part of grub are broke, you're going to need to yank the disk and rely on the other one
[07:42] <Lathiat> erich^^: probably not :)
[07:42] <Lathiat> psusi: that still brings up the issue of grub then looking at the second disk and not the first can it do that now?
[07:42] <Lathiat> (without having a second separate config)
[07:43] <psusi> you just configure grub to load from the first disk... if the first disk fails, yank it out and the second becomes the first
[07:43] <Lathiat> no it doesnt
[07:43] <Lathiat> well, depends what kind of drives they are
[07:43] <dilinger> bypass the problem entirely.  boot from a usb memory stick :p
[07:44] <Lathiat> dilinger: :)
[07:44] <psusi> if you remove whatever drive is bios disk 80, then bios disk 81 should become bios disk 80
[07:44] <Lathiat> psusi: mm well i'll have to try it
[07:44] <psusi> or in modern bioses you can usually tell them to boot a drive other than 80
[07:49] <psusi> though I'm not sure if they just load the mbr from the other drive, or if they remap the other drive to 80
[07:49] <psusi> if it does not remap... then the MBR would still try to load the stage 1.5/2 loader from the bad disk
[07:50] <Lathiat> heres a Q, why is there a stage 1.5 and then 2?
[07:51] <psusi> stage 1.5 is optional... stage2 can be built to understand several filesystems... that can make it too large to fit in the 62 sectors that are usually free between the MBR and the first partition
[07:51] <psusi> so you can use a stage 1.5 loader that fits there because it only udnerstands the one filesystem that holds the stage 2 loader
[07:52] <Lathiat> ah
[07:52] <Treenaks> OpenFirmware yay :)
[07:52] <Lathiat> i toyed with a xeon server at work with this cracky EFI thing
[07:52] <Lathiat> which is terribly slow
[07:53] <Lathiat> and apparently not very usefull
[07:53] <erich^^> psusi: still, the bios needs to detect the drive being broken. and you currently might need different grub config files on both disks for grub to access the right one (I mean, you could install grub on the first drive, and have your kernel on the second; but you don't want that in this case...) - so you'll need different /boot dirs on both disks. Which then makes renumbering them bad... :-(
[07:53] <erich^^> anyway. see ya
[07:53] <Lathiat> bye, thanks for the talk :)
[07:54] <Lathiat> sounds like what im doing now might be the way to go :)
[07:54] <psusi> erich^^: you don't want a seperate /boot on each drive is the point
[07:54] <Lathiat> psusi: but you might have to, being the point :)
[07:54] <Lathiat> as sucky as that is
[07:54] <psusi> I don't see what that actually buys you
[07:54] <erich^^> psusi: yeah, but that will only work if you plug the good drive back in as primary; not if you replace the broken primary and keep the second second.
[07:54] <Lathiat> being able to get a gimp to reset a machine and get it up with a disk dead without shuffling disks around inside a case without hotswap bays
[07:55] <psusi> erich^^: if the bios can only boot from the primary, then it won't work even if you have a seperate /boot
[07:55] <psusi> if it can boot from the secondary, then it should work without a seperate /boot
[07:55] <erich^^> psusi: on scsi it probably can boot from the secondary; and with a IP-KVM you might even be able to do that remotely.
[07:56] <psusi> erich^^: exactly... and once it boots from the secondary drive, you don't need it to have a seperate /boot because the secondary drive BECOMES bios disk 80, which is where grub will load the kernel from
[07:57] <Lathiat> psusi: mm, i have to try that
[07:57] <Lathiat> will do soon
[07:57] <psusi> grub is configured to access bios disk 80... if you boot from the primary drive, that is disk 80.. if you boot from the secondary, that becomes disk 80... so either way, the same grub config will read the kernel from whichever disk you boot from
[07:58] <psusi> in fact... if you do use two seperate grub configs, it WON'T work if the primary fails
[07:59] <psusi> because when you install grub, if you configure the one on the second disk to load the kernel from /dev/sdb, grub will think that is bios disk 81
[07:59] <psusi> but if you try to have the bios boot from the second disk, grub will be trying disk 81 but it won't be there, it will be disk 80 instead
[08:41] <robertj> i've been thinking about the best way to handle mounting an smb share as /home and from what I can tell setting smbmount and smbumount suid root and writing ~/.pam_mount.conf before loading pam_mount seems to be the best way, any other ideas?
[08:41] <robertj> I can
[08:42] <robertj> er I can't think of a way that could be done without having an upgrade trounce all over it
[08:43] <psusi> what's wrong with just adding the mounts to /etc/fstab?
[08:43] <robertj> psusi: I want to manage it at login time through a directory
[08:45] <psusi> what do you mean?  manage it through a directory?
[08:45] <robertj> psusi: I want to specify per-user & per-group mounts
[08:46] <robertj> so those could be written to ~/.pam_mount.conf after checking with ldap
[08:46] <psusi> ohh... you mean like mount differntly dependong on who logs in?
[08:46] <robertj> yup
[08:46] <psusi> add them all to fstab with the noauto and user options... then when the user logs in, then can mount it
[08:46] <robertj> and also it's just not good to have 900 users in /home on a desktop IMO
[08:47] <psusi> the one that they want that is... 
[08:47] <psusi> ohh, you want users to be able to have their home directories on the network?
[08:47] <robertj> yes
[08:48] <psusi> hrm... why not use nfs and mount all of /home on a server then?
[08:48] <robertj> psusi: well smb is the destination of choice for me and like I said, I don't like users navigating through 900 entries in /home
[08:48] <tseng> psusi: because windows doesnt speak nfs
[08:49] <robertj> but with pam_mount the sky is the limit and you can even use fuse mounts
[08:49] <robertj> so it's just a more granular way of controlling things
[08:49] <psusi> well, usually users don't look in /home, they just worry about their own home dir...  but yea... that's a problem with smb since the entire mount uses the same credentials
[08:50] <robertj> psusi: which means also that you have to mount it as root
[08:50] <psusi> though actually, windows does support NFS I believe... MS has a package called services for unix
[08:50] <robertj> which means if you get rooted...
[08:51] <psusi> ohh... yea... there is a facility in pam to mount isn't there?  so what's the problem then?
[08:51] <robertj> psusi: yeah, MS even supports hosting an Appletalk printer on NT4, they have all kinds of things no one knows about
[08:51] <robertj> psusi: well all the mounting programs used would need to be suid root
[08:51] <robertj> or use fuse which I haven't used alot
[08:52] <psusi> why would they need suid root?  the login program invoking pam is running as root
[08:52] <robertj> ahh
[08:52] <robertj> lemme try that then
[08:52] <psusi> so pam should be able to invoke mount as root... and mount is already suid root anyhow ;)
[09:08] <robertj> psusi: hrmmm
[09:08] <robertj> http://paste.ubuntu-nl.org/6337
[09:08] <robertj> apparently its not
[09:10] <robertj> or...
[09:10] <psusi> well login definately runs as root, so unless pam drops root privs before getting to the stuff you're adding, you should be root ;)
[09:14] <robertj> apparently it is
[09:15] <robertj> ls
[09:15] <robertj> doh ;)
[11:05] <a_monkey> I got some problem booting the install of ubuntu on my laptop
[11:05] <a_monkey> ive been told the bootcommand is "linux apic=off noacpi"
[11:06] <a_monkey> but, still it wont detect my CD-ROM after langage-settings 
[11:06] <Burgwork> a_monkey, please use #ubuntu, as this is not a support channel
[11:06] <a_monkey> arright
[11:06] <a_monkey> this was listed as the "laptop team" so... sorry
[11:07] <Burgwork> a_monkey, #ubuntu-laptop
[11:08] <a_monkey> thank you
[11:17] <ryanpg> pitti, hi bug 20564 is marked RESOLVED but I'm still experiencing it, should I reopen?
[11:18] <pitti> ryanpg: yes, sure
[11:18] <ryanpg> k
[11:19] <pitti> ryanpg: so your device permissions are still wrong?
[11:21] <ryanpg> pitti, sorry I forgot the details but they're currently brw-r----- 1 root plugdev 8, 3 2005-12-29 16:09 /dev/sda3
[11:26] <ryanpg> pitti, does permission still appear to be the issue? (I'm not sure what they should be) should I click "commit" ;)
[11:26] <pitti> ryanpg: wait
[11:26] <ryanpg> k
[11:27] <pitti> ryanpg: ok, the permissions seem fine now
[11:27] <pitti> ryanpg: well, just commit your current comment
[11:27] <pitti> ryanpg: can you please do the https://wiki.ubuntu.com/DebuggingRemovableDevices steps again?
[11:28] <ryanpg> pitti, sure should I attach output in a single file or seperate?
[11:29] <pitti> ryanpg: the big ones in separate files, please
[11:29] <ryanpg> doing so now
[11:29] <pitti> the small stuff (mount output, etc.) just as a paste
[11:29] <pitti> thank you
[11:32] <ryanpg> and only the relevant end part of dmesg needed yes?
[11:34] <pitti> right
[11:34] <pitti> good night everybody
[11:34] <tseng> bye pitti