[00:00] <SpamapS> for instance , if I have a maintainer script that stops a service which takes a long time to stop said service, and then tries to stop it again.. it shouldn't fail if the service is already stopped.. but mysql seems to fail this way all the time.
[00:01] <SpamapS> the scenario is prerm stops job, then preinst stops job ...
[00:06]  * SpamapS realizes its friday night and he's mostly typing into the emptiness of space
[01:26] <smoser> cjwatson, so, if you see this, i've now (i think) sufficiently verified bug 581760
[01:28] <smoser> now we can get on to important bugs like bug 671097
[01:34] <anon^_^> this is probably a question that's been asked a few times, apologize if it is.
[01:35] <anon^_^> What's the status on [RFC/RFT PATCH] discussed on LKML being backported to Ubuntu 10.04?
[01:35] <anon^_^> http://lkml.org/lkml/2010/10/19/123
[01:39] <Pici> anon^_^: Likely not, see http://askubuntu.com/questions/13562/how-do-we-get-this-magic-performance-boosting-200-line-patch
[01:40] <psusi> stop reading linux tabloids ;)
[01:40] <ebroder> Why is everybody so excited about this patch all of a sudden?
[01:41] <psusi> because Phoronix is hyping it
[01:41] <Sarvatt_> (and slashdot)
[01:41] <psusi> with videos of the amayzing difference it makes when you are stupid enough to run make -j64
[01:41] <ebroder> (and slashdot)> Ah, that makes sense
[01:41] <psusi> yea, that too
[01:41] <ebroder> I've never heard of Phoronix and I stopped reading /. about a year ago :)
[01:42] <Pici> Only a year ;)
[01:42] <psusi> I've not bothered reading /. in a few years unless I'm board, and I finally checked out phoronix once or twice before this broke because I keep hearing so much about it
[01:42] <ebroder> I spent longer thinking about stopping :-P
[01:42] <maco> phoronix is a benchmarking software maker who uses their software to benchmark OSes and hardware and people care about the results
[01:42] <psusi> I was fairly certain it was crap before this story broke... now I'm sure
[01:43] <Pici> Anyway, that askubuntu link above has comments by a bunch of our kernel people in it.
[01:43] <psusi> in this case, they are showing a test case involving make -j64, which is absurd, and the uneducated masses see it and generalize it to achieving the same results for normal desktop loads
[01:44] <anon^_^> ebroder, because desktop responsiveness is an issue are load
[01:44] <psusi> I've been meaning to high five Kees when I see him on irc for his comment pointing out that this will have no effect for the average desktop user
[01:44] <anon^_^> whether this patch actually addresses that issue is a different question
[01:44] <psusi> anon^_^, sure, but your average Ubuntu user does not typically have a load avg of 64 ;)
[01:45] <ebroder> It's not just that. Your average Ubuntu user doesn't have non-trivial processes associated with more than one tty
[01:45] <psusi> exactly
[01:46] <anon^_^> well and then there's the issue of what constitutes load.
[01:47] <anon^_^> try running an luks lvm encrypted system moving writing files to disk and watch a desktop crawl to a halt when multitasking
[01:49] <psusi> the NT kernel has always had a feature where it boosts the scheduling priority of the thread that owns the foreground window.. I've long wondered why we don't have something like that
[01:49] <Sarvatt_> the userspace implementation of it very noticeably helps while dist-upgrading or building packages while browsing the web on an atom :)
[01:51] <psusi> you know... I've been running dist-upgrades a lot lately testing out lvm snapshots and don't have bad lag on gui apps while that goes on... I've also been playing with the apt options to make more use of triggers and avoid re-triggering the same things 12 times during the dist-upgrade, as well as using libeatmydata to stop the damn fsyncing dpkg normally does that slows things down so much for the last release or two
[01:53] <superm1> would it maybe make sense to use libeatmydata during initial install when using d-i?
[01:53] <superm1> but still leave default fsync behavior otherwise
[02:04] <psusi> there's a patch working its way into dpkg to add a switch to stop the extraneous syncs, but yes, it would
[02:06] <superm1> well a proper patch to turn it off sounds much better than hacking in an additional library
[02:06] <psusi> yep
[02:06] <psusi> either way, I'm happy to dist-upgrade in 10 minutes or less instead of 20 or more
[02:07] <cody-somerville> Omg. Call of Duty Black Ops ships zork in it as an easter egg.
[02:07] <cody-somerville> along with a 'linux console'.
[02:12] <d_ed> cody-somerville: just found a video of it...ooh
[02:12]  * psusi fires up tonight's dist-upgrade to natty
[02:13] <cody-somerville> d_ed, apparently you can do all sorts of stuff in it - even rlogin into other systems, lol.
[02:13] <d_ed> I've not played the game, but I feel I should get it now.
[02:13] <cody-somerville> Its an easter egg, hidden in an easter egg, hidden in an easter egg. Its bloody brilliant.
[02:14] <cody-somerville> Anyhow, back to my xbox. :)
[02:15] <d_ed> and back to hacking
[02:16]  * psusi has proposed an updated lvm2 for merge
[03:39] <ScottK> Nice to see Canonical has taken over dealing with sponsorship.
[10:54] <geser> cjwatson: how far up in your TODO queue is the review (and sponsoring) of bug #617885? The current suggested "fix" from other bug comments are links to the natty debs
[12:00] <asac> here is a tool (perf) that doesnt properly resolve symbols from dbgsym it seems ...
[12:00] <asac> how can i manually resolve it?
[15:24] <weasel> is there some system that has sources of older packages (akin to snapshot.debian.org)?  specifically I'm looking for openssl_0.9.8k-7ubuntu8.3.dsc
[15:26] <elmo> weasel: https://launchpad.net/ubuntu/+source/openssl/0.9.8k-7ubuntu8.3
[15:26] <weasel> thanks
[15:26]  * weasel goes to interdiff 8.3 and 8.4, since you broke all tor relays
[15:26]  * weasel hates libssl
[15:27] <elmo> jdstrand/mdeslaur/kees: ^-- fyi (since sbeattie is away)
[15:29] <weasel> [ http://archives.seul.org/or/talk/Nov-2010/msg00128.html ]
[15:29] <mdeslaur> weasel: if you downgrade to 0.9.8k-7ubuntu8.3 it starts working again?
[15:30] <weasel> hard to tell, I don't have an ubuntu system :)
[15:30] <weasel> guess I'll set up a chroot somewhere
[15:30] <weasel> but several people have reported that their relays stopped working after upgrading libssl, so /something/ is going on
[15:30] <mdeslaur> weasel: well, that distro are you running? the 8.4 release adds the same upstream security fix that everyone else used
[15:31] <weasel> mdeslaur: yup, so it appears
[15:31] <weasel> so the next step should probably be reproducing it in an ubuntu chroot and reproducing it with 0.9.8p from upstream
[15:32] <mdeslaur> weasel: could you please open a bug and track your progress in it? or ping me again if there's something we need to do?
[15:33] <weasel> mdeslaur: should it turn out to be ubuntu specific I'll complain some more.  and I'll try to keep you updated
[15:33] <mdeslaur> weasel: thanks
[15:33] <weasel> debian sid should have the same patch.  /me goes to check that
[15:37] <mdeslaur> weasel: yes, sid has the same patch, I just checked
[15:37] <weasel> yup
[15:44] <weasel> ok, not ubuntu specific.
[15:45] <mdeslaur> weasel: well, let me know if there's anything I can go to help
[15:46] <mdeslaur> s/go/do/
[15:46] <weasel> will do
[15:46] <mdeslaur> thanks weasel
[15:46] <weasel> I suspect our (Tor's) libssl guru will have to add yet another work around to work everywhere
[19:14] <gaurava> Hi All ..  I am a newbie here
[19:15] <gaurava> I am trying to create a sample patch file following this article <https://wiki.ubuntu.com/Bugs/HowToFix>  but facing some issue ..
[19:16] <gaurava> nyone .. can help me out .. or point me to the right channel
[19:19] <ScottK> gaurava: #ubuntu-motu is probably more active on a weekend.
[19:20] <gaurava> thanks Scottk ... will try my luck over there .
[19:20] <gaurava> #join ubuntu-motu
[21:05] <buxy> cjwatson, pitti: should ubuntu switch to use data.tar.xz by default instead of stripping changelog and all that kind of intrusive changes?
[22:03] <ebroder> buxy: The issue isn't in the size of the .debs. It's the size of the unpacked .debs - the desktop CD is all of the CDs unpacked, and then squashfs'd
[22:05] <ebroder> buxy: And using lzma to compress the filesystem is infeasible because it breaks zsyncing from one daily CD to another (zsyncing becomes basically equivalent to re-downloading the whole CD, which isn't feasible for some of our QA procedures)
[22:06] <ScottK> lzma does help on the alternate CD though.
[22:07] <ScottK> (not that it particularly needs it)
[22:17] <JanC> ebroder: zsyncing is only relevant for daily builds though, so it shouldn't be an issue for final builds?
[22:19] <ebroder> JanC: But you can't switch from one compression algorithm to another for the final build. That defeats the point of doing the testing
[22:19] <ebroder> (Keybuk was very proud of having found a bug in squashfs's gzip compression in the past)
[22:19] <JanC> well, you can still test both
[22:20] <JanC> but use the zsync'able version for "daily tests" with testdrive
[22:21] <JanC> and people who don't care about bandwidth can use the lzma version
[22:21] <JanC> it would require changing the build process of course
[22:22] <JanC> and testing how much more space we get that way  ;)
[22:23] <ebroder> Well, I can at least give an estimate for the space savings. I'm rolling custom live CDs at work, and I tried building one with lzma just out of curiosity. lzma image was about 83% the size of the gzip image
[22:24] <ebroder> I was working with a custom set of packages, so your mileage may vary
[22:26] <ebroder> But in any case, the people at the UDS session were fairly adament that we only generate one desktop spin, and that it must be incrementally updatable
[22:33] <JanC> ebroder: so basically this is dictated by the QA people (which I understand)
[22:33] <ebroder> JanC: Yeah, basically
[22:34]  * penguin42 wonders what % of users actually install from a physical CD these days
[22:35] <JanC> penguin42: depends on if I have a live-USB with me or not  ;)
[22:35] <penguin42> JanC: Are you more likely to have a spare USB thumb or a blank CD with you?
[22:36] <JanC> penguin42: I almost always have an official CD with me  ;)
[22:37] <penguin42> I was just wondering that given the minimum spec of machines these days very few can't boot off USB, heck there are probably more machines with no-optical drive rather than ones that can't boot off USB
[22:37] <JanC> as I have our locoteam's box of official-CDs-for-redistribution at my home...  :P
[22:38] <JanC> penguin42: when we are at computer fairs we also distribute free CDs/CD-Rs, not free USB sticks  :P
[22:39] <JanC> so I imagine a lot of people install from CD still
[22:39] <ebroder> penguin42: That came up at UDS, too. cjwatson made the point that even if people aren't using CDs, the size of a CD presents *some* limitation to keep the image size from ballooning out of control
[22:39] <penguin42> ebroder: Yeh that is true
[22:40] <JanC> of course if they bring an USB stick, we'll make it a bootable live-USB for them
[22:40] <ebroder> And also, with the rise of netbooks, you can no longer assume that computers have arbitrarily inflated specs. Disk space post-installation is a legitimate concern for netbooks, and controlling the CD size also conveniently controls the post-install size
[22:41] <penguin42> ebroder: Indeed post-install size is as important
[22:41] <JanC> IIRC Windows 7 is about 20 GiB post-install now...  :P
[22:42] <ebroder> Hmm...I don't think it's actually that bad. More like 10
[22:42] <JanC> maybe depends on what version of Win7
[22:42] <JanC> and maybe what options you install
[22:43] <JanC> oh, and that was after all the updates
[22:44] <penguin42> heck, apparently you can get 1GB usb sticks for 59p now :-)
[22:44] <JanC> (and after more than 3 hours of installing updating)
[22:44] <ebroder> penguin42: Sure, but you can get CDs for a fraction of that
[22:44] <penguin42> ebroder: Indeed, I was just being pleasently shocked by what they had come down to
[22:44] <JanC> fortunately it took only 15 minutes to install + update Ubuntu  ;)
[22:46] <JanC> Pendulum: where do you find 1 GB USB sticks still?
[22:47] <JanC> penguin42: where do you find 1 GB USB sticks still?
[22:47] <JanC> (sorry Pendulum ;) )
[22:47] <penguin42> JanC: Google found me http://www.kikatek.com/product_info.php?products_id=34487&source=froogle
[22:48] <penguin42> bah, if you click check availability it says sold out
[22:52] <JanC> penguin42: cheapest I can find are 5-6 £
[22:53] <JanC> taht's like 10× more...
[22:55] <JanC> well, cheaper too, if you order a 1000 of them  :P