[14:37] <highvoltage> stgraber, mgariepy: any comments on http://edubuntu.org/sites/default/files/banner-slideshow/precise-release.html for a release banner on edubuntu.org? http://edubuntu.org/sites/default/files/banner-slideshow/precise-release.html
[14:37] <stgraber> highvoltage: looks good
[14:38] <highvoltage> stgraber: I also think that we should have an "About Edubuntu" paragraph at the end of the announcement
[14:38]  * highvoltage adds it so long
[14:38] <stgraber> highvoltage: good idea, you can probably copy/paste some of what I put in the wiki announcement
[14:38] <highvoltage> hmm, link?
[14:41] <stgraber> highvoltage: https://wiki.ubuntu.com/PrecisePangolin/ReleaseNotes/Edubuntu though looking at it again, there's only the first sentence you might re-use
[14:41] <highvoltage> great
[14:44] <highvoltage> stgraber: I'm going to change the spin slightly and assert Edubuntu more as a project than a product
[14:45] <mgariepy> highvoltage, it looks good :)
[14:49] <stgraber> highvoltage: I'll need to get the weblive VMs to run with libeatmydata ... it's so slow to build with all the fsyncs() and it's not like I care about the data
[14:49] <highvoltage> heh, libeatmydata avoids the fsyncs? :)
[14:50] <stgraber> yeah, it restores your system to pre-ext4 fiasco where you hope that the fs autosyncing is enough to get your data on disk
[14:50] <stgraber> but prevents dpkg (well, everything) from calling fsync everytime they write to disk
[14:50] <highvoltage> doesn't data=writeback do that already?
[14:51] <stgraber> highvoltage: isn't that only for metadata syncing?
[14:52] <highvoltage> from what I understand, data=writeback basically makes it work like reiserfs and if you have a power outage or something, you just travel back in time and lose all recent writes
[14:55] <highvoltage> the quick usage instructions doesn't seem to disagree with either of us, but it does seem like data=writeback is the "ldm_directx" of ext4 :)
[14:55] <highvoltage> http://kernel.org/doc/Documentation/filesystems/ext4.txt
[14:56] <highvoltage> it's interesting how increasing the journal size could increase performance. I just go ahead and disable it completely on slow disks, like on my acer one.
[14:58] <highvoltage> I've also had some success with commit=60 on slow disks where I wouldn't particular care losing a minutes worth of work (like also on the netbook)
[14:58] <highvoltage> (but I'm sure you know about all of that already :p)
[15:01] <stgraber> apparently I need to use cache=none on my VM disk
[15:01] <stgraber> that should save quite a bit of cpu by avoiding use of the host cache
[15:01] <stgraber> (I'm giving a partition directly to the VM)
[17:27] <highvoltage> stgraber: mgariepy, alkisg, (and others): we have an edubuntu meeting at 19:00 UTC today, will you be able to make it?
[17:28] <stgraber> should be
[17:28] <alkisg> highvoltage: I don't think I can make it
[17:29] <alkisg> highvoltage: one idea I'd like to discuss for 12.10 is that the edubuntu dvd could support ltsp fat clients with the existing squashfs image now
[17:30] <alkisg> This would make the dvd smaller, the clients boot faster, and it would support fats along with thins
[17:30] <highvoltage> alkisg: yeah knoppix does that and I always thought it was really cool
[17:30] <highvoltage> ah you mean thin clients too, ok.
[17:30] <alkisg> Now we have a chroot in the dvd
[17:31] <alkisg> Instead of that, we could use the whole casper squashfs image as the chroot
[17:31] <highvoltage> alkisg: and what happens when we want to install LTSP?
[17:31] <alkisg> We'd need to have both ltsp-server and ltsp-client installed in the dvd for that to work
[17:31] <alkisg> We just remove ltsp-client in that case
[17:31] <highvoltage> I mean, when we install LTSP during the system installation?
[17:31] <highvoltage> currently the installer uses the chroot that is shipped on the dvd for that
[17:32] <alkisg> Two ideas:
[17:32] <alkisg> 1) use the squashfs image of the cd,
[17:32] <alkisg> or better yet,
[17:32] <alkisg> 2) just clone the server disk
[17:32] <alkisg> After ubiquity finishes, we can run ltsp-update-image --export-root
[17:32] <alkisg> That uses the server disk, while cleaning it up a bit
[17:33] <alkisg> (I've implemented it as part of ltsp-pnp, putting it upstream currently, will finish in a few weeks)
[17:33] <alkisg> So people would have support for thin and fat clients out of the box and even with a smaller DVD
[17:33] <highvoltage> server's disk? as in the server's /?
[17:33] <alkisg> Yes
[17:34] <alkisg> So it automatically takes the language settings and everything
[17:35] <alkisg> The user accounts and other sensitive data are removed of course before being put in the chroot
[17:36] <alkisg> Another thing I'd like in the 12.10 Ubuntu CDs (and edubuntu DVD) is ipxe
[17:36] <alkisg> Would allow people that don't have bios netboot option to use the edubuntu dvd to netboot all clients
[17:37] <alkisg> I think I've filed a bug report for that and it received positive comments...
[17:42] <highvoltage> ok, interesting
[18:12] <stgraber> highvoltage: can you help testing the amd64 image? weblive is taking me longer than expected
[18:12] <highvoltage> stgraber: on it...
[18:13] <highvoltage> (syncing now, I can't think that it would take long)
[18:13] <stgraber> highvoltage: Ideally, do the following:
[18:13] <stgraber>  - Boot to live session in a non-english language
[18:13] <stgraber>  - Start LTSP-Live
[18:13] <stgraber>  - Login from a thin client => check session language
[18:13] <stgraber>  - Shutdown thin client
[18:14] <stgraber>  - Install to disk in another language
[18:14] <stgraber>  - Use manual partitioning
[18:14] <stgraber>  - Use LTSP + gnome-fallback
[18:14] <stgraber>  - Unselect tertiary
[18:14] <stgraber>  - Select encrypted home
[18:14] <stgraber> That's the easiest way of covering 95% of the code paths (just missing OEM)
[18:15] <highvoltage> ok
[18:15] <stgraber> if that all works, the image is good to ship
[19:02] <highvoltage> stgraber, mgariepy: should we do a quick release roundup in #ubuntu-meeting since we have a meeting scheduled?
[19:02] <stgraber> sure
[19:02] <highvoltage> great.
[19:40] <stgraber> highvoltage: install done
[19:40] <stgraber> highvoltage: doing some post-install test of the codec fix
[19:40] <mhall119> stgraber: highvoltage: I linkified some of the project names on https://wiki.ubuntu.com/PrecisePangolin/ReleaseNotes/Edubuntu if you want to look over it
[19:40] <highvoltage> I'm getting some weird graphics issues on ltsp-live, it looks like some weird compositing issues where any square is just drawn as a triangle
[19:41] <stgraber> highvoltage: what video card are you using?
[19:41] <stgraber> I'm using cirrus here and it looked good, using Precise's kvm
[19:42] <highvoltage> stgraber: it's one of those good HP thin clients (with the intel mobile 4 iirc cards)
[19:42] <stgraber> oh, ok, that's weird
[19:42] <highvoltage> indeed, especially since I've never seen that so far on precise
[19:43] <highvoltage> my checksums match. this is the right iso, right? http://cdimage.ubuntu.com/edubuntu/dvd/current/precise-dvd-amd64.iso
[19:43] <stgraber> 20120425
[19:44] <highvoltage> ok they're the same, at least
[19:44] <highvoltage> well, ltsp-live itself works at least
[19:45] <stgraber> yeah, just some weird ltsp issue to look into
[19:45] <stgraber> apparently unity 3d is broken too, some people reported blank screens on LTSP unless you explicitly select ubuntu-2d
[19:45] <stgraber> so much for 16bit just working I guess...
[19:46] <stgraber> the dash overlay on the ubuntu wallpaper in 16bit is also extremely ugly, thankfully Edubuntu's wallpaper is quite a bit better
[19:48] <highvoltage> ok I'm on to installing... the edubuntu slideshow looks cool in French :)
[19:49] <stgraber> right, so I confirmed the codec fix, just checking that post-reboot oem still works and then waiting for your install result and will be consider as good to relesae
[19:50] <stgraber> considering the few LTSP issues I've seen so far, I think we should prioritize these before any 12.10 feature work so we have them all fixed for 12.04.1
[19:51] <stgraber> we won't be able to revert the 16bit change in an SRU but we should at least make sure the session always open fine and anything crashing because of 16bit gets fixed
[19:58] <Ziron> Is the 12.04 release still on schedule for tomorrow?
[20:00] <highvoltage> yep
[20:00] <highvoltage> there's been some hickups but it shouldn't cause a delay
[20:18] <stgraber> highvoltage: still installing?
[20:19] <highvoltage> stgraber: yep, just checked
[20:29] <highvoltage> stgraber: done, and I'm passing it despite the display problems I had in LTSP Live
[20:30] <stgraber> k
[20:30] <highvoltage> stgraber: I'll recheck that and file a bug if I can reproduce, if it's valid I guess we can SRU it for 12.04.1
[20:30] <stgraber> yeah, if we can reproduce, we should fix and SRU
[20:30] <stgraber> I don't have any LTSP hardware here so I'll need to wait to be back home before I can test any of that
[20:31] <highvoltage> the installed system didn't have that error on LTSP
[20:31] <stgraber> oh really? that's very odd then
[20:31] <highvoltage> yep. perhaps something didn't reset properly on the thin client.
[20:31] <stgraber> because we don't really change anything in lts.conf and the only thing we do to the image is run ltsp-update-image
[20:31] <highvoltage> yep
[20:32] <stgraber> yeah, sounds like graphical memory corruption or similar random weirdness
[20:32] <highvoltage> but that's why I have to check again, perhaps it happens intermittently
[20:32] <stgraber> anyway, can you post your results on the tracker before I move it?
[20:32] <highvoltage> or perhaps it was just a fluke
[20:32] <highvoltage> the tracker still says rebuilding though?
[20:33] <stgraber> highvoltage: http://iso.qa.ubuntu.com/qatracker/milestones/214/builds/16053/testcases
[20:33] <stgraber> the one in Precise Final is disabled until the new one in Precise Pre-release is tested
[20:33] <highvoltage> ok
[20:33] <stgraber> once you post your results I'll replace the current one in Precise Final by the one currently in Pre-release
[20:33] <highvoltage> done
[20:34] <stgraber> (yeah I know kind of complicated, skaet's idea ;) I think we'll discuss that at UDS and decide not to do it again)