[09:19] <ev> stgraber: thanks for the ubiquity fixes!
[11:34] <CIA-45> ubiquity: evand * r4952 trunk/ (3 files in 3 dirs):
[11:34] <CIA-45> ubiquity: Manually connect the activate-link signal on the advanced
[11:34] <CIA-45> ubiquity: partitioning tool link on the automatic partitioning page.
[11:34] <CIA-45> ubiquity: GtkBuilder signal mapping appears to be broken (LP: #852054).
[14:14] <stgraber> cjwatson: hi! I was wondering, is there an easy way I can run netcfg in the exact same way d-i does without starting d-i entirely?
[14:14] <cjwatson> not really, I'm afraid
[14:14] <cjwatson> netcfg is quite close to the start of d-i though so it's not too time-consuming
[14:15] <stgraber> ok, back to trying to get d-i to run in lxc then ;)
[14:15] <cjwatson> you need a debconf frontend in place, it probably relies on details of cdebconf, and I bet it isn't robust against all sorts of things only true in d-i
=== bdmurray_ is now known as bdmurray
[14:57] <stgraber> yeah! got d-i to start in an lxc container ;)
[14:57] <stgraber> and netcfg seems to work fine in there
[15:11] <highvoltage> hmm, that's interesting. so someone could install an lxc host like it's a VM?
[15:11] <highvoltage> *guest, that is
[15:14] <stgraber> well, if you have a separate block device for the container, I guess it'd work yes
[15:14] <stgraber> in my case, I don't and I didn't allow the container to see any of my block devices, so I can install using iscsi on my iscsi server but that's about it ;)
[15:33] <ev> I'm working on a fix for bug 837406, which actually appears to be the partitioning page's fault.  At the same time, I'm trying to construct a test to ensure that the installer window does not grow beyond the minimum netbook resolution we've set as the lower bound as we hit these sizing bugs multiple times a release, and I'd like to catch them a lot sooner :)
[15:33] <ubot2> Launchpad bug 837406 in ubiquity "installer dialogue buttons fall off bottom of the screen on netbooks with small displays" [High,In progress] https://launchpad.net/bugs/837406
[15:43] <CIA-45> debian-installer: cjwatson * r1532 ubuntu/ (57 files in 2 dirs): Update help text translations from Launchpad.
[15:45] <CIA-45> debian-installer: cjwatson * r1533 ubuntu/debian/changelog: releasing version 20101020ubuntu64
[15:59] <cjwatson> ev: do you have any notion about bug 848938?  I looked at it a bit on Friday, but I couldn't find documentation of the NM API we're using
[15:59] <ubot2> Launchpad bug 848938 in ubiquity ""Download updates" unavailable after connecting to WiFi" [Critical,New] https://launchpad.net/bugs/848938
[15:59] <cjwatson> it's Critical and beta-2, so we need to either deal with it or explicitly defer it
[16:00] <ev> http://projects.gnome.org/NetworkManager//developers/api/09/spec.html
[16:00] <ev> I'll have a look after this page sizing stuff
[16:01] <ev> should be fairly straightforward
[16:01] <ev> might take the opportunity to better handle online/offline globally within ubiquity while I'm there
[16:01] <cjwatson> aha.  great, thanks
[16:02] <cjwatson> if you don't do an upload before you finish today, BTW, please let me know whether there's anything blocking me doing one, since I expect beta-2 CD builds to start tomorrow morning
[16:03] <CIA-45> ubiquity: cjwatson * r4955 trunk/debian/changelog: typo
[16:03] <ev> cjwatson: will do, thanks
[17:15] <stgraber> cjwatson: http://www.stgraber.org/download/installer01.tar.gz if you unpack that in /var/lib/lxc, then run rebuild.sh from /var/lib/lxc/installer01 and finally run lxc-start -n installer01, you'll get d-i running in a container
[17:15] <stgraber> cjwatson: everything seems to work fine as long as you don't try to partition anything (as the container isn't allowed access to block devices)
[17:16] <cjwatson> what's in that tarball?
[17:17] <stgraber> LXC config, a replacement init script, an initrd.gz from the installer (guess I could just have left a link to archive.u.c for that one), a preseed file and a script that unpacks the initrd and converts it into a container
[17:18] <stgraber> rebuild.sh basically unpacks the initrd into the rootfs directory (root of the container), then replaces /sbin/init by a script that tweaks the environment to simulate what we usually have on a cd and copies a preseed file to the / of the container
[17:19] <stgraber> lxc-start then simply starts that container showing you d-i executing the preseed. calling "lxc-stop -n installer01" from another shell is currently the only safe way of killing the container (reboot/halt don't work at the moment)
[17:23] <cjwatson> might be worth a target in debian-installer/build/Makefile to build it
[17:23] <cjwatson> or debian-installer/build/config/x86.cfg or whatever's appropriate
[17:24] <cjwatson> grabbed, thanks, may have a look at some point :)
[17:28] <CIA-45> ubiquity: evand * r4956 trunk/ (4 files in 4 dirs):
[17:28] <CIA-45> ubiquity: The first partitioning page does not translate its options ahead of
[17:28] <CIA-45> ubiquity: time, so ensure that the placeholder text isn't so long as to create
[17:28] <CIA-45> ubiquity: the installer window to grow excessively large before partitioning
[17:28] <CIA-45> ubiquity: is reached (LP: #837406).
[17:31] <stgraber> oh, that was the problem, interesting
[17:59] <CIA-45> ubiquity: evand * r4957 trunk/ (3 files in 3 dirs):
[17:59] <CIA-45> ubiquity: Use the new network-manager interface on the prepare page to disable
[17:59] <CIA-45> ubiquity: options that need an Internet connection when one is not available
[17:59] <CIA-45> ubiquity: (LP: #848938).
[18:24] <CIA-45> ubiquity: evand * r4958 trunk/ (151 files in 3 dirs): Update translations from Launchpad.
[18:26] <CIA-45> ubiquity: evand * r4959 trunk/ (d-i/manifest debian/changelog):
[18:26] <CIA-45> ubiquity: Automatic update of included source packages: flash-kernel
[18:26] <CIA-45> ubiquity: 2.28ubuntu32.
[18:30] <ev> flash-kernel is busted, uploading a fix now
[18:30] <ev> ubiquity's shell syntax checker caught it (yay)
[18:37] <CIA-45> ubiquity: evand * r4960 trunk/ (d-i/manifest debian/changelog): Update to fixed flash-kernel.
[18:39] <CIA-45> ubiquity: evand * r4961 trunk/tests/pyflakes.exclude: ignore unused import in tests/test_ubi_partman. It's needed by GtkBuilder.
[18:48] <stgraber> cjwatson: I'm finally doing some ipv6 tests with d-i. I'm getting a permission denied on dhcpv6-info as it's 644 instead of 755, I'm guessing that's a bug? :)
[18:49] <stgraber> (result is no working ipv6 when using dhcpv6 at least)
[18:49] <cjwatson> oops
[18:53] <CIA-45> netcfg: cjwatson * r1270 ubuntu/ (debian/changelog print-dhcp6c-info print-dhcpv6-info): Make print-dhcp6c-info and print-dhcpv6-info executable.
[18:53] <stgraber> cjwatson: when making them executable before starting netcfg, dhclient seems to run just fine but after that I still don't have an IPv6 address on the interface or DNS config in /etc/resolv.conf
[18:53] <stgraber> cjwatson: though all the details are correct in /var/lib/dhcp/dhclient6.leases
[18:55] <stgraber> syslog: http://paste.ubuntu.com/693245/
[18:55] <cjwatson> le sigh
[18:55] <stgraber> dhclient6.leases: http://paste.ubuntu.com/693246/
[18:56] <stgraber> dhclient.leases: http://paste.ubuntu.com/693249/
[18:56] <stgraber> resolv.conf: http://paste.ubuntu.com/693250/
[18:57] <cjwatson> hm, the syslog looks plausible enough
[18:58] <cjwatson> this usually requires strace
[18:58] <CIA-45> netcfg: cjwatson * r1271 ubuntu/debian/changelog: releasing version 1.68ubuntu3
[18:58] <cjwatson> last time I saw this it was the DHCP client tearing stuff down, but I thought I'd fixed that
[18:59] <stgraber> let me see if I can get the same happening in a container (that was in a VM), if I do, strace/gdb/whatever will be extremely easy :)
[19:00] <cjwatson> heh, yeah, though strace in d-i is easy enough
[19:00] <cjwatson> anna-install strace-udeb
[19:00] <cjwatson> or anna/choose_modules=strace-udeb as a boot parameter
[19:00] <CIA-45> ubiquity: evand * r4962 trunk/debian/changelog: releasing version 2.7.30
[19:00] <ev> right, home. Text me if the world catches fire.
[19:01] <cjwatson> k
[19:01] <cjwatson> thanks
[19:06] <stgraber> cjwatson: http://paste.ubuntu.com/693257/ (strace -fF of netcfg)
[19:08] <cjwatson> hmm, I misread the syslog anyway, print-dhcp6c-info apparently didn't print anything
[19:08] <cjwatson> ... or even print-dhcpv6-info
[19:09] <cjwatson> stgraber: I could use -v on strace as well
[19:10] <cjwatson> where the heck did my CD-RWs go?  having to use a DVD-RW and it takes forever to blank
[19:10] <stgraber> cjwatson: with -v http://paste.ubuntu.com/693261/
[19:11] <cjwatson> 14733 execve("/lib/netcfg/print-dhcpv6-info", ["/lib/netcfg/print-dhcpv6-info"], ["pid=14732", "reason=PREINIT6", "interface=eth0", "/build/buildd/isc-dhcp-4.1.1-P1/"...]) = 0
[19:11] <cjwatson> wonder if that's truncated
[19:12] <stgraber> cjwatson: I'm guessing print-dhcpv6-info isn't supposed to print "end" before printing useful stuff?
[19:13] <cjwatson> would sort of expect stuff like PATH
[19:13] <cjwatson> it prints end at the end
[19:13] <stgraber> well, for me it prints it at the beginning and at the end
[19:13] <cjwatson> it's the last line in the script ...
[19:13] <cjwatson> and your strace doesn't show it printing anything else
[19:14] <stgraber> running: "dhclient -6 -cf /varlib/netcfg/dhclient6.conf -sf /lib/netcfg/print-dhcpv6-info" gives me: http://paste.ubuntu.com/693263/
[19:14] <cjwatson> ARGH.  accidentally SIGQUITting dvd+rw-format is NOT GOOD
[19:14] <cjwatson> one coaster
[19:14] <cjwatson> in fact one unejectable coaster, eek
[19:15] <cjwatson> oh, it's still running in the background I think
[19:15] <cjwatson> hate hardware
[19:16] <cjwatson> oh, that's interesting, it runs print-dhcpv6-info twice
[19:16] <cjwatson> I think print-dhcpv6-info needs to check $reason
[19:17] <cjwatson> try http://paste.ubuntu.com/693267/ ?
[19:20] <stgraber> ok, I only see it run once now, though I still don't get an IP on eth0 and resolv.conf still only has ipv4 dns servers
[19:21] <cjwatson> can I have an updated strace?
[19:21] <stgraber> cjwatson: http://paste.ubuntu.com/693273/
[19:22]  * stgraber needs to make a udeb for pastebinit (though that probably means porting it to C ;))
[19:22] <CIA-45> netcfg: cjwatson * r1272 ubuntu/ (debian/changelog print-dhcpv6-info):
[19:22] <CIA-45> netcfg: print-dhcpv6-info: Exit immediately if reason is not BOUND6, RENEW6, or
[19:22] <CIA-45> netcfg: REBIND6.
[19:26] <cjwatson> urgh, truncated strings
[19:26] <cjwatson> stgraber: (a) for subsequent straces could I have -s1024 as well (b) could I have an updated syslog?
[19:26] <cjwatson> silly strace defaults
[19:26] <cjwatson> 14921 sendto(7, "<15>Sep 19 19:20:59 netcfg[14921"..., 84, MSG_NOSIGNAL, NULL, 0) = 84
[19:26] <cjwatson> ^- tantalising
[19:27] <stgraber> cjwatson: http://paste.ubuntu.com/693275/ (I wiped the syslog just before my last netcfg test)
[19:28] <CIA-45> netcfg: cjwatson * r1273 ubuntu/ (autoconfig.c debian/changelog):
[19:28] <CIA-45> netcfg: Don't bother unlinking /var/lib/netcfg/dhcp6c-finished unless we're
[19:28] <CIA-45> netcfg: using dhcp6c.
[19:29] <cjwatson> Sep 19 19:21:06 main-menu[332]: (process:14919): ip: RTNETLINK answers: File exists
[19:29] <cjwatson> hmm, may not be relevant
[19:29] <stgraber> cjwatson: new strace http://paste.ubuntu.com/693278/
[19:29] <cjwatson> wait, with ISC-DHCP the dhclient-script is meant to *actually bring the interface up* isn't it?
[19:30] <cjwatson> I think WIDE was entirely different
[19:30] <cjwatson> that would explainn the lack of IPv6 address but not the lack of IPv6 nameservers
[19:31] <stgraber> oh, indeed. Just ran dhclient with the -sf and everything got set just fine (now to reset everything so we can still debug ;))
[19:31] <cjwatson> oh, in fact with ISC, dhclient-script deals with resolv.conf as well
[19:31] <cjwatson> sigh
[19:32] <cjwatson> I suspect this may not have worked properly with WIDE either :-/
[19:35] <cjwatson> however, it's not very obvious what to do if both DHCPv6 and DHCPv4 give us nameservers
[19:35] <cjwatson> the stock dhclient-script keeps nameservers from only the second one to run
[19:36] <cjwatson> though isc's udeb script could use some improvements on that front
[19:37] <stgraber> ideally you'd also like to handle the case where you get 3 DNS servers from either dhcpv4 or dhcpv6 so that the first 3 nameserver entries in resolv.conf contain a mix of ipv4 and ipv6
[19:42] <cjwatson> so I guess we want http://paste.ubuntu.com/693284/ in netcfg and http://paste.ubuntu.com/693285/ in isc-dhcp, at least to start with
[19:42] <cjwatson> it doesn't handle combining results from the two, but it should be better than what we have now
[19:43] <stgraber> yeah, I guess that'd be a good start. Can you paste the complete files so I can copy them to my VM for testing?
[19:46] <cjwatson> and is closer to the current main dhclient-script
[19:46] <cjwatson> (I couldn't bear to copy the horrible sed thing, and I doubt it would work with busybox sed anyway.  grep -i is much simpler)
[19:47] <stgraber> yeah, I also noticed that d-i's busybox doesn't have an ipv6 aware version of ping, doesn't really help for automated testing :) at least netcat seems fine
[19:47] <cjwatson> http://paste.ubuntu.com/693290/
[19:47] <cjwatson> http://paste.ubuntu.com/693291/
[19:47] <cjwatson> netcfg and isc-dhcp respectively
[19:52] <stgraber> ok, I have an IPv6! and resolv.conf is the one received from dhcpv4
[19:53] <cjwatson> whee.  that's the best we'll manage for beta-2 IMO
[19:53] <cjwatson> anything else that should be rolled in at the same time?
[19:53] <stgraber> if you can wait 10 more minutes, I can quickly test all my testcases on that patched netcfg
[19:54] <cjwatson> ok, sure
[19:54] <cjwatson> I'll go downstairs then, back in a while
[20:04] <stgraber> cjwatson: http://paste.ubuntu.com/693299/
[20:05] <stgraber> cjwatson: so only problem I saw happening (just once) was eth0 being brought down by something
[20:09] <stgraber> trying a v6-only install now to make sure there isn't something else horribly broken later in the installer
[20:19] <CIA-45> ubiquity: evand * r4963 ubiquity/ (d-i/update-control debian/changelog debian/control): Build-depend on the GTK3 version of libindicator.
[20:22] <stgraber> yeah, got ubuntu server to install on a v6-only network! will have to try again once we have these fixes in a new netcfg as that VM is far from clean ;)
[20:39] <cjwatson> OK, that sounds a hell of a lot better than the prior state, so I'll go ahead and upload that lot
[20:39] <cjwatson> thanks!
[20:39] <CIA-45> netcfg: cjwatson * r1274 ubuntu/ (debian/changelog print-dhcpv6-info):
[20:39] <CIA-45> netcfg: print-dhcpv6-info: Actually apply IPv6 address and resolv.conf entries
[20:39] <CIA-45> netcfg: received from the DHCPv6 server.
[20:42] <CIA-45> netcfg: cjwatson * r1275 ubuntu/debian/changelog: releasing version 1.68ubuntu4
[21:03] <CIA-45> apt-setup: cjwatson * r222 ubuntu/debian/ (changelog control):
[21:03] <CIA-45> apt-setup: Make apt-setup-udeb Architecture: any so that we can use different
[21:03] <CIA-45> apt-setup: multiarch defaults on different architectures.
[21:05] <CIA-45> apt-setup: cjwatson * r223 ubuntu/debian/changelog: last commit fixes LP: #853738
[21:06] <CIA-45> apt-setup: cjwatson * r224 ubuntu/debian/changelog: releasing version 1:0.49ubuntu8
[21:42] <infinity> Anyone fixing the ubiquity FTBFS?
[21:43] <infinity> I'll take that as a "no, please fix".
[21:45] <infinity> Or..
[21:45] <infinity> ev: You already have a fix comitted?
[21:45] <infinity> ev: Is that being tested and uploaded?
[22:19] <CIA-45> apt-setup: cjwatson * r225 ubuntu/debian/ (changelog rules): Fix non-i386 builds.
[22:20] <CIA-45> apt-setup: cjwatson * r226 ubuntu/debian/changelog: releasing version 1:0.49ubuntu9
[22:20] <cjwatson> infinity: 20:00 <ev> right, home. Text me if the world catches fire.
[22:20] <cjwatson> (3h20m ago)
[22:21] <infinity> cjwatson: Ahh, I wasn't in the channel to see that.
[22:21] <cjwatson> infinity: can you pastebin the test failure?
[22:21] <cjwatson> yeah, I noticed you joined later
[22:21] <infinity> cjwatson: Though, you'd think he was aware of the FTBFS, since he committed a fix.
[22:21] <cjwatson> maybe he didn't notice the subsequent failure
[22:22] <infinity> And I think I closed that terminal...  It's been a long day.
[22:23] <infinity> Is there a make target to run the testsuite?
[22:23] <infinity> Ahh, tests.
[22:23] <infinity> I think.
[22:23] <cjwatson> 'debian/rules tests', I think, although I can't vouch for its dependencies
[22:24] <cjwatson> look not too bad
[22:25] <infinity> http://lucifer.0c3.net/~adconrad/tests.log
[22:26] <infinity> Just looks like a seriously self-hating testsuite.
[22:38] <cjwatson> hm, mysterious
[22:38]  * cjwatson fires at pbuilder
[22:38] <cjwatson> I mean it basically means some bit early on fell over, but not clear what
[22:39] <infinity> I blame... Everything.
[22:39] <infinity> People shouldn't be expected to think on Monday.
[22:41] <cjwatson> that's OK, most people don't
[22:42] <infinity> *rimshot*
[22:43] <cjwatson> grr.  $ sudo pbuilder --update
[22:43]  * cjwatson yawns
[23:12] <cjwatson> infinity: OK, I think I remember how to fix this kind of thing
[23:18]  * cjwatson cargo-cults code to mock plugin methods
[23:58] <CIA-45> ubiquity: cjwatson * r4964 trunk/ (4 files in 4 dirs):
[23:58] <CIA-45> ubiquity: Move setup_power_watch out of the ubi-prepare plugin into a new
[23:58] <CIA-45> ubiquity: ubiquity.upower module, allowing us to patch it out more easily while
[23:58] <CIA-45> ubiquity: running tests. Patch ubiquity.nm.add_connection_watch and
[23:58] <CIA-45> ubiquity: ubiquity.nm.has_connection in the GTK frontend tests as well.