| ev | stgraber: thanks for the ubiquity fixes! | 09:19 |
|---|---|---|
| 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). | 11:34 |
| 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 | 14:14 |
| 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 | 14:15 |
| === bdmurray_ is now known as bdmurray | ||
| stgraber | yeah! got d-i to start in an lxc container ;) | 14:57 |
| stgraber | and netcfg seems to work fine in there | 14:57 |
| highvoltage | hmm, that's interesting. so someone could install an lxc host like it's a VM? | 15:11 |
| highvoltage | *guest, that is | 15:11 |
| 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 ;) | 15:14 |
| 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 | 15:33 |
| CIA-45 | debian-installer: cjwatson * r1532 ubuntu/ (57 files in 2 dirs): Update help text translations from Launchpad. | 15:43 |
| CIA-45 | debian-installer: cjwatson * r1533 ubuntu/debian/changelog: releasing version 20101020ubuntu64 | 15:45 |
| 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 | 15:59 |
| ev | http://projects.gnome.org/NetworkManager//developers/api/09/spec.html | 16:00 |
| ev | I'll have a look after this page sizing stuff | 16:00 |
| 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 | 16:01 |
| 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 | 16:02 |
| CIA-45 | ubiquity: cjwatson * r4955 trunk/debian/changelog: typo | 16:03 |
| ev | cjwatson: will do, thanks | 16:03 |
| 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) | 17:15 |
| cjwatson | what's in that tarball? | 17:16 |
| 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 | 17:17 |
| 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 | 17:18 |
| 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) | 17:19 |
| 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 | 17:23 |
| cjwatson | grabbed, thanks, may have a look at some point :) | 17:24 |
| 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). | 17:28 |
| stgraber | oh, that was the problem, interesting | 17:31 |
| 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). | 17:59 |
| CIA-45 | ubiquity: evand * r4958 trunk/ (151 files in 3 dirs): Update translations from Launchpad. | 18:24 |
| 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. | 18:26 |
| ev | flash-kernel is busted, uploading a fix now | 18:30 |
| ev | ubiquity's shell syntax checker caught it (yay) | 18:30 |
| CIA-45 | ubiquity: evand * r4960 trunk/ (d-i/manifest debian/changelog): Update to fixed flash-kernel. | 18:37 |
| CIA-45 | ubiquity: evand * r4961 trunk/tests/pyflakes.exclude: ignore unused import in tests/test_ubi_partman. It's needed by GtkBuilder. | 18:39 |
| 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? :) | 18:48 |
| stgraber | (result is no working ipv6 when using dhcpv6 at least) | 18:49 |
| cjwatson | oops | 18:49 |
| 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 | 18:53 |
| stgraber | syslog: http://paste.ubuntu.com/693245/ | 18:55 |
| cjwatson | le sigh | 18:55 |
| stgraber | dhclient6.leases: http://paste.ubuntu.com/693246/ | 18:55 |
| stgraber | dhclient.leases: http://paste.ubuntu.com/693249/ | 18:56 |
| stgraber | resolv.conf: http://paste.ubuntu.com/693250/ | 18:56 |
| cjwatson | hm, the syslog looks plausible enough | 18:57 |
| 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 | 18:58 |
| 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 :) | 18:59 |
| 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. | 19:00 |
| cjwatson | k | 19:01 |
| cjwatson | thanks | 19:01 |
| stgraber | cjwatson: http://paste.ubuntu.com/693257/ (strace -fF of netcfg) | 19:06 |
| cjwatson | hmm, I misread the syslog anyway, print-dhcp6c-info apparently didn't print anything | 19:08 |
| cjwatson | ... or even print-dhcpv6-info | 19:08 |
| cjwatson | stgraber: I could use -v on strace as well | 19:09 |
| 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/ | 19:10 |
| 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 | 19:11 |
| stgraber | cjwatson: I'm guessing print-dhcpv6-info isn't supposed to print "end" before printing useful stuff? | 19:12 |
| 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 | 19:13 |
| 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 | 19:14 |
| cjwatson | oh, it's still running in the background I think | 19:15 |
| cjwatson | hate hardware | 19:15 |
| cjwatson | oh, that's interesting, it runs print-dhcpv6-info twice | 19:16 |
| cjwatson | I think print-dhcpv6-info needs to check $reason | 19:16 |
| cjwatson | try http://paste.ubuntu.com/693267/ ? | 19:17 |
| 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 | 19:20 |
| cjwatson | can I have an updated strace? | 19:21 |
| stgraber | cjwatson: http://paste.ubuntu.com/693273/ | 19:21 |
| * 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. | 19:22 |
| 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 | 19:26 |
| stgraber | cjwatson: http://paste.ubuntu.com/693275/ (I wiped the syslog just before my last netcfg test) | 19:27 |
| 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. | 19:28 |
| 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? | 19:29 |
| 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 | 19:30 |
| 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 | 19:31 |
| cjwatson | I suspect this may not have worked properly with WIDE either :-/ | 19:32 |
| 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 | 19:35 |
| cjwatson | though isc's udeb script could use some improvements on that front | 19:36 |
| 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 | 19:37 |
| 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 | 19:42 |
| 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? | 19:43 |
| 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) | 19:46 |
| 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 | 19:47 |
| stgraber | ok, I have an IPv6! and resolv.conf is the one received from dhcpv4 | 19:52 |
| 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 | 19:53 |
| cjwatson | ok, sure | 19:54 |
| cjwatson | I'll go downstairs then, back in a while | 19:54 |
| stgraber | cjwatson: http://paste.ubuntu.com/693299/ | 20:04 |
| stgraber | cjwatson: so only problem I saw happening (just once) was eth0 being brought down by something | 20:05 |
| stgraber | trying a v6-only install now to make sure there isn't something else horribly broken later in the installer | 20:09 |
| CIA-45 | ubiquity: evand * r4963 ubiquity/ (d-i/update-control debian/changelog debian/control): Build-depend on the GTK3 version of libindicator. | 20:19 |
| 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 ;) | 20:22 |
| 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. | 20:39 |
| CIA-45 | netcfg: cjwatson * r1275 ubuntu/debian/changelog: releasing version 1.68ubuntu4 | 20:42 |
| 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. | 21:03 |
| CIA-45 | apt-setup: cjwatson * r223 ubuntu/debian/changelog: last commit fixes LP: #853738 | 21:05 |
| CIA-45 | apt-setup: cjwatson * r224 ubuntu/debian/changelog: releasing version 1:0.49ubuntu8 | 21:06 |
| infinity | Anyone fixing the ubiquity FTBFS? | 21:42 |
| infinity | I'll take that as a "no, please fix". | 21:43 |
| infinity | Or.. | 21:45 |
| infinity | ev: You already have a fix comitted? | 21:45 |
| infinity | ev: Is that being tested and uploaded? | 21:45 |
| CIA-45 | apt-setup: cjwatson * r225 ubuntu/debian/ (changelog rules): Fix non-i386 builds. | 22:19 |
| 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) | 22:20 |
| 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 | 22:21 |
| infinity | And I think I closed that terminal... It's been a long day. | 22:22 |
| 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 | 22:23 |
| cjwatson | look not too bad | 22:24 |
| infinity | http://lucifer.0c3.net/~adconrad/tests.log | 22:25 |
| infinity | Just looks like a seriously self-hating testsuite. | 22:26 |
| 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 | 22:38 |
| infinity | I blame... Everything. | 22:39 |
| infinity | People shouldn't be expected to think on Monday. | 22:39 |
| cjwatson | that's OK, most people don't | 22:41 |
| infinity | *rimshot* | 22:42 |
| cjwatson | grr. $ sudo pbuilder --update | 22:43 |
| * cjwatson yawns | 22:43 | |
| cjwatson | infinity: OK, I think I remember how to fix this kind of thing | 23:12 |
| * cjwatson cargo-cults code to mock plugin methods | 23:18 | |
| 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. | 23:58 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!