| mario_limonciell | evand, glancing through a pruned diff on the logs, i'm baffled at how ubiquity.components.apt_setup.AptSetup doesn't get called in noninteractive | 00:49 |
|---|---|---|
| mario_limonciell | my gut would say that some type of error isn't being caught. have you narrowed it down to a particular call that fails or anything? | 00:50 |
| mario_limonciell | by dropping in debugging print statements etc around that configure_apt() function as it drops down into the ubiquity component? | 00:51 |
| evand | superm1: it's not getting called for you? It's getting call for me. | 03:13 |
| evand | called* | 03:14 |
| superm1 | evand, well if its being called, it's not indicating it in syslog | 04:01 |
| superm1 | i didnt add any debug code in or anything | 04:01 |
| superm1 | just compared basic logs | 04:01 |
| evand | hrm. | 04:08 |
| superm1 | is yours indicated in syslog? | 04:08 |
| evand | Are you referring to the starting up messages? | 04:12 |
| superm1 | there is a message when the apt-setup shell script gets called | 04:13 |
| superm1 | that is logged to syslog | 04:13 |
| superm1 | that message is not present when launched with noninteractive | 04:14 |
| superm1 | you know - come to think of it - this exact issue plagued the mythbuntu frontend too during gutsy | 04:18 |
| superm1 | never identified it completely | 04:18 |
| superm1 | well actually no it wasn't identical, it was just cdrom sources that didnt work | 04:19 |
| superm1 | but ended up adding an extra manual call to apt-cdrom | 04:19 |
| superm1 | but it might be one in the same, because network sources may not be available in noninteractive | 04:19 |
| superm1 | have you tried to run dhclient in parallel at all as soon as ubiquity/noninteractive starts? | 04:20 |
| evand | the starting up messages aren't there because I disable debug logging in noninteractive (re-enabling it doesn't help, I just tried). As far as network sources go, I have an active network connection (I'm testing from the desktop), but during that part of the install (for all frontends) network sources are commented out. I'll try re-enabling them later. | 04:28 |
| evand | oh hrm, I still don't get that message for AptSetup though. I'll look into that as well. | 04:29 |
| evand | I know it runs though. | 04:30 |
| superm1 | well when you try to re-enable them, just remember since you don't have NM running, you may not have an address and manually need to run dhclient and/or set something static | 04:32 |
| evand | but I do have n-m running. | 04:36 |
| evand | I'm running ubiquity from the desktop | 04:36 |
| evand | by typing ubiquity noninteractive in gnome-terminal | 04:37 |
| superm1 | oh running noninteractive from the desktop | 04:37 |
| evand | indeed | 04:37 |
| superm1 | for some reason i didn't think of doing it that way :) | 04:37 |
| superm1 | that's probably a heck of a lot easier to debug from | 04:37 |
| evand | exactly | 04:37 |
| tjaalton | hmm, d-i consoles become useless (no characters visible) after a VT change | 09:13 |
| tjaalton | restarting the console works, until the next VT change | 09:14 |
| xivulon | are today's dailys/upgrades good? | 09:35 |
| xivulon | TheMuso, any insight in the mounting thing? | 10:45 |
| === cjwatson_ is now known as cjwatson | ||
| evand | xivulon: bug 202169 . Can you provide links to the discussion you had with the artwork team and send a heads up email to ubuntu-doc@lists.ubuntu.com and ubuntu-translators@lists.ubuntu.com ? | 14:43 |
| ubotu | Launchpad bug 202169 in ubuntu "Freeze exception: wubi and umenu string and artwork changes" [Undecided,New] https://launchpad.net/bugs/202169 | 14:43 |
| xivulon | evand added a comment to that, see if it is enough | 14:58 |
| xivulon | at the end I did not change the logos anyway, since I asked the artwork team several times to provide url/path for svg and they didn't | 14:59 |
| xivulon | I only "zoomed" the vertical image as instructed by kwwii and generated proper windows icons (with multiple layers). | 14:59 |
| evand | ah, ok | 14:59 |
| xivulon | windows "icons" means the .ico file used in taskbar and windows explorer | 15:00 |
| evand | indeed | 15:00 |
| superm1 | evand, if the solution for noninteractive isn't in place yet, do you think just adding a script into our late scripts would be an adequate solution for now? | 15:01 |
| superm1 | we need to test some of our stuff destined that won't run without noninteractive and see if locale support works out properly and such | 15:01 |
| evand | yes, I think that's perfectly reasonable | 15:05 |
| superm1 | oem-config-gtk, languages, and what else get installed typically then? | 15:05 |
| evand | oem-config, oem-config-gtk, localechooser-data | 15:10 |
| xivulon | evand bean123 just release the new grub4dos code with the bugfixes | 15:10 |
| xivulon | I will test and upload any changes tonight | 15:11 |
| evand | xivulon: fantastic, thanks! | 15:11 |
| xivulon | I also plan to merge in the available translations | 15:11 |
| cr3 | is there a way to install xorg-driver-fglrx automatically from the alternate image? instead, people would have to reboot from the console on some machines with ATI graphics controllers, run apt-get, then modify xorg.conf. I'm trying to improve the user experience in these situations. | 15:11 |
| xivulon | evand can you have a quick look at bug #201750 ? | 15:12 |
| ubotu | Launchpad bug 201750 in wubi "Cannot remount loopfiles inside of vfat" [High,Confirmed] https://launchpad.net/bugs/201750 | 15:12 |
| superm1 | cr3, you can handle it in a late command if you are preseeding | 15:13 |
| xivulon | my take on 201750 is: unless you know about some ways to remount the loopdevices the workaround should be applied before beta | 15:14 |
| cr3 | superm1: might there be a way to edit the preseed when installing from the cdrom? | 15:14 |
| superm1 | cr3, you'd have to provide the seed in an external means | 15:15 |
| superm1 | or remaster the disk in some sense to include it | 15:15 |
| cjwatson | cr3: user experience> our preferred solution is that it installs vesa, vesa copes (or is fixed to cope), and jockey displays a message saying "we can't drive your hardware very well, but this restricted driver can" | 15:15 |
| cjwatson | you can just pass pkgsel/include on the kernel command line from the CD though | 15:15 |
| cr3 | cjwatson: if vesa can't handle the graphics controller, is that worth reporting a bug? | 15:16 |
| cjwatson | absolutely | 15:16 |
| cr3 | cjwatson: will do, thanks | 15:16 |
| superm1 | cjwatson, unfortunately some of the RadeonHD cards won't handle at this point even in vesa | 15:16 |
| cr3 | superm1: good call! that's exactly what I'm working on :) | 15:16 |
| cjwatson | or indeed, in general if X can't autoconfigure something in the default install without free drivers, that's a bug | 15:16 |
| evand | xivulon: remounting NTFS ro works for me. | 15:16 |
| superm1 | 3450 and 3650 particularly | 15:16 |
| cjwatson | er, without non-free drivers | 15:16 |
| cr3 | I think mine is a X1200 | 15:16 |
| xivulon | evand might be the new ntfs-3g | 15:16 |
| cjwatson | superm1: right, but I gather that's something AMD care about and it ought not to be rocket science to fix | 15:16 |
| xivulon | did not test ntfs, only vfat lately | 15:17 |
| superm1 | but yeah, cr3 yeah please file a bug | 15:17 |
| superm1 | i'll add some feedback to it on the cards i've got as well that dont go | 15:17 |
| evand | ntfs-3g 1.2216 | 15:17 |
| evand | but indeed | 15:17 |
| xivulon | evand there are 2 separate issues: remounting ntfs-3g, remounting loopfiles (whether on vfat or other fs) | 15:17 |
| xivulon | #201750 involves the second issue | 15:17 |
| xivulon | I seemed to remember that splitting mount -o loop in losetip + mount /dev/loop# would do the trick | 15:18 |
| evand | ah, ok | 15:18 |
| xivulon | but for some reason it does not work in my tests (see bug) | 15:18 |
| xivulon | strange thing is that TheMuso mentione that he gets a rw root using explicit losetup! | 15:18 |
| xivulon | I cannot reproduce that | 15:19 |
| xivulon | will try again tonight | 15:19 |
| evand | ok | 15:21 |
| evand | if we do end up supporting fat32, I'd like to see it get more testing than it is currently. Unless I am mistaken, the code path for separate partitions in Wubi isn't nearly as well tested as a root and swap disk. | 15:23 |
| xivulon | well if that go sour I can always restrict everything to a single 4GB root... | 15:23 |
| xivulon | but we used that since 7.04 and nobody complained so far, also we had a separate /home for a long time | 15:24 |
| xivulon | cjwatson, iirc we went through some loopdevice remounting issues also in 7.10, and you suggested to try some command which was not losetup... | 15:26 |
| cjwatson | I don't remember - maybe try irclogs.ubuntu.com? | 15:26 |
| xivulon | will have a look | 15:27 |
| xivulon | http://irclogs.ubuntu.com/2007/09/07/%23ubuntu-installer.txt | 15:30 |
| mario_limonciell | evand, looking at the manifest for today's live disk, ubiquity 1.7.18 isn't included, is this correct? The image was mastered today but that manifest is two days old | 15:35 |
| evand | mario_limonciell: when you see a manifest that's behind the rest of the CD, that means that something isn't installable in the livefs build | 15:36 |
| evand | one second | 15:36 |
| mario_limonciell | (we hit that need for the signal import testing another locale) | 15:36 |
| evand | http://people.ubuntu.com/~ubuntu-archive/livefs-build-logs/hardy/ubuntu/20080314/livecd-20080314-amd64.out | 15:36 |
| xivulon | cjwatson: I think the command I you mentioned was blockdev --setrw http://irclogs.ubuntu.com/2007/08/30/%23ubuntu-installer.txt | 15:37 |
| mario_limonciell | oh so it looks like hwdb-client-gnome is in universe not main. | 15:38 |
| cjwatson | xivulon: mm, that fiddles with the kernel's idea of the writability of the underlying block device, which isn't the same as the read-only-ness of the mount | 15:38 |
| cjwatson | mario_limonciell: replaced with hwtest; I think ubuntu-meta's since been uploaded | 15:38 |
| cjwatson | yep, it has | 15:38 |
| mario_limonciell | cjwatson, could you guys remaster the daily with that in today when it clears the archive? | 15:39 |
| cjwatson | yeah, should be doable | 15:40 |
| mario_limonciell | great thanks | 15:40 |
| evand | it has already cleared the archive, no? | 15:40 |
| mario_limonciell | well my mirror here is always out of sync by a few hours, so i'm not sure if its on a.u.c yet | 15:44 |
| xivulon | I think I have a theory on the remount thing: | 15:45 |
| xivulon | mount -o remount,rw /host && losetup /dev/loop5 /host/loop_file && mount -o ro /dev/loop5 /root | 15:45 |
| xivulon | now /root should be remountable | 15:45 |
| xivulon | I was missing the first command... | 15:45 |
| xivulon | the catch though is that I do not think you can now remount /host ro... | 15:46 |
| xivulon | unless you delete the loop device... | 15:46 |
| xivulon | which we should do in umountroot | 15:46 |
| xivulon | hmm better not | 15:46 |
| xivulon | the issue is that you cannot change the ro/rw status of the loop device itself... | 15:47 |
| xivulon | that is where blockdev --setrw might help | 15:48 |
| cjwatson | evand: go ahead and rebuild it if you've checked | 15:56 |
| evand | cjwatson: will do | 15:56 |
| cjwatson | ah, yes, it's in | 15:56 |
| xivulon | cjwatson do you think that applying blockdev --setrw in the context above would make sense? | 15:59 |
| cjwatson | it's worth a try | 16:00 |
| cjwatson | but I don't magically know whether it'll work - you will have to try it :-) | 16:00 |
| xivulon | cannot do it now | 16:00 |
| xivulon | the test is: mount -o loop loop_file_with_ext3 /mnt | 16:00 |
| xivulon | herm mount -o ro,loop loop_file_with_ext3 /mnt | 16:01 |
| xivulon | then blockdev --setrw | 16:01 |
| xivulon | then mount -o remount,rw /mnt | 16:01 |
| xivulon | and the other way around for ro, so that the host device can also be remounted ro (required by umountroot) | 16:02 |
| xivulon | evand, if ntfs can be remounted you should merge in the patch in bug #186114 | 16:23 |
| ubotu | Launchpad bug 186114 in wubi "umountroot cannot handle fuse host mounts" [Medium,Fix committed] https://launchpad.net/bugs/186114 | 16:23 |
| xivulon | ...and merge new lupin code (that fixes mounthost). | 16:23 |
| evand | will do | 16:26 |
| xivulon | of course we will also need to patch umountroot to use blockdev --setro when appropriate (assuming the above holds) | 16:27 |
| xivulon | have added a comment as reminder | 16:32 |
| xivulon | something along the lines of: for d in /dev/loop*; do blockdev --setro "$d" || true; done | 16:35 |
| xivulon | ah it looks like the blockdev trick works! thanks cjwatson for the (last year) tip! | 16:49 |
| cjwatson | heh | 16:49 |
| cjwatson | you're welcome | 16:49 |
| xivulon | with that we can close several bugs! | 16:49 |
| xivulon | good thing I listen carefully to what you say... :P | 16:50 |
| xivulon | evand forget about lupin and umountroot for the time being, I will need to edit a few things to make good use of blockedv | 17:19 |
| evand | ok, fair enough | 17:20 |
| evand | mario_limonciell: forgot to mention, new CDs are up. Sorry about that. | 20:12 |
| mario_limonciell | evand, yeah i just saw by checking here myself. our factory guy just left a few minutes ago, so i'll just be able to experiment with a local install on it | 20:18 |
| CIA-24 | ubiquity: evand * r2568 ubiquity/ (4 files in 3 dirs): | 20:26 |
| CIA-24 | ubiquity: * Change --no-continue option to a more general --only option for use | 20:26 |
| CIA-24 | ubiquity: in only-ubiquity and automatic-ubiquity modes. | 20:26 |
| CIA-24 | ubiquity: * Disable the minimize button when using --only (LP: #200034). | 20:26 |
| xivulon | TheMuso, will upload soon a new mounthost script, would appreciate if you could test it | 21:46 |
| TheMuso | xivulon: Sure, likely not this weekend, but I can certainly have a look first thing Monday morning. | 21:46 |
| xivulon | great, not sure if you read the logs of today discussion | 21:48 |
| TheMuso | xivulon: I scanned voer them, but I'll read them more thoroughly later. | 21:49 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!