20070412
Log
TZ UTC+1
10:59 mdz cjwatson,Keybuk: is everyone here who's expected?
11:00 heno here
11:00 Riddell hi
11:00 asac i am here ... in case i am expected ;)
11:00 kwwii hi
11:00 iwj Hello everyone ...
11:00 mvo hello
11:01 doko seems to be a slow start today =)
11:01 cjwatson BenC said he wanted to be here, but he may have keeled over with exhaustion
11:01 Keybuk mdz: yes
11:01 fabbione it's also 3/4am there
11:01 cjwatson everyone else is here
11:01 cjwatson fabbione: he was aware of that
11:02 mdz fabbione: 5, but wasn't expecting him at this meeting
11:02 pitti I saw him an hour ago
11:02 mdz BenC: good morning
11:02 fabbione here he is
11:02 cjwatson 00:47 <BenC> would this meeting be opportune time to discuss the kernel upload?
11:02 cjwatson 00:47 <cjwatson> I certainly won't hold it against you if you aren't there, as long as I have all the information needed to represent kernel interests
11:02 cjwatson 00:47 <cjwatson> yes
11:02 cjwatson 00:47 <BenC> then I'll make it
11:02 cjwatson aha
11:02 BenC good morning
11:02 mdz so, welcome everyone. thanks for being available even though this was a bit last-minute
11:03 mdz the release is creeping up on us, and I think we're a little bit behind where we ought to be right now, so it seemed wise to get everyone together to go over the outstanding issues
11:03 mdz Keybuk and Mithrandir have been reviewing our progress
11:04 mdz Keybuk: would you give an overview of where we stand?
11:04 Keybuk Current status is that the release candidate images prepared yesterday had several problems with them
11:04 Keybuk notably things like the "About Ubuntu" menu option being missing
11:04 pitti #105593 :(
11:04 Keybuk and /etc/lsb, /etc/issue, etc. all being wrong
11:05 pitti ^ is the worst one I found
11:05 seb128 about ubuntu, launchpad integration have been fixed yesterday
11:05 seb128 bug #105593
11:05 Mithrandir seb128: and accepted into the archive this morning.
11:05 Keybuk we also haven't performed several of the steps leading to the candidate, including hardware certification testing
11:05 seb128 Mithrandir: rock on
11:05 Keybuk fabbione: I believe you've got that started now?
11:05 mdz (incidentally, is anyone else having trouble with firefox stalling for a very long time at "Connecting to launchpad.net..."?)
ubotu Malone bug 105593 in restricted-manager "claims that nvidia is in use on live system" [Undecided,Confirmed] https://launchpad.net/bugs/105593
11:05 fabbione Keybuk: yes already started. we got only one bug so far
11:05 Keybuk mdz: yes, problem is at LP end
11:06 seb128 mdz: yeah, it's really really really slow since yesterday :/
11:06 fabbione Keybuk: i think Marc has almost completed the first round
11:07 mdz pitti: the lrm change fixes the immediate problem, right?
11:07 asac mdz: what do you mean by stalled? like no redraw of ui ... or just waiting for server?
11:07 pitti mdz: right, there are actually two problems
11:07 pitti (1) r-m warns you about non-free drivers being in use, taints kernel, prone to raise political discussions
11:07 pitti (2) as soon as an user actually enables nvidia, it is going to break their X because of ABI incompatibility (except if they have a very *very* new card which is not supported by nvidia-glx)
11:07 mdz asac: I did some packet level debugging which I posted to the launchpad list; let's talk about it later
11:07 asac mdz: k
11:08 pitti mdz: thus we would break many X.orgs instead of showing off compiz&co for nvidia users :(
11:08 Mithrandir pitti: but the lrm fix in the queue fixes this, right?
11:08 pitti right
11:08 pitti it's not built yet, though
11:09 Mithrandir I'm driving the publisher by hand now
11:09 pitti but I saw that the source is in the archive
11:09 pitti 'no builds recorded' instead of 'needs build', hmm
11:10 mdz fabbione: is there any estimate of how long the cert tests will take
11:10 fabbione mdz: Marc is working hard to catch up. he promised to be as fast as he can, but no numbers for an ETA
11:10 fabbione mdz: i expect at least another day
11:10 pitti so, we have the l-r-m problem, the hwdb notification, about ubuntu menu items, lsb; we can probably live with any single one of them, but all together look really bad IMHO
11:11 mdz fabbione: I think it is irrelevant for now; either there will be no problems discovered and we can release, or there will be problems and we'll need to delay
11:12 fabbione mdz: as i said.. one bug found so far in n-m already reported by other people. I will be able to tell you more in a few hours when Montreal wakes up
11:12 fabbione mdz: i didn't know about the meeting till this morning, otherwise i would have gathered more info
11:12 seb128 pitti: hwdb and about ubuntu are supposed to be fixed now, no?
11:13 pitti seb128: right, should be in the archive
11:13 mdz fabbione: bug number?
11:13 Mithrandir seb128: they are, if you run apt-get update.
11:13 seb128 Mithrandir: ok, just making sure because pitti listed them
11:13 Mithrandir note however that ubuntu-docs still have "feisty fawn" in them in some places.
11:13 fabbione mdz: https://bugs.launchpad.net/ubuntu/+source/network-manager/+bug/105688
11:13 pitti seb128: argh, no, hwdb-client is still the old one
11:13 Mithrandir (with the version numbers, mind, so it's less bad)
ubotu Malone bug 105688 in network-manager "NetworkManager applet shows No network connection when there is a network connection (dup-of: 105234)" [Undecided,Needs info]
11:13 ubotu Malone bug 105234 in network-manager "Netowrk manager says disconnected but is connected and working" [Medium,Confirmed]
11:14 seb128 that's network-manager change is confusing a lot of users
11:14 seb128 that network-manager change
11:14 Mithrandir seb128: but it's confusing them rather than leaving them with broken networks.
11:14 sladen NM is teh rock.
11:14 pitti seb128: red herring, drescher has the new hwdb-client, so that's fixed as well
11:14 ogra sladen, the rock we all get buried under ?
11:15 seb128 Mithrandir: how hard would it be to display no icon when it's not used?
11:15 Mithrandir seb128: probably doable. I'll look into it.
11:15 seb128 would be great
11:15 ogra seb128, it already does that if i shut down my WLAN (but keep the static iface up)
11:15 ogra as soon as i load the bcm module again, its back
11:16 ogra so juat abusing that function could probably work
11:16 cjwatson Mithrandir: could the n-m bug be due to your most recent change?
11:17 Mithrandir cjwatson: yes, and while the confusion isn't intentional, it is behaving correctly.
11:17 Mithrandir so we need to address the confusion.
11:17 cjwatson Mithrandir: is there an explanation in the bug of the confusion?
11:17 seb128 cjwatson: network-manager does the right thing, it says there is no network configured though
11:18 seb128 which means is not managing any
11:18 Mithrandir what seb says.
11:18 seb128 but have an icon saying "no network configured" confuses users
11:18 cjwatson the bug above only has confused comments from users, AFAICS
11:18 cjwatson it needs an authoritative comment from somebody who understands what is going on
11:18 seb128 because there is a network configured and working
11:19 heno I tested it with my father. He was slightly confused, but shrugged when the network worked anyway
11:20 Mithrandir there seems to be a real bug related to the "disable networking" menu item, I need to investigate what's the problem there.
11:20 fabbione wouldn't be possible to change the string to match reality?
11:20 fabbione like: "hey dude.. i can't manage your setup. kthxbye"
11:20 Mithrandir fabbione: it would break translations, but otherwise that's a possibility.
11:20 fabbione we can always unbreak them with langpacks update
11:21 fabbione it's probably the less intrusive change we can do
11:21 heno yes, the string is more confusing than the icon
11:21 fabbione dunno how many langs that string is translated, but i am pretty sure we can fix the major ones within the team
11:22 seb128 fabbione: what's wrong with hiding the icon when there is nothing to manage?
11:22 seb128 that doesn't break any string
11:22 heno 'Please replug your network cable' would work, because that 'fixes' it for the session :)
11:22 mdz heno: how does that fix it?
11:22 pitti seb128: but you'd lose the menu entry for static configuration or enabling wlan, etc.
11:22 fabbione seb128: nothing.. you need to change code and possibly add other regressions
11:22 iwj BenC: Is bug 99648 the same as bug 53268 ?
11:22 cjwatson heno: sounds more specific than the problem in fact indicates?
11:22 mdz this isn't the time to think about redesigning the UI
11:22 seb128 pitti: can you enable wlan when there is not connexion managed?
ubotu Malone bug 99648 in linux-source-2.6.20 "MMC/SD memory card reader does not work" [High,Fix committed] https://launchpad.net/bugs/99648
ubotu Malone bug 53268 in hal "On Thinkpad X60s and Z60 SD card reader doesn't work" [Undecided,Confirmed] https://launchpad.net/bugs/53268
11:23 heno mdz: if you unplu and replug ethernet it behaves as expected until you reboot
11:23 pitti seb128: if you disabled it before, then you'll be in that situation
11:23 mdz it seems there is a genuine issue, but we ought not rush in a poorly thought out change
11:23 mdz heno: why? I thought in this situation, n-m was ignoring the interface
11:23 pitti disabling wlan temporarily should not cause the icon to go away
11:23 heno I think it ignores it bacause the network is brought up during boot, before n-m
11:24 heno if you 'let n-m find it' it's happy
11:25 mdz heno: but if it's already being managed by ifupdown, that would result in confusion (like two dhclients)
11:26 Mithrandir mdz: except the fact that network-manager kills it as part of its startup.
11:27 Mithrandir void nm_system_kill_all_dhcp_daemons (void)
11:27 Mithrandir { nm_spawn_process ("/usr/bin/killall -q dhclient");
11:27 Mithrandir }
11:27 mdz ...
11:27 heno Mithrandir: so when it does that it should also set a flag so it remembers to check for an active network later
11:27 mdz what a piece of work
11:27 fabbione Mithrandir: and it doesn't always work because sometime we spawn dhclient3
11:28 Keybuk Mithrandir: isn't it dhclient3?
11:28 ogra Mithrandir, could you call that nm_system_kill_all_dhcp_client ??
11:28 Mithrandir fabbione: we can kill those too.
11:28 Mithrandir ogra: no.
11:28 fabbione Mithrandir: we should
11:28 ogra Mithrandir, thats confusing :)
11:28 Mithrandir ogra: I'm not going to change the name of NM-internal interfaces.
11:28 fabbione ogra: no you can't. the nm backends export a standard API towards the rest
11:28 cjwatson Mithrandir: n-m ubuntu5 was a further fix for the problem with people with manually configured interfaces having trouble with n-m claiming they were offline, yes?
11:29 Treenaks what if n-m only manages one of several DHCP interfaces?
11:29 ogra Mithrandir, oh, i thought it was a new function ... ignore me :)=
11:29 Mithrandir cjwatson: it was a revert to the ubuntu3 behaviour, except we don't claim to be offline if there are disabled devices available.
11:30 cjwatson Mithrandir: remind me, was there a bug about that?
11:30 Mithrandir and then the applet is patched to check both the online/offline status and the number of devices managed. If the number of managed devices is zero, it goes into "we are offline" mode.
11:30 Mithrandir cjwatson: yes, it broke LTSP setups for instance.
11:30 Mithrandir let me find the bug #
11:31 cjwatson we should have a policy that late changes must have a bug# referenced in the changelog
11:31 ogra cjwatson, it hogged static interfaces that are configured in /e/n/i and teared them down because NM allows only one iface to be up
11:31 Keybuk Mithrandir: why do you kill dhclient, rather than invoke ifdown?
11:32 fabbione Keybuk: that's what upstream does in all backends IIRC
11:32 fabbione Keybuk: you don't want to call ifdown because the interface might be in use for nfs mounts
11:32 fabbione Keybuk: just take over the dhcp process
11:32 Keybuk redhat don't have something quite like our ifup though
11:32 Mithrandir Keybuk: I believe that code is either from upstream or from you, I haven't written it.
11:32 Keybuk since ifdown won't work for that interface after you kill its dhclient
11:32 Mithrandir cjwatson: 100021
11:33 cjwatson thanks
11:34 cjwatson BenC: are you still awake?
11:37 fabbione am i still connected?
11:37 seb128 fabbione: yes
11:37 fabbione thanks
11:37 mvo fabbione: no
11:41 Mithrandir I could make it use the connected icon and change the tooltip if there are manual devices, would that be good enough?
11:41 ogra sounds like
11:41 ogra as long as it doesnt *always* show the connected icon then at least :)
11:42 Mithrandir ogra: it'll always show it if you have manually configured devices.
11:42 heno Mithrandir: by manual, do you mean eth0 or static IP?
11:42 ogra thats fine
11:42 ogra heno, static ip and auto in /e/n/i
11:42 heno this problem occurs with dhcp too
11:43 heno which is the most common for basic users
11:43 Mithrandir heno: anything not managed by NM.
11:43 ogra you mean if /e/n/i is empty ?
11:43 heno ok, that will work then
11:43 ogra then it shouldnt happen ...
11:44 cjwatson so we have a possible proposal to skip release candidate and release Herd 6 instead
11:44 cjwatson at this point, if we have to delay release candidate, there'll be very little time to squeeze in validation
11:45 mdz we need to hear from Ben about these kernel bugs
11:45 mdz the changes which are proposed are too intrusive for 7 days before release
11:45 cjwatson so Herd 6 would be a much-less-validated set of images
11:45 mdz so if the problems are truly showstoppers, then we are looking at a delay of the release
11:45 fabbione what kind of delay we are talking about?
11:45 fabbione one week? two weeks?
11:45 mdz today was the first I had heard of these issues
11:45 mdz fabbione: unknown
11:46 cjwatson HPA has been on the table for some time and Ben has tested the patch on his array of non-HPA hardware
11:46 mdz fabbione: this change is proposing to switch to an entirely different IDE driver for a huge proportion of Intel based systesm
11:46 mdz (bug 96857)
11:46 cjwatson the PIIX thing I think we should drop
ubotu Malone bug 96857 in linux-source-2.6.20 "Feisty Beta: debian-installer does not mount cdrom" [High,Fix committed] https://launchpad.net/bugs/96857
11:47 mdz so we have:
11:48 mdz kernel issues: we will have a talk with Ben when he wakes up and review those
11:48 mdz network-manager: Keybuk is looking over the patches and doing some further analysis
11:48 mdz what else is on the table?
11:48 fabbione server installer issues: Keybuk is on it already. bug #105623
ubotu Malone bug 105623 in devmapper "udev rules missing from udeb" [High,Confirmed] https://launchpad.net/bugs/105623
11:49 mdz certification: we will await results, and have to decide on whether to delay if there are problems
11:49 mdz (fabio tracking)
11:49 mdz fabbione: this is a showstopper?
11:49 mvo I would like to add additional sanity checking into the release upgrader for python, but that should not be controversal I hope?
11:49 fabbione mdz: for RC no (would be enough a release note). For final yes
11:50 Mithrandir fabbione: fwiw, it just worked fine for me.
11:50 mdz fabbione: note that there is unlikely to be an RC unless we delay
11:50 fabbione Mithrandir: it works.. it only takes ages to work.. so if you don't look at it and go back after a while you don't even notice
11:50 Mithrandir fabbione: ok.
11:51 fabbione Mithrandir: it's about waiting 3 minutes for each devmapper device to be created
11:51 mdz mvo: for the CDs?
11:51 fabbione mdz: yeah of course..
11:51 cjwatson just adding those rules to the udeb stands a good chance of breaking the partitioner. I'd veto that.
11:51 cjwatson (because it would bring LVs/whatever up by default which the partitioner isn't expecting)
11:51 mvo mdz: having it on the CD is a bonus, but if its available online that should be fine
11:51 fabbione cjwatson: they are the exact same rules that lands in the installed system
11:52 cjwatson fabbione: the installed system is different in important ways
11:52 doko mdz: bug 105642, bug 105828
11:52 cjwatson release note and point release if we can get the change tested. It's annoying, but fixing it is very risky
ubotu Malone bug 105642 in linux-source-2.6.20 "kernel panic starting edubuntu dvd in live mode with VGA" [High,Unconfirmed] https://launchpad.net/bugs/105642
ubotu Malone bug 105828 in Ubuntu "wrong keyboard layout after LTSP chroot install" [Undecided,Unconfirmed] https://launchpad.net/bugs/105828
11:52 Mithrandir while it's annoying, I don't think it's a showstopper either.
11:53 fabbione cjwatson: i think Keybuk can tell you more about the status
11:53 mdz doko: you are saying that these are showstopper bugs?
11:53 cjwatson fabbione: I've already spoken with Scott about this.
11:53 fabbione cjwatson: ok
11:53 cjwatson keyboard layout> totally not release-critical. dpkg-reconfigure console-setup
11:53 ogra mdz, i'm not yet sure 105828 is a valid one, i havent seen it in any installs here
11:53 doko mdz: well, a kernel panic for the default install? IMO rc for the final; the edubuntu install not.
11:54 ogra mdz, doko and i are still investigating
11:54 mdz doko: I'm asking you; you just gave bug numbers without context
11:54 ogra doko, the panic doesnt happen witjh the CD here
11:54 mdz doko: a kernel panic under what circumstances?
11:54 ogra mdz, framebuffer
11:54 ogra it immediatley dies if it tries frambuffer stuff
11:55 ogra i.e. directly after gfxboot
11:55 doko mdz: 105642: insert the edubuntu amd64 dvd, let it boot alone.
11:55 ogra it either is specific to dokos HW or to the DVD
11:55 ogra i dont see it on two different machines with the CD
11:55 doko ogra: I can recheck with another DVD, but not other hardware
11:56 Mithrandir I can check it here once I finish downloading the edubuntu dvd
11:57 heno I have the Edubuntu DVD, I'll check too
11:57 mdz surely this is not specific to edubuntu
11:57 ogra doko, i find it weird that you have so many bugs i dont see anywhere else ...
11:57 mdz if it is, it's even less potentially serious
11:57 ogra especially the keyboard thing i see no technical way how that could happen
11:58 mdz it's easy enough to test, should not be hardware-specific
11:58 cjwatson I'm burning an Ubuntu amd64 DVD now
11:59 ogra mdz, still i'm missing a logical explanation how a setting in a chroot can influence a confiog file on the host system
11:59 cjwatson ogra: it's not release-critical. please worry about it later
11:59 cjwatson we do not have time to get into extensive research regarding non-release-critical problems
11:59 ogra cjwatson, well ... having all edubuntu installs with US keyboard is something i'd see as RC
11:59 cjwatson I disagree
12:00 cjwatson you can reconfigure it trivially, and furthermore you already said it's not happening to you
12:00 ogra 80% of your users are non english
12:00 cjwatson unless you feel that your testing is in some way inadequate?
12:00 ogra not really ... it worked the last releases :)
12:01 cjwatson I'm not dismissing doko's problem, but I do think that it is not something to delay release for
12:01 ogra if i start seeing it too i'd consider it for edubuntu
12:02 ogra but i dont (yet ?) so this discussion is pointless
12:03 mdz ok, so we have a current direction for each of the outstanding issues I mentioned
12:03 mdz last call for other issues we need to be aware of for release
12:04 pitti bug 104602?
12:04 mdz we are in a delicate state right now, and it's important to know how serious it is
12:04 ubotu Malone bug 104602 in sysvinit "root password visible at emergency console" [Undecided,In progress] https://launchpad.net/bugs/104602
12:04 pitti can be fixed in a security update as well, though
12:04 fabbione ah so i was not the only one
12:04 fabbione that's also with upstart
12:05 pitti it's an usplash bug really
12:05 Mithrandir mdz: were you able to reproduce the NM resume from hibernate problem?
12:05 pitti it's not a release blocker, but important IMHO
12:05 fabbione mdz: there is a bug in multipath-tools but i already uploaded the fixed version. it's not worth more discussion here.
12:06 Keybuk pitti: we don't set root passwords by default, and you have to be at the console anyway
12:06 kwwii not a show stopper, but I was hoping to finish a usplash/CD install pic for Xubuntu
12:06 Keybuk definitely not stopping the show ;)
12:06 mdz pitti: definitely a serious issue which should be fixed
12:06 mdz pitti: but I think it can be a security update
12:06 cjwatson CD install images can go in very late, usplash less so
12:06 kwwii is that still possible?
12:06 pitti I agree, but you wanted to be aware of important issues, so I thought I'd mention it
12:06 mdz pitti: thanks
12:07 kwwii cjwatson: in any case, it should be done by tomorrow at the latest - if it is too late, so be it
12:08 mdz if you don't have a release-critical task to work on, please join #ubuntu-bugs
12:08 mdz and help keep an eye out for unexpected issues as they arise
12:09 mdz if you discover something which looks serious, please talk to Colin, Scott or myself
12:09 mdz there will be further communication about the kernel, certification and other outstanding issues as it is available
12:09 pitti so, will we get new RC images?
12:09 fabbione or Herd-6?
12:09 fabbione what's the consensum now?
12:09 pitti well, it's just a name
12:09 mdz we will get new daily builds
12:09 mdz it is unknown whether we will delay the release or not
12:10 mdz and that will determine the plan for remaining milestones
12:10 cjwatson I do not feel we should have a release candidate at this point
12:10 fabbione pitti: well herd-6 will involve a new RC set
12:10 cjwatson unless the release is delayed
12:10 mdz there is no point doing a full validation test of an RC at this point when there are known blockers
12:10 cjwatson pitti: herd-6 => not a complete round of validation
12:11 mdz has everyone done at least one test of each of their assigned ISOs?
12:11 mdz please confirm
=== fabbione did
12:11 pitti confirm
12:11 mvo yes
12:11 Riddell yes
=== dholbach did too
12:11 pkl_ pkl: yes
12:11 asac confirm
12:11 cjwatson I will have done in about one minute when I reboot
12:11 mdz heno: are you able to tell if any tests were outstanding?
12:11 Mithrandir server, not the kubuntu amd64 dvd yet (since cdimage was excessively slow yesterday)
12:12 fabbione mdz: https://www.stgraber.org/ubuntu/isotesting has a pretty good overview
12:12 heno mdz: https://www.stgraber.org/ubuntu/isotesting
12:12 pitti https://www.stgraber.org/ubuntu/isotesting shows quite a number of (0)s
12:12 asac oh wait ... not yet finished ... will finish later today
12:12 heno netboot is undertested and DVDs
12:12 iwj doko: Thanks for doing those dvd tests btw.
12:13 heno and upgrades, I'm doing some of those now
12:13 iwj Err, dholbach I mean.
12:13 iwj Duh.
12:13 mdz heno: please review with cjwatson and determine whether we need to push some specific cases
12:13 doko iwj: they were assigned to me :)
12:13 dholbach iwj: np
12:13 heno right
12:13 mdz ok, we've taken long enough with this meeting, let's reconvene later once we've filled in some of the unknowns
12:13 mdz adjourned, thanks allMeetingLogs/UbuntuDev/20070412 (last edited 2008-08-06 16:19:06 by localhost)