15:00:17 <jlaska> #startmeeting F-12Beta Blocker Bug Review meeting
15:00:17 <zodbot_> Meeting started Fri Oct  9 15:00:17 2009 UTC.  The chair is jlaska. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:17 <zodbot_> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:00:53 <jlaska> #topic gathering minds ...
15:01:20 <wwoods> oh no! my mind!
15:01:32 <jlaska> that's right ... we've got it now
15:01:33 <adamw> duuuhhhwwee?
15:01:37 * jlaska tightens lid on jar
15:01:58 <jlaska> lemme see if we can stir up a notting
15:02:31 <jlaska> not on irc ... thought I heard him earleir
15:02:33 <jlaska> earlier
15:03:00 <jlaska> might be too early for Oxf13 and poelcat?
15:03:08 <jlaska> I believe denise is present
15:03:19 <denise> yes, lurking ..
15:03:43 <jlaska> denise: howdy
15:03:50 <denise> hiya!
15:04:19 <jlaska> anyone else want to loan their mind or say hello to zodbot?
15:05:40 <jlaska> well, let's get started and we can pull in folks as needed
15:05:56 <jlaska> The usual procedure here ... I'm going to walk through https://bugzilla.redhat.com/showdependencytree.cgi?id=F12Beta&hide_resolved=1
15:06:06 <jlaska> the list has 5 bugs on it presently
15:06:16 <jlaska> starting with ...
15:06:19 <jlaska> #topic https://bugzilla.redhat.com/show_bug.cgi?id=517260
15:06:21 <buggbot> Bug 517260: medium, low, ---, dlehman, ASSIGNED, liveinst fails at partitioning screen
15:06:48 <jlaska> dlehman posted a patch to anaconda-devel-list to address this issue https://www.redhat.com/archives/anaconda-devel-list/2009-October/msg00163.html
15:07:08 <jlaska> I was trying to test it for him, but have unfortunately not been able to reproduce the underlying issue (since the storage test day last week)
15:07:23 <adamw> he's just posted a new comment explaining how to reproduce
15:07:33 <adamw> #23
15:08:00 <jlaska> yeah, we spoke about it before the meeting ...
15:08:09 <jlaska> that proceedure is not reproducing the issue
15:08:24 <adamw> ah :/
15:08:25 <jlaska> dlehman noted that some other changes in the UI might have worked around the underlying problem
15:08:39 <jlaska> so .. I'm inclined to push this off until post-beta
15:08:39 <adamw> are there any other known sufferers at this point?
15:08:50 <jlaska> I spoke to harald who also hit this issue ... and we tried his reproducer
15:08:53 <jlaska> no luck
15:09:01 <jlaska> but we were running into another bug that I added for review (down the list)
15:09:56 <jlaska> denise, dunno if you have any thoughts here ... the choice would be to take that patch post-beta or now
15:10:03 <jlaska> (or not at all I guess
15:10:05 <jlaska> )
15:10:17 <jlaska> but kind of disconcerting that we weren't able to reproduce the bug anymore
15:10:27 <jlaska> good from a user standpoint though :)
15:10:43 <denise> jlaska, as long as you take it!
15:11:05 <jlaska> #info jlaska unable to reproduce 517260 using the procedure posted to the bugzilla
15:11:16 <adamw> brb, call of nature
15:11:35 <jlaska> then I'd vote for post-beta (under the less change banner)
15:11:39 <jlaska> any objections?
15:12:07 <denise> +1
15:12:24 <jlaska> alrighty ... I'll go ahead and change the bug to block F12Blocker
15:13:27 <jlaska> #agreed 517260 will be moved to F12Blocker and accepted post-beta
15:13:36 <jlaska> #topic https://bugzilla.redhat.com/show_bug.cgi?id=526208
15:13:38 <buggbot> Bug 526208: low, low, ---, mclasen, ASSIGNED, preupgrade failed from old release(f10, f11)
15:14:11 <adamw> erf
15:14:22 <adamw> i'm not a huge fan of post-beta anaconda patches
15:14:22 <jlaska> who has the latest on this one?
15:14:22 <adamw> especially for things we are not sure are broken
15:14:29 <wwoods> it's not an anaconda bug.
15:14:38 <adamw> wwoods: i'm talking about 517260 sorry
15:14:40 <wwoods> it's a bug in gtk2/preupgrade in F11.
15:14:43 <adamw> since i was away when the vote happened :)
15:14:53 <wwoods> gotcha, sorry!
15:15:06 <adamw> beta is going to be the last point at which we get really wide-based installer testing
15:15:16 <adamw> any patch we throw into the installer after that is something of a wildcard
15:15:24 <wwoods> anyway there are gross workarounds (preupgrade-cli, downgrade gtk2)
15:15:52 <jlaska> adamw: I'll just back to that bug in a sec then ...
15:15:55 * poelcat lurking, but at another meeting
15:15:58 <jlaska> let's just finish out the preupgrade bz
15:16:08 <jlaska> s/just/jump/
15:16:22 <adamw> k, sorry
15:16:27 <jlaska> np np
15:16:40 <jlaska> wwoods: so we have some ugly workarounds?
15:16:57 <wwoods> and as discussed previously, preupgrade can be fixed in an F11 updates push after Beta
15:17:08 <wwoods> so it's probably not worth delaying the release over
15:17:11 <wwoods> err, delaying Beta
15:17:27 <jlaska> how close does that one feel to having a real fix
15:17:35 <jlaska> is that days or weeks away
15:18:03 <wwoods> yeah, the workarounds are not good - downgrading gtk2 is nasty, preupgrade-cli is a totally different user experience.. or you just just wait patently and only watch the progrss via the status icon
15:18:09 <wwoods> I have no idea
15:18:16 <jlaska> #info "as discussed previously, preupgrade can be fixed in an F11 updates push after Beta" -wwoods
15:18:30 <wwoods> the only comment I've got is "rewrite to use a mainloop" which is a fairly major refactoring effort
15:18:45 <wwoods> and no actual assertion that "this will definitely fix the problem because ..."
15:19:15 <wwoods> I'm gonna take some time today to actually try to trace the problem
15:19:45 <jlaska> wwoods: okay
15:19:48 <wwoods> I feel like it's probably a pretty simple fix - just need to find someone who knows what the fix is
15:19:51 <adamw> wwoods: i think that looked like a red herring from the bug discussion
15:20:06 <adamw> mclasen's last comment suggests he didn't think your code had a mainloop at all
15:20:44 <jlaska> would it help to have mclasen here?
15:21:50 <adamw> i don't think so - like i said he kinda took himself out of being a useful analyst of the problem :) it looks to me like we need a different expert, heh
15:22:08 <jlaska> is skvidal plugged in on this one still
15:22:22 <wwoods> yeah, but neither of us are pygtk experts
15:22:33 <rjune_wrk> ?
15:22:41 <rjune_wrk> what's up?
15:22:42 <adamw> it's kind of taken a turn since it was discovered to be a gtk/pygtk bug - it looks like it needs the input of someone who's knowledgeable in that area
15:22:46 <adamw> rjune: ?
15:22:48 <rjune_wrk> I'm no expert, but I'm familiar with pygtk
15:23:03 <adamw> rjune: https://bugzilla.redhat.com/show_bug.cgi?id=526208
15:23:04 <buggbot> Bug 526208: low, low, ---, mclasen, ASSIGNED, preupgrade failed from old release(f10, f11)
15:24:05 <rjune_wrk> this is going from f10/f11 to f12?
15:24:10 <adamw> yes.
15:24:37 <notting> is it a bug in the *old* preupgrade?
15:24:53 <rjune_wrk> that would be the next question. is it fixed already?
15:25:06 <adamw> no. it's a bug in gtk or pygtk.
15:25:14 <notting> adamw: but the old version
15:25:19 <mclasen> that is in no way clear to me...
15:25:26 <wwoods> old version of *what*?
15:25:35 <notting> it's an error running preupgrade on f10/f11, right?
15:25:38 <wwoods> preupgrade runs on the (unupgraded) host system
15:25:41 <adamw> notting: the bug is in the f10/f11 package base, yes, as far as we know.
15:25:47 <wwoods> it doesn't run on the target at all
15:25:50 <notting> so, there is nothing we can possibly change in the beta to fix it. ergo, it's not a beta blocker.
15:25:56 <rjune_wrk> mclasen: if the bug is in the version of gtk / pygtk installed in f10 and f11, *BUT* there is a newer version in f12, the bug may be fixed and it's a simple backport
15:26:07 <adamw> notting: we went through that at the last meeting.
15:26:33 <adamw> notting: it is possible for us to refuse to release the beta unless a bug which is not technically in the beta but relates to it is fixed.
15:26:48 <mclasen> rjune_wrk: there is no simple backporting between gtk 2.18 and 2.16
15:27:09 <adamw> notting: we can do that either by setting it as a beta blocker and remember it doesn't block us building the beta, just releasing it; or not set it as a blocker and remember that we shouldn't release until it's fixed. either depends on us being people with brains. it didn't seem that important which we picked.
15:27:37 <adamw> rjune: we can't tell if the problem happens when preupgrading from f12 as there's nothing to preupgrade *to*, yet.
15:27:39 <rjune_wrk> mclasen: ok so it's not simple. it's still a backport and the bug is likely documented
15:27:53 <rjune_wrk> adamw: so this is from f10 to f11 ?
15:27:56 <mclasen> rjune_wrk: I don't know where you take that knowledge from
15:27:56 <adamw> no. from f11 to f12.
15:28:00 <rjune_wrk> not from f10 or f11 to f12
15:28:14 <adamw> or possibly f10 to f12, i think, but we're focussing on the f11 case afaik.
15:28:32 <rjune_wrk> ok.
15:28:34 <adamw> the point is we don't know if the code in f12 also has this problem because you can't do a preupgrade from f12 to anything else, yet. there's nowhere to go.
15:28:39 <mclasen> I think the first step toward identifying the problem would be to narrow 'between 2.16.0 and 2.16.6 to a single version step
15:28:44 <wwoods> we've got a reliable reproducer with F11; haven't tried to confirm it with F10 yet
15:28:50 <rjune_wrk> ok.
15:29:06 <wwoods> if there's any known changes in behavior between those gtk versions I'd really like to know about it
15:29:07 <jlaska> #info "I think the first step toward identifying the problem would be to narrow 'between 2.16.0 and 2.16.6 to a single version step" -mclasen
15:29:10 <wwoods> changelog the best place to look for that?
15:29:33 <adamw> i think the idea would be just to test with each intermediate step and see which breaks
15:29:44 <adamw> assuming we can pull packages from koji, or just rebuild them
15:29:46 <wwoods> I don't think we packaged every intermediate step
15:29:48 <mclasen> wwoods: not sure what you are after; of course there are changes, otherwise we wouldn't release a new version...
15:30:19 <wwoods> mclasen: right, but specifically I'm looking for changes that involve the way gtk.events_pending() and gtk.main_iteration() would work
15:30:23 <wwoods> since AFAICT that's what's failing
15:31:03 <mclasen> wwoods: there's no changes in that area
15:31:04 <wwoods> it has a Very Long Running Callback where we just manually run the event queue rather than returning to the mainloop
15:31:04 <mclasen> http://git.gnome.org/cgit/gtk+/log/?h=gtk-2-16
15:31:23 <jlaska> is there a smaller simpler test script as a reproducer?
15:31:26 <jlaska> to avoid reinstalls etc...
15:31:26 <wwoods> not yet
15:31:42 <wwoods> but I've got a couple F11 VMs that I've been cloning and messing with
15:31:49 <jlaska> ah good call
15:32:29 <jlaska> okay, to summarize ...
15:32:55 <jlaska> we still need to isolate root cause, to do that mclasen recommended testing against certain versions
15:33:15 <jlaska> and I'm hearing from adamw, that this should stay on the list for now
15:33:26 <wwoods> disagree
15:33:28 <jlaska> objections/misreads?
15:33:34 <adamw> jlaska: no, that's not my position
15:33:52 <adamw> jlaska: was just explaining to notting that _in theory_ blocker bugs don't have to be in the f12 package base
15:33:55 <wwoods> I agree with the point that technically something can be a release blocker despite it not being part of the release
15:34:00 <adamw> whether this is one of those cases isn't the same point
15:34:05 <jlaska> adamw: sure
15:34:17 <wwoods> And we also made the point that this bug is a *release* blocker which is, technically, distinct from a *compose* blocker
15:34:29 <wwoods> but that's kind of just academic
15:34:37 <adamw> it feels kinda on the bubble to me as to whether we consider it a beta release blocker or not
15:34:43 <wwoods> I vote we move the bug to F12Blocker
15:34:56 <adamw> i'm abstaining :)
15:34:58 <wwoods> document the workarounds and push a fix ASAP
15:35:07 <wwoods> the fix can be pushed post-Beta
15:35:20 <mclasen> depends on if you want to test upgrades with the beta
15:35:29 <mclasen> and if you can live with a documented workaround for now
15:35:48 <adamw> mclasen: as noted, since the fix would go into f11 not f12, once it's fixed people would still be able to test with the beta
15:36:03 <mclasen> sure
15:36:06 <wwoods> we definitely want to test upgrades with the beta, but I don't think we need that *to the exclusion of everything else*
15:36:06 <adamw> but of course mostly people are likely to test around release time...
15:36:09 <wwoods> you can still upgrade with media / boot.iso
15:36:42 <wwoods> plus - it still *WORKS* is the thing
15:36:49 <jlaska> let's push out a fix that pops up a window ... "this will take a while and won't be responsive" :)
15:36:53 <wwoods> you can just minimize the window and walk away and it'll work exactly as expected
15:36:59 <wwoods> and the status icon even provides correct status
15:37:18 <wwoods> if we knew how to update the UI to pop up a window, we wouldn't have this problem
15:37:24 <adamw> =)
15:37:30 <jlaska> wwoods: heh
15:38:14 <jlaska> alrighty ... so I agree, this shouldn't block compose, but I think we've got ~3-4 work days where we can still chase this issue down and have a fix in time for beta
15:38:31 <wwoods> I'll put other things aside for this - at least for a day or two
15:38:39 <jlaska> would more people help solve this ... or wwoods are you comfortable going to town on those package versions?
15:39:09 <jlaska> adamw: common_bugs still alive and well?
15:39:19 <mclasen> one thing that _might_ be related is e.g. this: http://git.gnome.org/cgit/gtk+/commit/?h=gtk-2-16&id=52325ccd669ebd24de873bb0949d68d6dad17fb0
15:39:41 <mclasen> if preupgrade previously relied on getting random events to pump the event loop...
15:40:00 <adamw> jlaska: i'll be updating it soon, want to get an accurate snapshot of what will/won't be broken in the beta :)
15:40:08 <jlaska> adamw: agreed :)
15:40:12 <adamw> this'll definitely go in there if it's still broken at release time
15:41:03 <jlaska> I think I'd prefer to keep this on the list until beta goes public
15:41:22 <jlaska> and see about trying to line up help to isolate the problem
15:42:00 <jlaska> which bug this blocks seems like semantics at this point ... we are trying to work the issue for beta
15:42:08 <jlaska> but it doesn't block beta compose
15:42:27 <jlaska> while we are at least working it for beta, let's just keep it on the list?
15:42:40 <adamw> sure, as we said last week, just relies on us having brains to remember the situation :)
15:42:45 <jlaska> right on
15:43:24 <wwoods> mclasen: hrm, maybe it was! so maybe events_pending() is always returning false/[]
15:43:33 <jlaska> okay ... folks ready to move on (or back)?
15:43:38 <adamw> sure
15:43:41 <wwoods> we aren't actually firing off any events..
15:43:46 <wwoods> yeah please do, sorry
15:44:17 <jlaska> #agreed wwoods and mclasen will continue to identify root cause on 526208 with the goal of a new F-11 update in time for F-12-Beta
15:44:22 <mclasen> wwoods: normally, you would expect some queue_draw calls, or such, or just setting progress values will cause the progressbar to do that
15:44:26 <jlaska> adamw: ready to jump back to first bz?
15:44:34 <jlaska> #topic https://bugzilla.redhat.com/show_bug.cgi?id=517260
15:44:35 <buggbot> Bug 517260: medium, low, ---, dlehman, ASSIGNED, liveinst fails at partitioning screen
15:44:51 <adamw> yeah...i just wanted to note that i'm leery about adding patches post-beta to 'fix' things that we can't reproduce as being broken any more
15:45:03 <jlaska> Spoke to dlehman ... who suggested that if we take this fix for Beta ... we have to get a specific set of test results to confirm this isn't introducing regressions
15:45:15 <jlaska> adamw: you're point is valid
15:45:18 <jlaska> your point
15:45:45 <adamw> regression testing is good but by the nature of the beast we can't test in as wide a range of scenarios as the beta will be tested
15:46:08 <adamw> if the anaconda guys really think it's ok to put this in then i guess i'll live with it. really i just want my leeriness noted so i can say 'i told you so' if it all goes south =)
15:46:42 <jlaska> hey, we're all in this together :)
15:47:07 <jlaska> so ... any changes to the plan here? ... we can take it for beta ... but will need to scramble on validating the patch
15:47:07 * mclasen points out that it is somewhat disingenuous to be very adamant about not taking patches between beta and final when we are just opening the floodgates again after final
15:47:37 <adamw> mclasen: eh? why?
15:47:58 <adamw> you can break stuff in rawhide after we release f12. that's fine. breaking stuff is what rawhide's there for...
15:48:36 <jlaska> adamw: wanna help test this patch with me?
15:48:44 <mclasen> adamw: I'm talking about opening the f12 updates floodgates...
15:48:54 <jlaska> or if we can get help from others ... we can get confidence in the fix and take it for beta
15:49:04 <adamw> mclasen: oh. this is an anaconda issue. anaconda doesn't get updated after release.
15:49:11 <mclasen> right
15:49:12 <adamw> jlaska: i can do what i can do, but all i can do at present is virtual installs
15:49:17 <mclasen> anaconda is different, ignore me
15:49:21 <adamw> jlaska: i have no spare hardware for to-the-metal installs :/
15:49:29 <jlaska> adamw: hey, I offered :)
15:49:39 <jlaska> alright ... so I'll take an action to work this w/ dlehman
15:49:41 <adamw> yeah, maybe i should expense a couple of spare hard disks
15:50:07 <jlaska> if we can get confidence that this won't invalidate all of the beta testing so far ... we'll take it
15:50:15 <adamw> jlaska: anyway, my personal preference would be to have it in the beta so it gets wider testing, but do override me if you think i'm wrong. =)
15:50:27 <jlaska> nope, I agree with that
15:50:35 <jlaska> but want to test it *before* it lands in the beta
15:51:00 <jlaska> #action jlaska will work w/ dlehman to verify the bug#517260 patch before it hits next anaconda
15:51:00 <adamw> sure. i can definitely do a couple of virt installs with it to help test. i'll cc myself on the bug to stay in the loop.
15:51:01 <buggbot> Bug https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=517260 medium, low, ---, dlehman, ASSIGNED, liveinst fails at partitioning screen
15:51:15 <jlaska> okay moving on ...
15:51:17 <jlaska> #topic https://bugzilla.redhat.com/show_bug.cgi?id=526320
15:51:18 <buggbot> Bug 526320: medium, low, ---, jkeating, MODIFIED, ppc64.img and ppc32.img missing from tree
15:51:32 <jlaska> I moved this from CLOSED -> MODIFIED today since the issue doesn't seem fixed from what I can tell
15:51:36 <jlaska> is Oxf13 around?
15:52:09 <jlaska> sure enough, it might have just needed the newer anaconda build in the compose
15:52:20 <jlaska> 'cause it seems to be here for todays recently finished compose
15:52:23 <jlaska> http://kojipkgs.fedoraproject.org/mash/rawhide-20091009/development/ppc/os/images/netboot/
15:52:26 <jlaska> nice, that was easy
15:53:00 <jlaska> I just closed that bug so it's now off the list
15:53:06 <jlaska> next up ...
15:53:15 <jlaska> #topic https://bugzilla.redhat.com/show_bug.cgi?id=526899
15:53:16 <buggbot> Bug 526899: medium, low, ---, anaconda-maint-list, NEW, Encrypted partitions has "Unknown" type in table and ext4 in editor
15:53:31 <jlaska> this out of greenlion1's testing from the anaconda storage test day
15:53:40 <jlaska> I added this for review during this meeting
15:53:47 <greenlion1> yep
15:53:50 <greenlion1> it's on livecd
15:54:13 <jlaska> the use case seems to be that using the live image, encrypted partitions don't get a mount point anymore
15:54:24 <jlaska> so the only encrypted isntalls you can do are autopart+encrypted ... it seems no manual encrypted partitioning
15:54:51 <jlaska> see screenshot posted by greenlion1 - https://bugzilla.redhat.com/attachment.cgi?id=363477
15:55:20 <jlaska> this is a regression from the Beta test compose ... which is why I bumped this into the meeting requesting blocker
15:55:40 <jlaska> denise: what's your take?
15:56:11 <denise> dlehman is on it, but not a lot to report
15:56:39 <Oxf13> jlaska: I'm somewhat around now
15:56:43 <Oxf13> jlaska: still trying to do the breakfast thing
15:56:51 <jlaska> Oxf13: mornin' :)
15:56:52 <denise> could document around it for beta if we had to i guess, but this feels like one we should fix to get the testing
15:57:37 <Oxf13> what's the bug here?
15:58:08 <adamw> Oxf13: we already got past the one jlaska pinged you for :)
15:58:08 <jlaska> Oxf13: seems like we can no longer create encrypted partitions in the manual disk partition screen
15:58:08 <adamw> Oxf13: now we're on https://bugzilla.redhat.com/show_bug.cgi?id=526899
15:58:08 <buggbot> Bug 526899: medium, low, ---, anaconda-maint-list, NEW, Encrypted partitions has "Unknown" type in table and ext4 in editor
15:59:19 <jlaska> any other opinions/concerns on this issue?
16:00:02 <Oxf13> hrm.
16:00:07 <Oxf13> let me preface something here.
16:00:13 <Oxf13> We have a high chance of getting into RC phase today
16:00:13 <jlaska> uh oh :)
16:00:26 <Oxf13> I think the last of the blockers were "fixed" by the anaconda build I tagged last night
16:00:37 <Oxf13> so lets be really sure something is a blocker before we cause a slip
16:00:48 <jlaska> right on, that's why we're here
16:02:03 <Oxf13> but I don't think today's rawhide has landed yet, so there may be surprises :/
16:02:14 <jlaska> Oxf13: I built a live image this morning from rawhide-static
16:02:21 <Oxf13> k
16:02:23 <jlaska> that's where I'm seeing this issue
16:02:27 <Oxf13> k
16:02:43 <Oxf13> I can deal with a new anaconda build on the side to enter RC phase
16:02:50 <Oxf13> that's easy to manage.
16:02:54 <Oxf13> anything more gets less so
16:04:03 <jlaska> are you in agreement on this bug staying on the list and dlehman continues to investigate?
16:04:47 * jlaska reconfirms ... there is something unpleasant going on with the partition detail screen
16:05:16 <Oxf13> yeah
16:05:25 <Oxf13> I think it would be crappy to go out like this
16:05:31 <jlaska> unfortunate I know ... but I ... yeah exactly
16:06:05 <jlaska> #agreed 526899 will stay on beta blocker awaiting review from dlehman
16:06:10 <jlaska> okay ... next up ...
16:06:15 <jlaska> #topic https://bugzilla.redhat.com/show_bug.cgi?id=528077
16:06:16 <buggbot> Bug 528077: high, low, ---, rstrode, NEW, No graphical boot (verbose text boot only) on ix86 live images 20091007+
16:06:27 <jlaska> adamw: you want to take this?
16:07:00 <adamw> well, it's pretty straightforward
16:07:09 <adamw> you think it may be fixed with today's builds so i'll check it out
16:07:25 <jlaska> confirming on _real_ hardware now
16:08:15 <adamw> the bug is just as the subject says - 20091007 and 20091008 i686 live images gave verbose text boot, on all systems tested (by me and jlaska)
16:08:21 <jlaska> having a non-functioning plymouth seems like a no brainer
16:08:25 <adamw> but the matching x86-64 images gave nice working graphical boot
16:08:42 * mclasen welcomes halfline
16:08:56 <adamw> i'll test with 20091009 images when there are some
16:09:03 <adamw> jlaska's using a homebrew image
16:09:22 <jlaska> adamw: using the test day instructions (https://fedoraproject.org/wiki/QA/Test_Days/Live_Image)
16:10:03 <jlaska> any objections to not keeping this on the list?
16:10:17 <jlaska> rather ... anyone suggesting we remove this from the F12Beta list
16:10:54 <Oxf13> ugh...
16:10:57 <halfline> i chimed on the bug report
16:11:03 <Oxf13> that would be tougher to slip in if it wasn't already fixed
16:11:33 <halfline> it's pretty hard for me to imagine a plausible explanation for this bug
16:11:47 <adamw> halfline: thanks, will update you later
16:11:54 <Oxf13> I seem to remember getting text plugin booting live image I made yesterday in kvm, i686
16:12:17 <adamw> note you don't just get the text progress bar - you actually get a full verbose text boot
16:12:27 <notting> adamw: w/modeset, though, right?
16:12:27 <adamw> as if you removed 'rhgb quiet' from the parameters
16:12:29 <adamw> yes
16:12:36 <adamw> modeset kicks in
16:13:00 * adamw didn't actually check what the default kernel parameters were...wonder if it was as simple as they got screwed somehow...
16:13:12 <halfline> that could be
16:13:22 <adamw> anyway, will provide updated results from 20091009 later today
16:13:22 <Oxf13> hrm.
16:13:25 <halfline> the thing is...the fallback splash by default is "text" not "details"
16:13:33 <Oxf13> actually I think I see this on yesterday's image I built in kvm
16:13:36 <halfline> so if there was some failure, i'd expect to see animated bars
16:13:42 <adamw> right, that's what made it seem odd to me.
16:13:50 <Oxf13> but if jlaska doesn't see it today, maybe it fixed itself.
16:14:32 <jlaska> so, let's leave this on the list and follow-up with more testing this afternoon?
16:14:38 <Oxf13> k
16:14:59 <adamw> yeah
16:15:10 <halfline> i have a gdm fix i'd like to get in for the beta if possible
16:15:12 <Oxf13> rawhide is still doing the rsync, should be done soon
16:15:17 <halfline> fixes the "other" item
16:15:18 <Oxf13> halfline: erm...
16:15:19 <jlaska> #agreed retest 528077 on live images built today and provide feedback in bz
16:15:47 <jlaska> okay, lemme reload the F12Beta page ...
16:16:16 * greenlion1 has added one bug there at start of meeting...
16:16:17 <jlaska> we've got a new bug added to the list
16:16:43 <jlaska> halfline: we'll talk to greenlion1's bug, then come to yours
16:16:48 <jlaska> #topic https://bugzilla.redhat.com/show_bug.cgi?id=498026
16:16:50 <buggbot> Bug 498026: medium, low, ---, anaconda-maint-list, NEW, RuntimeError: Returning partitions to state prior to edit failed
16:17:19 <greenlion> I've hit in this many times during latest test day
16:17:44 <jlaska> looks like so is IBM
16:17:50 <Oxf13> while this is bad, I don't necessarily think it's a beta blocker.
16:17:51 * jlaska reading https://bugzilla.redhat.com/show_bug.cgi?id=498026#c17
16:17:52 <buggbot> Bug 498026: medium, low, ---, anaconda-maint-list, NEW, RuntimeError: Returning partitions to state prior to edit failed
16:18:01 <greenlion> basically, if all disk is filled with partitions, most operations will result in crash
16:18:14 <jlaska> greenlion: any work arounds?
16:18:22 <greenlion> it is hard to create any complex partition layout
16:18:34 <Oxf13> greenlion: so you're getting it without trying to revert any previous edits?
16:18:52 <greenlion> jlaska, only if you don't fill up fully disk
16:19:14 <jlaska> greenlion: comment#17 has a reproducer that seems to be intentionally validating an error case
16:19:32 <greenlion> Oxf13, not only revert, but when trying to edit partitions too
16:19:42 <jlaska> 4. Increase the size of the partition and press OK.
16:19:42 <jlaska> * This should give an error of saying " not enough free space on disk". Which
16:19:43 <jlaska> 5. Now click on 'OK' for the error. This will crash the anaconda.
16:19:45 <jlaska> is expected.
16:19:48 <jlaska> 
16:20:00 <greenlion> crash is expected?
16:20:14 <jlaska> no, the dialog was expected
16:20:16 <jlaska> afterwards, it crashed
16:20:20 <Oxf13> the is expected line got out of sync
16:20:29 <Oxf13> so basically there is a crash in the 'out of space' handler routine
16:20:34 <Oxf13> that might be easier to find/fix
16:20:47 <greenlion> it's even worse than bug with encryption on liveinst - there it doesn't crash
16:21:30 <jlaska> just reproduced using that procedure
16:21:53 <jlaska> but that procedure doesn't seem like a common use case  ... is this encountered some other way
16:22:16 <Oxf13> it's not common in that normally you'd reduce the size of something before increasing the size of something else
16:22:20 * greenlion will try to write more common use case
16:22:22 <Oxf13> but I can see this being hit
16:22:42 <greenlion> there is no need to resize partitions to crash it in this way
16:22:52 <jlaska> something disruptive was introduced into anaconda partitioning recently
16:24:14 <jlaska> I'd agree with keeping this if the failure case didn't involve negative testing -- a failure while testing whether anaconda let's you create a partition that's larger than your free space
16:24:22 <jlaska> Oxf13: your take?
16:24:24 <denise> dlehman says that he is pulling together an F12 tree to test a fix for this one
16:24:50 <jlaska> denise: geez, that was fast
16:24:55 <denise> he's good ;-)
16:25:02 <Oxf13> jlaska: from looking at it, I think the error can be isolated pretty easily, since it appears to be in the routine for when you have not enough free space
16:25:08 <jlaska> denise: sounds like we have a few folks reproducing this as well if he needs help
16:25:17 <Oxf13> if it crashed from finding out you had not enough, that might be worse, but this is a crash in handling the fact
16:25:18 <denise> ok will let him know
16:25:52 <jlaska> okay, I'm hearing this stays on the list from Oxf13 and greenlion
16:25:59 <jlaska> is that right?
16:26:13 <adamw> i haven't experienced this bug
16:26:19 <denise> jlaska, dlehman will be in touch
16:26:24 <jlaska> denise: thx
16:26:35 <adamw> but my only question would be, are we sure the bug greenlion is talking about, and the crash with the specific procedure outlined above, are both the _same_ bug?
16:26:39 <jlaska> adamw: I haven't either ... until I just tested the steps in comment#17
16:26:49 <adamw> jlaska: that was just a preface to my next remark =)
16:27:04 <jlaska> ;)
16:27:15 <greenlion> adamw, good question. I guess I need to reproduce and compare traceback?
16:27:23 <adamw> i think so, yeah
16:27:31 <jlaska> greenlion: did anaconda automatically cc you to that?
16:27:58 <greenlion> jlaska, I think, I've found it in bugzilla and cc'ed to in myself
16:28:37 <Oxf13> adamw: given that every reproducer seems to involve hitting the "you don't have enough space for this action" routine, I'd imagine they're the same bug
16:29:33 <adamw> Oxf13: ok
16:29:43 <jlaska> greenlion: if you can confirm, that would help too
16:30:26 <greenlion> ok
16:30:42 <jlaska> alright, so this is staying on the list then
16:31:08 <jlaska> any objections?
16:32:08 <jlaska> alrighty ... it stays
16:32:16 <jlaska> #agreed 498026 will remain on list - dlehman will have a patch available soon for verification
16:32:26 <jlaska> that's it for my list
16:32:33 <jlaska> halfline: you had a bug to review?
16:32:46 <halfline> jlaska: yea bug 527920
16:32:47 <buggbot> Bug https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=527920 high, low, ---, jmccann, MODIFIED, gdm doesn't show user list and crashes - unable to login
16:32:52 <jlaska> #topic https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=527920
16:32:55 <buggbot> Bug 527920: high, low, ---, jmccann, MODIFIED, gdm doesn't show user list and crashes - unable to login
16:32:58 <Oxf13> that's for folks with LDAP right?
16:33:06 <halfline> well it's for everyone
16:33:09 <halfline> it just affects them worse
16:33:17 <halfline> the "Other" item is absent right now
16:33:25 <halfline> which means you can only log into users that show up in the list
16:33:27 <Oxf13> I see.
16:33:46 <halfline> fix is already built
16:33:48 <Oxf13> lots of gdm packages :/
16:33:59 <adamw> what's the use case for logging in as a user that isn't in the list?
16:34:06 <adamw> (aside from LDAP)
16:34:14 <jlaska> krb/nis ?
16:34:21 <halfline> yea
16:34:54 * jlaska curses the live image terminal bug
16:34:54 <halfline> or for people who toggle root login back on
16:34:55 <Oxf13> ok, if we take this, I think we can do it on the side as well, and not have to wait another day
16:35:02 <warren> jlaska: you're still hitting that?
16:35:08 <warren> jlaska: what version of dracut?
16:35:33 <warren> jlaska: it is very maddening to debug because it rarely happens for me, I don't know if if what I tweaked fixed it or not
16:37:31 <jlaska> so it's seems possible to take this for Beta ... but does it hold it's own?
16:37:37 <jlaska> votes?
16:37:48 <Oxf13> jlaska: gdm ?
16:37:51 <jlaska> yeah
16:37:55 <Oxf13> Take it
16:38:51 <adamw> i guess take it
16:39:19 <jlaska> ldap and krb use seems like it would be low for a Fedora Beta ... but I'm not up on the usage data there
16:40:13 <mclasen> warren: I just hit it too
16:40:17 <mclasen> on a livecd built last night
16:40:30 <warren> mclasen: I want to know if dracut -13 in today's rawhide has the same issue
16:40:39 <warren> mclasen: we haven't narrowed it down to dracut or initscripts yet
16:40:57 <jlaska> #agreed 527920 affects all ldap and krb users during gdm login ... accepted as a beta blocker
16:41:25 <mclasen> warren: I can spin another live cd later
16:41:34 <jlaska> alrighty folks ... that's all I have on the list
16:41:42 <jlaska> #topic Open discussion
16:41:47 <greenlion> jlaska, about that anaconda bug: it attaches traceback to bug# 519212 which is closed as duplicate of 498026
16:41:48 <jlaska> anything else we didn't touch on yet?
16:41:49 <buggbot> Bug https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=519212 medium, medium, ---, anaconda-maint-list, CLOSED DUPLICATE, RuntimeError: Returning partitions to state prior to edit failed
16:42:41 <jlaska> greenlion: aha thanks ... so it seems to be the same issue then
16:42:55 <jlaska> okay folks ... we'll close up shop here and get to work if no other comments ...
16:43:16 <Oxf13> so here's my take on today
16:43:37 <Oxf13> if we get gdm fixed (which sounds like it is), and we get this partition traceback fixed, we can enter RC phase and I'll do a compose with these packages in a side repo
16:43:39 <jlaska> #topic Beta Compose Readiness
16:44:05 <adamw> if we can get the other candidate anaconda changes in the side compose that'd be good too
16:44:14 <adamw> (the others discussed in the meeting)
16:44:19 <Oxf13> oh shit
16:44:33 <Oxf13> there is the live partition thing too that dlehman posted but nobody reviewed :/
16:44:42 <Oxf13> but that's still anaconda
16:44:44 <jlaska> Oxf13: dlehman and I have been discussing that
16:44:51 <Oxf13> so looking at anaconda build and gdm build
16:45:12 <warren> hmm, no rawhide today yet?
16:45:27 <adamw> the compose completed during the meeting apparently...right jlaska?
16:45:34 <warren> Oxf13: I haven't successfully been able to run liveinst in order to test that patch
16:45:38 <Oxf13> warren: rsyncing still
16:45:41 <jlaska> adamw: is rawhidebrokrn hasn't fired yet
16:46:28 <jlaska> but we got the email, that usually means it's done right?
16:46:28 <jlaska> https://www.redhat.com/archives/fedora-test-list/2009-October/msg00132.html
16:47:31 <jlaska> okay, I think we're good to go here
16:47:39 <jlaska> any other issues/concerns?
16:47:55 <Oxf13> jlaska: I haven't gotten the mail here.  weird
16:47:58 <adamw> jlaska: that's yesterday's mail
16:48:09 <jlaska> haha
16:48:13 <adamw> just to note i still have to confirm the plymouth-not-firing bug has gone away
16:48:16 <Oxf13> rsync is still going
16:48:23 <jlaska> Oxf13: okay
16:48:31 <Oxf13> we've had rather lengthy rsyncs or composes lately, not sure which but need to investigate
16:48:36 <jlaska> we're good to go when https://bugzilla.redhat.com/showdependencytree.cgi?id=F12Beta&hide_resolved=1 is all MODIFIED
16:48:46 <jlaska> (minus the preupgrade bug)
16:48:48 <Oxf13> yeah
16:48:52 <adamw> jlaska: now i'm confused about that one, because any live image you're testing today should be from yesterday's package set, still? unless you built from the hourly repos or a sekrit internal one or something?
16:49:04 <jlaska> adamw: I used koji static repo
16:49:08 <adamw> ah k, that'd explain it
16:49:55 <jlaska> #info once the F12Beta bugs are all in MODIFIED (exluding preupgrade 526208) - RC compose will proceed
16:50:00 <jlaska> okay ... we're done
16:50:02 <jlaska> thanks folks
16:50:12 <jlaska> #endmeeting