ReleaseEngineering/Meetings/2008-jan-07

Fedora Release Engineering Meeting :: Monday 2008-01-07

IRC Transcript

-!- Irssi: #fedora-meeting: Total of 78 nicks [1 ops, 0 halfops, 0 voices, 77 normal]		13:03
poelcat	f13: meeting today?	13:03
f13	yeah	13:05
-!- f13 changed the topic of #fedora-meeting to: Fedora Release Engineering Meeting		13:06
f13	ping notting jeremy jwb rdieter wwoods spot warren poelcat lmacken	13:06
jwb	yo	13:06
warren	yo	13:06
* lmacken		13:06
* spot is heah.		13:07
rdieter	word	13:07
-!- f13 changed the topic of #fedora-meeting to: Fedora Release Engineering - F9 Alpha!		13:08
f13	yeah, freeze for Alpha is 8 days away.	13:09
warren	yet another secondary arch?	13:09
notting	<klaxon starts sounding>	13:09
f13	my plan for today is to do a compose across the arches and see what pops as unresolvable deps	13:10
f13	those we'll want to focus on getting fixed	13:10
notting	jwb: are you doing the ppc composes?	13:10
spot	i'm going to do perl 5.10 soon.	13:10
warren	f13,  should we kick maintainers into fixing F8 > F9 issues?	13:10
spot	soon as in, today or tomorrow	13:11
jwb	notting, i believe that was the plan	13:11
f13	warren: if you'd like.	13:11
warren	f13, I'm still on vacation	13:11
notting	beyond unresolvable deps, do we have an idea as to what is completely hosed?	13:11
jwb	however i wouldn't be opposed if f13 did some	13:11
jwb	f13, do we have koji tags for this already?	13:12
f13	no	13:12
f13	we create the tag on Tuesday	13:12
notting	if we're not freezing, why do we need tags?	13:12
f13	or more likely monday night.	13:12
jwb	so you're just composing rawhide?	13:12
f13	notting: we need a tag to compose from.	13:12
f13	notting: it's a non-blocking freeze, we're not blocking rawhide from moving on	13:12
f13	we're just taking a snapshot of it and updating as releng sees fit to get the alpha out	13:13
f13	jwb: until tuesday yes.	13:13
notting	is it on releng to find new things that fix stuff, or people to ping releng?	13:13
f13	notting: installer is pretty hosed right now.  url method dead, network on x86_64 dead (haven't tried ppc)	13:13
f13	notting: both, and with QA involved.	13:13
jwb	video is dead	13:13
f13	yeah, X is not starting well during install	13:13
notting	aside from that, how was the play. mrs. lincoln?	13:13
f13	and logging into X is pretty dead from GDM	13:14
f13	at least logging into gnome	13:14
spot	intel X is pretty much hosed up too	13:14
jwb	gdm as a whole is dead	13:14
spot	GL ops fail, and X won't restart cleanly	13:14
f13	notting: so all in all we're in pretty good shape!	13:14
f13	oh and we're missing a good chunk of firstboot due to things up in the air.	13:15
lmacken	ship it!	13:15
jwb	f13, your compose configs are where again?	13:16
jwb	pungi repo?	13:16
* jwb blames xmas for killing most of his braincells		13:17
f13	jwb: yes, pungi repo.	13:17
f13	conversely the live image configs are in the live repo	13:18
jwb	aye	13:18
jwb	f13, were you planning on using the quad to do composes or no?	13:18
f13	so it's going to be a fun alpha.	13:18
f13	jwb: if I'm going to do any ppc testing, sure.	13:19
jwb	are you?	13:19
f13	I wouldn't be opposed.	13:19
f13	mostly I only have the ps3 left for ppc testing though	13:19
warren	wow	13:19
jwb	understood.  if you make the isos available, i can give them to omar	13:19
f13	yeah, I was hoping to upload isos in progress to koji.fp.o/releng/	13:20
jwb	excellent.  i'll also be doing some composes on my own, but you have about 2x the machine i do :)	13:20
f13	heh	13:20
jwb	at least until i get my new one	13:20
f13	Seth also has a new createrepo for me to try out	13:21
f13	so I'll be throwing that at the compose process too and trying to turn up any weirdness, which my useage set usually does	13:21
skvidal	and it might be broken! :)	13:21
* skvidal goes back to lurking		13:21
f13	lmacken will be handling the live image creations	13:21
lmacken	indeed	13:21
jwb	lmacken, i assume you won't be doing ppc?	13:22
lmacken	jwb: correct.. unless we want to do them, and I get access to a machine to do them on	13:22
jwb	that largely won't be a loss, as they were ppc32 IIRC	13:22
f13	lmacken: you could have access to the quad	13:23
jwb	lmacken, as for access...  the quad or dwmw2 would gladly give you access :)	13:23
lmacken	cool, so if we want to spin them, I'd be happy to do so	13:23
f13	the quad needs updating badly though.  I should do that soon	13:23
-!- f13 changed the topic of #fedora-meeting to: Fedora Release Engineering - FUDCon		13:23
f13	I'll be at fudcon this weekend, willing to do either barcamp or hackfest stuff surrounding releng	13:24
f13	I do want to have a bit of a hackfest to re-evaluate how we do rawhide and find some better ways of doing it	13:24
f13	more robust, less hack-o-doom	13:24
f13	also we have to really look at reversing the netapp stream and what that's going to mean for releng and associated groups like QA	13:25
jwb	something like debian experimental, debian unstable, and debian stable?	13:25
f13	jwb: no no, just the methodology we use to compose rawhide	13:25
jwb	oh good	13:25
f13	jwb: it's all kinds of whacky	13:25
notting	that's putting it mildly	13:25
f13	but it "works" for now, which is why we haven't touched it much	13:25
notting	f13: although, is it really *that* different than how alpha/beta/final is composed, aside from being automated?	13:26
f13	notting: there are some significant differences.	13:26
f13	not that how we do the other composes is great mind you.	13:26
f13	we really need koji run-root capability, but there are other things we can clean up in the mean time.	13:26
jwb	from an outsider, the largest complication is that it does it all internally to the builder network	13:27
warren	run-root meaning?	13:27
jwb	which, of course, is the speedy part	13:27
notting	warren: a way to run commands on another arch inside a mock-created chroot	13:27
f13	warren: a koji process to run a specific command in a chroot created from a specific tag.	13:27
f13	like say pungi	13:28
notting	because, wow, is pungify a horror	13:28
f13	yes	13:28
f13	notting: so perhaps we can sit down and clean up some of that this weekend.	13:29
f13	you know, right before we freeze (:	13:29
notting	heh	13:29
-!- f13 changed the topic of #fedora-meeting to: Fedora Release Engineering - Open Discussion		13:30
f13	Anything anybody would like to discuss?	13:30
jwb	how are we going to handle the alpha tagging?	13:31
jwb	send email to rel-eng saying "i tagged this"?	13:31
f13	jwb: you mean releng people tagging things for the alpha?	13:31
jwb	yes	13:31
f13	We should use hte same methodology we did for previous freezes	13:32
f13	if we do one ourselves that's not externally requested IRC would likely work OK too	13:32
notting	subscribe rel-eng@fp.o to koji tag notifications :P	13:32
jwb	i don't foresee a large problem.  just curious	13:32
f13	notting: could do that.	13:33
f13	at least for that tag	13:33
jwb	except it doesn't tell you why	13:33
f13	right	13:34
f13	anything else?	13:37
wwoods	yeep. back	13:37
* wwoods reads scrollback quickly		13:38
f13	wwoods: I'll wait.	13:38
wwoods	I don't think url installs are broken (I just did one, and other than the 'user' command in kickstart causing a traceback, it worked)	13:39
wwoods	'course that means grubby didn't run, so it's not bootable...	13:40
wwoods	I'd really like to help straighten out the pungify / mash stuff. or at least help myself understand how we go from koji -> tree	13:40
wwoods	but yeah, I'll be going into Release Testing Mode soon	13:40
wwoods	and bothering people about bugs that screw up the alpha release	13:40
f13	wwoods: in boot.iso installs a sigsegv is thrown.	13:42
f13	wwoods: it's a newt issue, that is already fixed upstream.	13:43
f13	(upstream anaconda that is)	13:43
wwoods	gotcha	13:44
wwoods	this was network (using already-existing grub)	13:44
wwoods	anyway, yeah. nothing amazingly important to say. standard "oh shit oh shit release time" drill	13:44
f13	k	13:45
wwoods	I'll try to keep http://wwoods.fedoraproject.org/rawhide.html up to date	13:45
f13	k	13:45
wwoods	hopefully some of the hackfest time can go towards making that actually dynamic and automatic	13:46
wwoods	that's all I got.	13:47
f13	ok, if there's nothing else, ending in 30.	13:47

Generated by irclog2html.py 2.3 by Marius Gedminas - find it at mg.pov.lt!