Infrastructure/Meetings/2008-02-14

From FedoraProject

Jump to: navigation, search

Meeting of 2008-02-14


*** Time shown in EST

15:00 -!- mmcgrath changed the topic of #fedora-meeting to: Infrastructure =
-- Who's here?
15:00 < jima> not me!
15:00 < warren> m
15:00 -!- notting [n=3Dnotting@redhat/notting]  has joined #fedora-meeting
15:01 < mmcgrath> abadger1999: dgilmore f13 gregdek jcollie ivazquez iWolf =
J5 lmacken mbonnet paulobanon_ paulobanon ricky skvidal spoleeba yingbull  =
PING
15:01 < skvidal> hi
15:01  * dgilmore is here
15:01 < yingbull> pong
15:01  * abadger1999 is here
15:01 < spoleeba> mmcgrath, oh no
15:01  * lmacken is here
15:01 < skvidal> wow
15:01 < jcollie> mmcgrath: i'll brb in the middle of some trouble...
15:01 < skvidal> no one does the /me blank thing
15:01 < mmcgrath> jcollie: no problem.
15:01 -!- mbacovsk [n=3Dmbacovsk@44.252.broadband5.iol.cz]  has joined #fedo=
ra-meeting
15:01 < spoleeba> mmcgrath, im on a con call
15:01  * mmcgrath=20
15:01  * dgilmore=20
15:01 < mmcgrath> spoleeba: no worries.
15:01 < skvidal> ah, much better
15:01  * asgeirf *newbie* is here...
15:02 < mmcgrath> asgeirf: welcome!
15:02 < lmacken> skvidal: sorry, I forgot to flush the dircproxy users this=
week :(
15:02 < jima> ah, fresh meat!
15:02 -!- nim-nim [n=3Dnim-nim@fedora/nim-nim]  has quit Read error: 104 (Co=
nnection reset by peer)
15:02 < skvidal> lmacken: :)
15:02 < spoleeba> mmcgrath, trying to convince esmf developers to target fe=
dora
15:02 < mmcgrath> Ok, well lets get started.  We may have to put some thing=
s off until later as some people are busy right now, but thats not a proble=
m.
15:03 < skvidal> spoleeba: just keep the icbm developers from targeting fed=
ora
15:03 < jima> skvidal: i concur.
15:03 < mmcgrath> So lots has happened since our last meeting, because we m=
issed some in there.
15:03 < mmcgrath> .tiny https://fedorahosted.org/fedora-infrastructure/quer=
y?status=3Dnew&status=3Dassigned&status=3Dreopened&group=3Dmilestone&keywor=
ds=3D%7EMeeting&order=3Dpriority
15:03 < zodbot> mmcgrath: http://tinyurl.com/2hyyz6
15:03 < J5> pong
15:03 < f13> mmcgrath: howdy
15:03 < mmcgrath> so first ticket up is something notting asked about just =
today
15:03 < mmcgrath> .ticket 347
15:03 < zodbot> mmcgrath: #347 (Set localtime on all our servers to UTC) - =
Fedora Infrastructure - Trac - https://fedorahosted.org/projects/fedora-inf=
rastructure/ticket/347
15:03 < mmcgrath> J5: word
15:03 < dgilmore> mmcgrath: did i says id do it?
15:04 < mmcgrath> So when we do finally do this, we'll have cron jobs and a=
ll kinds of things to change.
15:04 < mmcgrath> dgilmore: yes you did.
15:04 < dgilmore> we need to look at cron job scheduling
15:04 < mmcgrath> I don't think anyone's against it, but we'll have to make=
sure we have our plan and do it.  Its going to take time to actually do it.
15:04 < dgilmore> mmcgrath: week of 25th ill be in boston.  so i think ill =
do it one night that week
15:05 < dgilmore> so lets work up a plan. and get word out there
15:05 < mmcgrath> dgilmore: solid, that works for me.  Go ahead and assign =
that ticket to you.
15:05 < mmcgrath> We *can* do this in steps if we want.  I'll leave that up=
to you (if you were volunteering for it that is)
15:05 < skvidal> mmcgrath: why do we have to change cron jobs? just to make=
sure they don't run at peak hours?
15:05 < mmcgrath> skvidal: yeah
15:05 < jima> oh, goody
15:05 < mmcgrath> especially with rawhide builds and stuff.
15:05 < skvidal> I mean have about 14 timezones where people are using fedo=
ra readily
15:06 < jima> my cron jobs run whenever
15:06 < skvidal> is there EVER a non-peak hour?
15:06 < f13> yes
15:06 < mmcgrath> For some machines yeah.  Not all of them though.
15:06 < jima> skvidal: non-peak US/EU is ideal, i think.
15:06 < f13> koji isn't nearly as busy between 1am -> 4am Eastern
15:06 < mmcgrath> I think many of our important cron jobs are running on th=
e hour anyway, but the buildsystem does have lulls
15:06 < f13> of course, doing rawhide during then makes koji a little busy =
(:
15:06 < jima> i do believe most of our contributors are in US/EU
15:06 < skvidal> jima: india is growing
15:07 < mmcgrath> skvidal: but the fact taht we're increasingly not having =
downtime is good :)  though it will cause us challenges in the future.
15:07 < mmcgrath> like when to back up what, etc.
15:07 < skvidal> right
15:07 < skvidal> all I'm wondering is maybe we leave the cron jobs alone
15:07 < mmcgrath> dgilmore: were you volunteering for that or did I misread=
you?
15:07 < skvidal> and see how things are
15:07 < skvidal> but change the system time
15:08  * iWolf slinks into the back
15:08  * mmcgrath thinks dgilmore might have been called away for a bit, we=
can get back to that if need be.
15:08 < dgilmore> mmcgrath: ill look at how it effects things
15:08 < mmcgrath> dgilmore: excellent, thanks.
15:08  * dgilmore was taking the ticket=20
15:09 < dgilmore> is slow due to no mouse
15:09 < mmcgrath> next ticket is 365 but jcollie is busy right now.
15:09 < mmcgrath> .ticket 270
15:09 < zodbot> mmcgrath: #270 (Fedora Wiki allows editing raw HTML) - Fedo=
ra Infrastructure - Trac - https://fedorahosted.org/projects/fedora-infrast=
ructure/ticket/270
15:09 < mmcgrath> paulobanon_: paulobanon: ricky: ping?
15:09  * mmcgrath hasn't seen them today much, might be busy.
15:09 < f13> skvidal: we'll have to change the rawhide compose cron
15:09  * mmcgrath notes he's going to have a whole special wiki talk after =
the tickets so we don't have to get into it too much.
15:09 < mmcgrath> moving on
15:09 < mmcgrath> .ticket 302
15:09 < zodbot> mmcgrath: #302 (Moin patches) - Fedora Infrastructure - Tra=
c - https://fedorahosted.org/projects/fedora-infrastructure/ticket/302
15:10 < f13> skvidal: we time that so that there is fresh rawhide content w=
aiting for RH folks to start work, who primarily start in UTC -5
15:10 < mmcgrath> Last I've heard these are still not upstream, I'm going t=
o un-meeting-ize this.
15:10 < dgilmore> mmcgrath: seems sane
15:10 < skvidal> f13: eastern standard tribe
15:10 < f13> yeah
15:10 < mmcgrath> netxt ticket
15:10 < mmcgrath> .395
15:10 < mmcgrath> also one for jcollie
15:10 < dgilmore> f13: we can work it
15:11 < mmcgrath> we actually did have that one working for a bit, expect i=
t to be deployed with the rest of asterisk.
15:11 < mmcgrath> next topic!
15:11 -!- mmcgrath changed the topic of #fedora-meeting to: Infrastructure =
-- The Wiki
15:11 < mmcgrath> so lets have a chat about this.
15:11 < abadger1999> Woo hoo!
15:11 < mmcgrath> I'd like to officially propose we move from Moin to Media=
wiki.
15:11 < skvidal> wikis are for losers
15:11 < skvidal> let's go with a big dir on fedora people
15:11 < notting> deep hurting.
15:11 < skvidal> anyone can edit
15:11 < mmcgrath> I've been doing some script conversions and thanks to iva=
zquez and unicode magic... I've been very happy with the results.
15:11 < jima> notting: deep stabbing?
15:12 < jwb> mmcgrath, how different are the sytaxes?
15:12 < dgilmore> mmcgrath: i want moin syntax
15:12 < mmcgrath> jwb: they're pretty different.
15:12 < jwb> uuff
15:12 < notting> mmcgrath: do  we have someone lined up for prettyfication?
15:12 < dgilmore> jwb: i personally hate mediawiki's syntax
15:12 < jwb> does mediawiki have a gui mode that makes that moot?
15:12 < lmacken> any way we can get a rich text editor plugin or something ?
15:12 < mmcgrath> Here's what we have so far - https://publictest1.fedorapr=
oject.org/wiki/Infrastructure
15:12 < mmcgrath> dgilmore: we might just be able to find a plugin for that.
15:13 < dgilmore> when ive had to use it ive had to copy paste and run rege=
xs in vi  to get it to do what is simple in moin
15:13 -!- glezos [n=3Dglezos@fedora/glezos]  has joined #fedora-meeting
15:13 < mmcgrath> There's going to be bits and pieces to change but for the=
most part.  Its been very good.
15:13 < dgilmore> mmcgrath: if you do ill be a happy man
15:13  * ivazquez apologizes for being late
15:13 < jwb> that looks like arse...
15:13  * glezos too
15:13 < mmcgrath> ivazquez: no worries
15:13 -!- mether [n=3Dask@fedora/mether]  has quit Read error: 113 (No route=
to host)
15:13 < dgilmore> lmacken: mediawiki does have soem wysiwyg editors
15:13 < mmcgrath> glezos: we're just talking about mediawiki - https://publ=
ictest1.fedoraproject.org/wiki/Infrastructure
15:13 < jwb> mmcgrath, maybe some of that doesn't work well from external c=
onnections?
15:13 < mmcgrath> jwb: we'll have someone make a template for it.
15:13 < warren> mmcgrath, would mediawiki be any easier or harder to integr=
ate with FAS?
15:14 < jwb> mmcgrath, no, i mean it looks broken
15:14 < mmcgrath> warren: easier.
15:14 < warren> cool
15:14 < mmcgrath> jwb: define broken?  as in there's the #! html box at the=
top?
15:14 < notting> hey, i could read InfrastructurePrivate w/o logging in
15:14 < jwb> mmcgrath, no... let me post a screenshot
15:14 < mmcgrath> warren: unlike moin, mediawiki has a full and documented =
api (someone has even written a fuse filesystem for it I hear) as well as a=
good plugin/extensions system.
15:15 < mmcgrath> Aside from dennis hating the markup, is anyone against th=
is?
15:15 < mmcgrath> poelcat: ping (thought you might want to know we're talki=
ng about this)
15:15 < skvidal> and an xml-rpc interface iirc
15:15 < notting> i am very ambivalent
15:15 < jwb> mmcgrath, http://jwboyer.fedorapeople.org/Screenshot.png
15:15 < notting> what do our most prominent wiki-ans think?
15:15 < skvidal> notting: eat more roughage
15:15 < ivazquez> I'm not against it, but that script needs quite a bit mor=
e tweaking.
15:15 < abadger1999> Do we know how it will scale under load?  Will the db =
be a bottleneck?
15:16 < yingbull> mmcgrath: I'm either way, but if it performs better that'=
s good.  Can we QA that?
15:16 < dgilmore> mmcgrath: we using mysql on the backend?
15:16 < iWolf> I just wonder if we will end up with a whole other set of pr=
oblems.
15:16  * abadger1999 wonders how new install smolt checkins and release day=
wiki traffic will work out
15:16 < mmcgrath> jwb: thats just the tables not being converted right, its=
still a work in progress.
15:16 < warren> iWolf, scalability wont be one of them...
15:16 < dgilmore> abadger1999: we should be able to cluster that.  and if w=
e are using mysql thats pretty easy to do
15:16 < ivazquez> iWolf: I don't doubt we will. It's just a matter of decid=
ing which set we can live with.
15:17 < yingbull> Would the mediawiki use an existing mysql farm, or would =
it have its own servers?
15:17 < notting> what do the docs people think?
15:17 < jwb> mmcgrath, ok...   i won't go posting that link profusely ;)
15:17 < mmcgrath> yingbull: yes, you can do a search on it now and its way =
way faster.
15:17 < mmcgrath> jwb: ;-)
15:17 < warren> Has moin ever corrupted itself?
15:17 < mmcgrath> dgilmore: right now yeah.  I don't want to put anything e=
lse on postgres for the time being.
15:17 < mmcgrath> warren: yes.
15:17 < yingbull> mmcgrath: that means its scales for content size, just wo=
ndering about actual load.
15:17 < warren> mmcgrath, how often?
15:17 < mmcgrath> warren depends, why?
15:17 < warren> just wondering
15:17 < dgilmore> mmcgrath: :)  i personally much prefer mysql
15:17 < abadger1999> dgilmore: heh.  I've had nothing but pain from cluster=
ing mysql but it could be how it was being used.
15:17 < warren> mmcgrath, because I would expect that from a non-RDBMS mult=
i-user concurrent system
15:18 < mmcgrath> warren: In fairness though the times it got corrupt we we=
re probably doing stupid things, there's been tickets about badness getting=
stuck in logs, people's accoutns not working right, etc.
15:18 < dgilmore> abadger1999: ive had success in the past clustering it
15:18 < dgilmore> ive not tried postgres
15:18 < mmcgrath> the other thing I like about mediawiki is we have multipl=
e options for scaling, caching, clustering, etc.
15:18 < warren> and they take security seriously
15:18 < dgilmore> mmcgrath: i do like that
15:19 < dgilmore> warren: they have to its php
15:19 < skvidal> dgilmore: :)
15:19 < warren> I'm guessing we would wall this into its own VM guest with =
strict selinux policies.
15:19 < mmcgrath> So if I went to other teams and said "The infrastructure =
team would like to migrate to mediawiki"  Thats not a false statement as fa=
r as anyone here is concerned?
15:19 < abadger1999> +1
15:19 < warren> +1
15:19 < yingbull> Sounds good.
15:19 < lmacken> +1
15:19 < J5> +1
15:20 < lmacken> what about plone!
15:20 < lmacken> (j/k)
15:20 < EvilBob> yeah thanks
15:20 < mmcgrath> ok, I'll get that conversation started to the various wik=
i users.
15:20 < skvidal> +1
15:20 < iWolf> +0 (defers to the more active members)
15:20 < glezos> mmcgrath: +1 from the L10n part -- will get back to you wit=
h more info in a week, after I meet with the mediawiki i18n folks at FOSDEM.
15:20 < ivazquez> +1
15:20 < mmcgrath> Its not a done deal that we'll move but if we can satisfy=
all of the teams, it will be.
15:20 < asgeirf> 0 :)
15:20 < glezos> I wonder whether the fact that mediawiki being a very activ=
e upstream project will make it harder for us to push any code of ours there
15:20 < mmcgrath> oh!
15:20 < mmcgrath> that reminds me.
15:21 < warren> glezos, nothing can be harder than upstream moin who doesn'=
t want any contributions
15:21 < ivazquez> It might also mean that it's less necessary.
15:21 < skvidal> ivazquez: +1
15:21 < mmcgrath> glezos talked to some of the mediawiki guys and will be t=
alking to them at fosdem (i think) so even before we deploy it, there might=
be a strong partnership between the Fedora Project and the mediawiki guys.=
Which is a very good thing and in stark contrast to the relationship with=
moin-upstream.
15:21 < dgilmore> +1
15:21 < iWolf> mmcgrath: that's a good thing
15:21  * abadger1999 changes his vote to +100 :-)
15:22 < mmcgrath> heh
15:22 < glezos> mmcgrath: the mediawiki guys have been *very* positive in d=
iscussing and looking at the issues we'll have, especially in terms of tran=
slations.
15:22 < mmcgrath> ok, well, there's not much else to talk about there, I'm =
going to continue working on that script.  I've got to get the tables worki=
ng right.  I'll try to get back with mizmo to see about a better template f=
or that place.
15:22 < glezos> I'll make sure to report back with the meeting's points.
15:22 < mmcgrath> glezos: I'm happy to hear that.
15:22 < mmcgrath> ok, next topic
15:23 -!- mmcgrath changed the topic of #fedora-meeting to: Infrastructure =
-- Fedora Hosted.
15:23 < mmcgrath> Two things here.
15:23 < spoleeba> telecon done
15:23 < mmcgrath> first, we now have mtn support on fedorahosted + trac - h=
ttps://fedorahosted.org/elfutils/browser
15:23 < mmcgrath> spoleeba: excellent.
15:23 < mmcgrath> The next (and bigger thing)
15:24 < mmcgrath> serverbeach (who provides all the hosting for fedorahoste=
d) would like to do a joint pressrelease about fedorahosted.
15:24 < dgilmore> we are taking on sourceforge?
15:24 < dgilmore> :)  cool
15:24 < mmcgrath> The last bit that we'd need to get done AFAIK, is the mai=
ling lists which jcollie is extremely close to completing.
15:24 < mmcgrath> https://fedorahosted.org/mailman/listinfo
15:24 < warren> yay!
15:24 < mmcgrath> Once thats available, I think we should do the announceme=
nt.
15:24 < mmcgrath> f13: what do you think?
15:24 < warren> mmcgrath, does that mean we have our own MTA?  control our =
own spam filtering?
15:24 < dgilmore> mmcgrath: sounds good
15:25 < mmcgrath> warren: we have that now, just no one's implemented the s=
pam filtering.
15:25 < spoleeba> mmcgrath, i officially have no problem with a joint press=
-release...do we get to look at it first?
15:25 < f13> hrm.
15:25 < warren> mmcgrath, ok, I'm very interested in that part.
15:25 < mmcgrath> spoleeba: I'm sure we could, max and paul are already inv=
olved in all of that.
15:25 < f13> there was one more thing I wanted done before we left 'beta' s=
tage.
15:25 < f13> that was raw webspace for hosted projects.
15:25  * mmcgrath actually wanted to not provide that.
15:26 < warren> we need that if we're going to have a complete upstream hos=
ting solution
15:26 < mmcgrath> f13: how much storage space did you have in mind for that=
?  I figured the wiki would be enough.
15:26 < f13> mmcgrath: wiki eh?
15:27 < mmcgrath> f13: if we're going to do that we need to come up with a =
solution quick, they want to do the release soon unless thats the sort of t=
hing that doesn't block the official announcement.
15:27 < iWolf> It could always be a later feature enhancement.
15:27 < f13> we have a working model in what seth setup for fedorapeople...
15:28 < mmcgrath> f13: we do have https://fedorahosted.org/releases/p/y/pyt=
hon-fedora/
15:28 < skvidal> we could nfs export across the vpn :)
15:28 < mmcgrath> f13: except that working model uses full shell accounts.
15:28 < mmcgrath> something which would require us to completely re-design =
our security settings.
15:29 < f13> mmcgrath: the wiki is a bit harder to automate attaching to
15:29 < f13> but it's not the worst option in the world.
15:29 < mmcgrath> f13: no need to attach to the wiki, we have the releases =
ability now.
15:29 < mmcgrath> people can scp the files they want up there.
15:29 < f13> mmcgrath: uh... so the people who are doing fedora hosted proj=
ects /already have/ fedora people space.
15:29 < warren> mmcgrath, and using fedorapeople can we sustainably have mu=
ltiple upstream project developers put files in a repository and get permis=
sions right?
15:29 < f13> I'm not sure what you're afraid of.
15:30 < warren> f13, 2 or more people working on the same project?
15:30 < skvidal> warren: acls
15:30 < f13> mmcgrath: I would have hoped the webspace would be more like '=
python-fedora.fedorahosted.org/'
15:30 < glezos> f13: that does sound like a nice feature to have for a host=
ed solution.
15:31 < mmcgrath> f13: is this something you really want to block the annou=
ncement?  if it is we need to come up with a solution for it now.
15:31 < skvidal> f13: A couple of hangups
15:31  * jcollie is back
15:31 < skvidal> 1. we need a box
15:31 < notting> how is fedorahosted.org/releases populated now?
15:31  * mmcgrath thinks this is something we "could" do but shouldn't.
15:31 < skvidal> 2. we need to say explicitly to the users: no php, cgi, etc
15:31 < f13> mmcgrath: I'd be ok announcing it, if we have at least agreed =
upon a future feature of having hosted raw webspace.
15:31 < warren> f13, we wouldn't want to put upstream project tarballs into=
the mirror structure?
15:32 < notting> f13: nah
15:32 < notting> erm
15:32 < notting> warren: nah
15:32 < warren> http://download.fedora.redhat.com/pub/fedora/projects/
15:32 < mmcgrath> notting: its in the faq.  If you don't have an actual she=
ll on there (which most people don't) you just 'scp filename-1.0.tar.gz fed=
orahosted.org:projectName' some voodoo on the backend sends it to the right=
place.
15:32 < warren> If we mirror it, we don't need to back it up. =3D)
15:32 < mmcgrath> warren: we could announce to mirrors its available though=
, run rsync fedorahosted::
15:32 < mmcgrath> its in there and available to be synced out.
15:32 < warren> mm
15:32 < f13> mmcgrath: can we do dns tricks so that <project>.fedorahosted.=
org works ?
15:33 < f13> I know it's asthetics but it /does/ look better
15:33 < mmcgrath> f13: what would people see at project.fedorahosted.org ?
15:33 < mmcgrath> the trac instance?
15:33 < f13> mmcgrath: anybody reading a spec file.
15:33 < lmacken> can we assume most of the fedorahosted projects are actual=
ly in Fedora?  if so, new releases can just entail pushing into fedora.  Th=
en we wouldn't need raw webspace ?
15:33 < f13> mmcgrath: spec files reference the url to the tarball release
15:33 < f13> lmacken: that is terrible for cross-distro
15:34 < f13> lmacken: it's extremely rude to tell Debian "get the tarball f=
rom the source rpm in Fedora release Blah"
15:34 < mmcgrath> f13: no I mean if I typed project.fedorahosted.org into m=
y browser, what comes up?
15:34 < lmacken> not if we link to the tarballs ?
15:34 < f13> mmcgrath: directory listing of https://fedorahosted.org/releas=
es/p/y/python-fedora/
15:34 < mmcgrath> lmacken: besides, we already have that
15:34 < skvidal> f13: not the hosted page?
15:34 < lmacken> or somehow autopopulate builds onto the wiki
15:34 < f13> ok, wait
15:34 < mmcgrath> ahhh
15:34 < skvidal> f13: ie: not the wiki instance?
15:34 < f13> in my /grand/ thought
15:34  * mmcgrath sees what f13 is saying here.
15:35 < f13> <projectname>.fedorahosted.org  would lead to raw webspace tha=
t hte project could do whatever within our rules
15:35 < f13> a pretty page
15:35 < f13> fedorahosted.org/project/ leads you to Trac
15:35 < f13> which often isn't nearly as pretty
15:35 < f13> and yes, this is a lot like how sourceforge works.
15:35  * mmcgrath thinks thats confusing.
15:35 < mmcgrath> so what is koji's home page?
15:36 < f13> mmcgrath: it depends on what the project wnats to do
15:36 < f13> mmcgrath: making every project's home page the Trac instance i=
s kind of lame
15:36 < skvidal> f13: why?
15:36 < mmcgrath> thats confusing though, some projects will use project.fh=
=2Eo some will use fh.o/project  If I want to check out elfutils, how will =
I know which one to go to?
15:37 < f13> mmcgrath: you're coming at this with prior knowledge that elfu=
tils is hosted with us
15:37 < f13> mmcgrath: invalid use case.
15:37 < f13> mmcgrath: people who are looking for the upstream of elfutils =
is going ot read the elfutils spec or documentation or whatever and be dire=
cted at their homepage
15:37 < f13> or you'll pull up the Trac instance which teh wiki there could=
say "See <here> for our homepage"
15:38 -!- stickster is now known as stickster_afk
15:38 < mmcgrath> f13: I even know elfutils is hosted with us because they =
told me, now I want to see it.  Which one do I go to?
15:38 < warren> mmcgrath, even downloading tarballs from sourceforge is a b=
it ambiguous because people have different URL's depending on how they down=
loaded it
15:39 < warren> mmcgrath, if we want to eliminate different URL's then we h=
ave to enforce it with redirects.
15:39 -!- stickster_afk is now known as stickster
15:39 < notting> w.r.t download space, if all we have is the wiki, how are =
files pushed through that, and does tracwiki scale for serving lots of larg=
e files?
15:39  * f13 sees this discussion going nowhere in a hurry
15:39 < mmcgrath> notting: actually the wiki is horrible for pushing files.
15:39 < lmacken> right now elfutils points directly to koji for new release=
s.
15:39 < abadger1999>  f13: Are we more like launchpador sourceforge?
15:39 < f13> mmcgrath: 'the wiki' are you talking about Moin or Trac?
15:39 < abadger1999> (Where we want to be)
15:39  * mmcgrath doesn't compare us to either of them.
15:39 < mmcgrath> f13: Trac.
15:39 < f13> IIRC Trac just is a direct link to the attachment on the files=
ystem
15:40 < warren> Ubuntu doenst bother to make tarballs, they build directly =
=66rom bzr on launchpad for things that they are upstream.
15:40 < f13> abadger1999: I'm more familiar with sf.
15:40 < f13> http://sourceforge.net/projects/pidgin vs pidgin.sf.net
15:41 < f13> hrm, perhaps not the best example
15:41 < abadger1999> Just wondering because sf provides web space via shell=
access while launchpad provides an external links section where you can li=
st a home page if you want more than they provide.
15:41 < f13> but when I dreampt up fedora hosted, I pictured Trac as one of=
the tools a hosted project would have, not the /only/ tool.
15:41  * warren entirely supports f13 on this.
15:42 < f13> and where a number of projects might have a rich home page for=
users, and use the Trac instance as a development tool
15:42  * mmcgrath saw it as a simple and complete set of tools.
15:42 < notting> ferexample, we may have a project that wants just 1) downl=
oad space 2) git + gitweb 3) a mailing list. how does trac help them?
15:42 < f13> that said, maybe our "customers" are perfectly happy with only=
having Trac.
15:42 < mmcgrath> f13: what can you do on raw space that you can't do with =
trac wiki?
15:42 < skvidal> mmcgrath: well, the page controls are better
15:42 < mmcgrath> notting: FWIW, we have lots of people that don't have a t=
rac instance.
15:42 < f13> mmcgrath: html, easily manage content programatically
15:42 < skvidal> f13: umm
15:42 < skvidal> wait
15:43 < lmacken> Hmm.. I've seen trac used as a pretty functional homepage =
as well
15:43 < lmacken> http://cherrypy.org
15:43 < warren> and some people don't want trac at all because it can be a =
distraction
15:43 < skvidal> f13:  are we talking about RAW stuff like php or cgis b/c =
that's not bloody likely
15:43 < notting> mmcgrath: right, but if they want to tie together a link t=
o gitweb, a link to the ML, etc. ... trac is the only solution atm
15:43 < mmcgrath> The bottom line is we can't provide everything to everyon=
e.
15:43 < mmcgrath> we have a "web" solution do we need 2?
15:43 < f13> skvidal: no of course not, there would be restriction around w=
hat could be used.
15:43 < warren> Hosted should be a menu of services that upstream projects =
can pick from.
15:43 < mmcgrath> warren: it is and will be.
15:43 < mmcgrath> but whats on the menu?
15:43 < f13> lets look at 108 as a great example
15:44 < f13> people hated 108 because to get anything on the web, they eith=
er had to fuck with wiki, or svn
15:44 < f13> what they really wanted was the ability to just rsync (via ssh=
) content into a dir
15:44  * mmcgrath is pretty sure thats not why people hated 108
15:44 < mmcgrath> it was closed source and complicated.  Trac is neither of=
those.
15:44 < f13> mmcgrath: that was on eof the big complaints I heard from a lo=
t of people trying to use it to host projects.
15:44 < f13> mmcgrath: from the usability side
15:44 < mmcgrath> Ok, lets take a step back.
15:44 < mmcgrath> is this a blocker to the announcement?
15:45 < skvidal> not imo
15:45 < f13> I'm leaning toward no, as long as we have a roadmap for someth=
ing like this we can make public
15:45 < warren> We should at least define exactly what we want and write it=
as a "coming soon" feature on the roadmap.
15:45 < mmcgrath> Ok, then we should discuss this at another time.
15:45 < notting> yeah, as long as we have a plan so we can tell people 'yes=
', 'no', 'yes, but in 3 months' when they ask for foo, bar, or baz
15:45  * mmcgrath is already tired of the slippery slope we're headed down =
with fedorahosted.
15:45  * ricky gets here.
15:46 < mmcgrath> everything some says 'wouldn't it be nice" we've implemen=
ted the feature, its got to stop.  There is beauty in simplicity.
15:46  * f13 notes that he's been making noise about having web space for p=
rojects since day 1.
15:46 < mmcgrath> and some how, over 100 projects have thrived while not ha=
ving web space.
15:46 < dgilmore> mmcgrath: i want a pony
15:46 < f13> mmcgrath: uh, other than adding additional SCMs, and a couple =
trac plugins, what exactly ahve we done?
15:46 < mmcgrath> f13: I'm sorry I just don't remember that at all.  I'm pr=
obably rembering it wrong but I don't remember anything about raw webspace =
ever.
15:46 < warren> mmcgrath, this is one of the few things that we really need=
to make hosted complete.
15:47 < J5> I have to agree.  By giving people the world we make it harder =
to create a good experience
15:47 < mmcgrath> f13: just that.  I want to make sure it stops there.
15:47 < mmcgrath> warren: I think hosted is complete.
15:47 < notting> is the scp fedorahosted:blah magic work for anything?
15:47 < notting> or just blah-1.2.3.tar.gz?
15:47 < mmcgrath> notting: only if you don't have a shell account.
15:47 < mmcgrath> scp anything fedorahosted:project
15:47 -!- mbacovsk [n=3Dmbacovsk@44.252.broadband5.iol.cz]  has quit Read er=
ror: 110 (Connection timed out)
15:47 < mmcgrath> warren: excpet for mailing.
15:47 < notting> so we have webspace ;)
15:47 < warren> f13, I do recall that from day 1 as well.
15:47 < warren> mmcgrath, which is already on the roadmap
15:48 < f13> mmcgrath: perhaps we should survey our current customers, and =
potential customers (108 holdouts, et.redhat.com folks, people.redhat.com f=
olks) about it.
15:48 < mmcgrath> f13: I think thats the wrong way to go.  We need to have =
in mind a solution that we want it to be.  If its the right tool for those =
people they will come, if not they won't.
15:48 < dgilmore> f13: most if not all people haviing people.redhat.com spa=
ce ahve fedorapeople.org space
15:48 < mmcgrath> we can't conform to everyone's needs on this or it will c=
onsume ALL of our time like 108's did the 108 people.
15:49 < skvidal> okay
15:49 < skvidal> let's chill
15:49 < skvidal> everyone
15:49 < warren> mmcgrath, this is not a slippery slope.  Raw web space is o=
ne of the fixed set of things people expect for an upstream project.
15:49 < dgilmore> skvidal: :)
15:49 < mmcgrath> this is a free 'value added' fedora service.  If we end u=
p spending lots of time on it, thats less time for other things.
15:49 < dgilmore> lets move on now
15:49 < skvidal> yes
15:49 < skvidal> we're not blocking announcing it
15:49 < skvidal> we can sort this out later
15:49 < dgilmore> no  we can announce as is
15:49 < skvidal> on a day when we're all less busy
15:49 < dgilmore> i think its done being a beta
15:49 < skvidal> and less stress-y
15:49 -!- mmcgrath changed the topic of #fedora-meeting to: Infrastructure =
-- Asterisk
15:50 < dgilmore> :) it works
15:50 < skvidal> we'll learn a lot about what people want from tickets
15:50 < mmcgrath> The Board has put a higher priority on our asterisk setup.
15:50 < skvidal> eyeroll
15:50 < skvidal> sorry, the board has an asterisk fetish for odd reasons :)
15:50 < mmcgrath> asterisk as it stands has enough shortcommings that it wo=
n't be usable for us without a few things.
15:50 < spoleeba> skvidal, i have an n810 now.. id like to use it for..some=
thing
15:50 < mmcgrath> 1) they've requested town-hall style meetings.
15:50 < dgilmore> skvidal: we do
15:50 < skvidal> dgilmore: :)
15:50 < mmcgrath> jcollie: do you want to talk about that for a moment and =
the implementation you tested?
15:50 < jcollie> ujh sure
15:50 < jcollie> erg
15:51 < jcollie> it's ticket 395
15:51 < mmcgrath> .ticket 395
15:51 < mmcgrath> :)
15:51 < zodbot> mmcgrath: #395 (Audio Streaming of Fedora Board Conference =
Calls) - Fedora Infrastructure - Trac - https://fedorahosted.org/projects/f=
edora-infrastructure/ticket/395
15:52 < jcollie> basically the board wants to have some conferences calls s=
treamed out so that 10s or maybe even 100s of people can listen in live
15:52 < dgilmore> taking feedback via irc
15:52 < jcollie> so i've figured out a way to get the audio out of the aste=
risk conference call and stream it using flumotion
15:53 < dgilmore> jcollie: and its all FOSS?
15:53 < jcollie> yup
15:53 < dgilmore> streaming ogg
15:53 < spoleeba> jcollie, thats a nice valentines day present
15:53 < jcollie> its streaming ogg/vorbis right now but i was going to test=
ogg/speex
15:54 < jcollie> fluendo even has a java applet for those unfortunate to be=
running windows
15:54 < jcollie> the java applet is gpl too i think
15:54 < dgilmore> cool
15:54 < warren> bbl
15:54  * mmcgrath notes he used this system the other day.  It actually wor=
ked, rather well.
15:54 < jcollie> the java applet isn't in fedora yet but i have a srpm unde=
r development
15:55 < lmacken> good deal
15:55 < ivazquez> Does IcedTea support audio yet?
15:55 < glezos> jcollie: sounds great
15:55 < spoleeba> jcollie, so where is it at right now?  does all the board=
members need to test it?
15:55 < jcollie> also, i'd prefer to use a development version of flumotion=
than what is in fedora right now but i'm not stuck on that
15:55 < mmcgrath> spoleeba: its not at that point yet
15:55 < mmcgrath> our asterisk deployment isn't even official yet, we'll ne=
ed to get that finalized and in puppet before we start considering these th=
ings.
15:56 < jcollie> i just did a technology preview to see if i could get the =
audio out of asterisk and into flumotion
15:56 < spoleeba> mmcgrath, the big concern was making sure connected audio=
source clients...weren't stupidly configed
15:56 < jcollie> yeah, we need to make sure that all the board members have=
decent headsets and a working SIP client setup
15:56 < spoleeba> jcollie, ha!
15:56 < spoleeba> jcollie, 'decent'
15:57 < jcollie> the mics built into laptops pick up too much ambient noise
15:57 < poelcat> mmcgrath: sorry on another call... yay for Mediawiki :)
15:57 < jcollie> spoleeba: actually you can get by with some really cheap o=
nes - i use these generic labtec ones
15:57 < spoleeba> jcollie, i want to test my n810 :->
15:57 < mmcgrath> I'm actually working on trying to get some dial in number=
s for that.  We'll see how it goes but areacodes in boston and raleigh woul=
d help a lot.
15:57 < jcollie> so no need to spend $100s
15:58 < spoleeba> jcollie, just let me know when you wwant me to test it
15:58 < jcollie> dial in numbers should be pretty easy to find
15:58 < notting> 888-fed-ora1?
15:58 < skvidal> umm
15:58 < skvidal> eww
15:58 < dgilmore> most of the board brought headsets when we looked at usin=
g asterisk for board meetings
15:58 -!- jmtaylor [n=3Djason@69.212.251.202]  has joined #fedora-meeting
15:58 < mmcgrath> notting: the 888 numbers require per minute charges.
15:59 < mmcgrath> Ok, hey guys we're running out of time.  There's a couple=
of other things I wanted to get to.
15:59 < dgilmore> notting: we have to pay for that
15:59 < jcollie> i'm also thinking about using an IRC bot to control the co=
nference
15:59 < dgilmore> jcollie: awesome
15:59 < dgilmore> mmcgrath: move on
15:59 < mmcgrath> jcollie: sorry to cut you off :(
15:59 < jcollie> yeah, the board doesn't want to do this until sometime in =
april anyway
15:59 -!- mmcgrath changed the topic of #fedora-meeting to: Infrastructure =
-- FAS2
15:59 < mmcgrath> we're moving to database with that, abadger1999 has been =
very helpful and ricky and I will be back on it hard core next week.
15:59 < mmcgrath> and last thing
16:00 -!- mmcgrath changed the topic of #fedora-meeting to: Infrastructure =
-- My Fedora
16:00 < mmcgrath> J5:  how's that going?
16:00  * mmcgrath hopes J5 is still around.
16:00 < J5> been working on the build page - it has been slow since I am le=
arning new things
16:00 < J5> But it is coming on nicely
16:00 < mmcgrath> is there a demo up anywhere yet?
16:01 < J5> just in git
16:01 < J5> I'll get something up by next week
16:01 < J5> the end of next week that is
16:01 < mmcgrath> <nod> no worries.  Do you have an ETA for when you'd like=
to deploy?
16:01 < ricky> Probably sometime after FAS2, I assume?
16:01 < J5> yes
16:02 < J5> well we can get the non-login stuff (it is still useful for sea=
rching packages and quickly navigating them)
16:02 < mmcgrath> <nod>
16:02 < J5> before fas2
16:02 < mmcgrath> J5: thanks for that.
16:02 < mmcgrath> alrighty, before we go
16:02 -!- mmcgrath changed the topic of #fedora-meeting to: Infrastructure =
-- Open Floor.
16:02 < mmcgrath> Anyone have anything else?
16:02 < skvidal> how about them backups?
16:03 < mmcgrath> ahh, yes dgilmore has been working on backups.
16:03 < skvidal> yay for all the work dgilmore and mmcgrath put in to make =
tape backups work
16:03 < mmcgrath> dgilmore: how's that going?
16:03 < dgilmore> skvidal: the backup on /mnt/koji is running very slowly
16:03  * mmcgrath notes tapes are in and AFAIK backing.
16:03 < mmcgrath> dgilmore: how long do you suspect a full koji backup will=
take?
16:03 < dgilmore> mmcgrath: at the rate we are going days
16:03 < mmcgrath> how many though?
16:04 < jcollie> what kind of tape hw do we have?
16:04 < mmcgrath> its  a TL2000 with LTO-3 tapes.
16:04 < jcollie> are we still using bacula?
16:04 < dgilmore> since 10pm cst we have backed up 300gb of 2tb
16:04 < dgilmore> jcollie: yeah
16:05 < jcollie> this just uses a single LTO drive then?
16:05 < mmcgrath> jcollie: yeah.
16:05 < mmcgrath> dgilmore: anything else?  We should probably wrap it up.
16:05 < jcollie> hmm too bad
16:06 < dgilmore> mmcgrath: its running and is working
16:06 < dgilmore> we will need to do some test restores
16:06 < mmcgrath> yeah
16:06 < dgilmore> but thats it
16:07 < mmcgrath> k.
16:07  * mmcgrath looks forward to that.
16:07 < mmcgrath> Ok, anyone have anything else?
16:07  * dgilmore has nothing
16:07 < jcollie> nope
16:08 < mmcgrath> allllllrighty
16:08 -!- mmcgrath changed the topic of #fedora-meeting to: Infrastructure =
-- Meeting Closed