Infrastructure > Meetings -> 2006-11-30
Meeting of 2006-11-30
*** Time shown in EST
15:07 < dgilmore> everyone here?
15:07 < mmcgrath> yo
15:07 * mmcgrath is late.
15:07 < mmcgrath> Who all is here?
15:07 < dgilmore> warren, lmacken, mmcgrath, c4chris, iWolf ping
15:07 * lmacken is here
15:08 * jcollie is here - i'm one of the "new guys"
15:08 < dgilmore> jcollie: welcome
15:08 < dgilmore> schedule is http://fedoraproject.org/wiki/Infrastructure/Schedule
15:08 * teknofile is a "new guy" too
15:08 < mmcgrath> welcome
15:09 < mmcgrath> dgilmore: you want to take it today?
15:09 < dgilmore> PckageDB c4chris, abadger1999 whats the good word?
15:09 < dgilmore> mmcgrath: sure
15:09 < mmcgrath> ;-)
15:09 < abadger1999> I'm orking on the schema
15:09 < abadger1999> f13 thinks we might get brew open sourced though.
15:10 < abadger1999> And that has implications both for buildsystem and for packageDB.
15:10 < dgilmore> abadger1999: :D im getting turbo gears book on monday so ill be jumping in more then
15:10 < dgilmore> abadger1999: yeah
15:10 < lmacken> dgilmore: mine is waiting for me up at school :)
15:10 * iWolf is here for a few minutes, overlapping meetings.
15:10 < abadger1999> dgilmore: Cool. I'm going to be playing a bit more with sqlalchemy vs sqlobject and post any questions I have to f-i-l so sopwith can tell me what I'm doing wrong ;-)
15:11 < dgilmore> :D sounds good
15:11 < dgilmore> so we are making some progress
15:11 < abadger1999> yes. If f13 is here I'd like to hear what's going on with brew though.
15:11 < dgilmore> abadger1999, f13: how goes the VCS stuff
15:11 < abadger1999> If brew comes out, it'll be a big lump of stuff that we'll have to figure out how to integrate.
15:12 < abadger1999> VCS is in a holding pattern for me as I work on the packageDB.
15:12 < dgilmore> abadger1999: yeah brew will throw a spanner in the works but give us alot of other functionality also
15:12 < dgilmore> abadger1999: PackageDB is more important now
15:12 < mmcgrath> on the topic of VCS....
15:12 < abadger1999> I think f13 made some good progress/finishing touches but he'll have to speak to that.
15:13 < mmcgrath> BTW guys, as far as I'm concerned cvs-int is back to where it was. If you hear of anything not working right or not syncing or killing someone's dog. Let me know.
15:13 < dgilmore> abadger1999: we will have to get an update from him on the list he said he was in another meeting this arvo
15:13 -!- japj [email@example.com] has joined #fedora-admin
15:13 < dgilmore> mmcgrath: :) seems to be working ok from my end
15:14 < abadger1999> mmcgrath: Did you work out all of f13's hg and git problems?
15:14 < mmcgrath> abadger1999: afaik yes.
15:14 < dgilmore> mmcgrath: any word on getting the old cvs server with an os so we can do db2?
15:14 < mmcgrath> though I still have no idea why the hg IP changed.
15:14 < mmcgrath> Yeah. I'll have to check with stacy and mgalgoci. I think its ready.
15:14 < mmcgrath> I'll be building the dedicated backup server soon.
15:15 < dgilmore> mmcgrath: :D any sign of the new builder hardware?
15:15 < iWolf> mmcgrath: just let me know when it's ready and I can get started on it (hardware for db2)
15:15 < mmcgrath> dgilmore: I thought it was there already.
15:15 < mmcgrath> mdomsch: ?
15:15 < dgilmore> mmcgrath: i have no idea
15:16 < dgilmore> i could be if it is ill get with stacy so i can install and configure it
15:16 < dgilmore> xenbuilder1 is running nicely
15:16 < mmcgrath> solid.
15:16 < dgilmore> i need to take a lvm snapshot of it
15:17 < dgilmore> mmcgrath: want to send an email to stacy and mgalgoci to see what the hardware status is ? or want me to ?
15:17 < abadger1999> dgilmore: Good work!
15:17 < mmcgrath> I'll drop them a note right now.
15:17 < dgilmore> mmcgrath: thanks
15:18 < dgilmore> we should have just the two boxes i think
15:18 < dgilmore> were we going to use xen on the box for DB2?
15:19 < iWolf> dgilmore: not sure what the final decision was. I can if that's what folks want.
15:20 < dgilmore> iWolf: what resouces will db2 need?
15:20 < dgilmore> i think it would be nice to do it in xen
15:20 < teknofile> what VCS are we going to?
15:21 < dgilmore> teknofile: we are looking at quite a few different ones
15:21 < iWolf> dgilmore: The dbs are small now, the largest backup is near 11mb I think...
15:21 < teknofile> ah, still eval
15:21 < dgilmore> teknofile: no decision has been made yet
15:21 < dgilmore> iWolf: the would thrive in xenguests i think
15:21 < abadger1999> teknofile: Yes. And the scope of evaluation is getting larger:
15:22 < dgilmore> with our resouces as they are we need to be extremly flexible
15:22 < iWolf> dgilmore: yeah, I don't think we would lose much by doing it in a xen guest. It just doesn't seem that busy or that intensive.
15:22 < abadger1999> teknofile: Before we were concentrating on just porting our existing functionality over. Now we are also talking about trying new thin
gs (expanded source trees, etc) Not sure how any of it will end up though.
15:23 < dgilmore> iWolf: :D were we going to cluster mysql and postgres in the new setup?
15:23 < teknofile> 'k
15:23 < iWolf> dgilmore: I was planning on getting things to the new stuff and then look at doing that. There was a guy that had some good cluster experience with DBs, not sure if he is still hanging out or not.
15:23 < jcollie> any word on when the new fedora-scm list is going to be set up?
15:24 < dgilmore> jcollie: probably after fedora 7 timeline now
15:24 < dgilmore> we are going to do a merge of core and extras in cvs
15:25 < jcollie> is that hg, git et al can't do partial checkouts?
15:25 < abadger1999> jcollie: There shouldn't be any problem there that I can see.
15:25 < mdomsch> mmcgrath, it was ordered a couple weeks ago, should have shipped
15:25 * mdomsch looks for the order number
15:25 < jcollie> the tree of repos seems to work?
15:26 < dgilmore> jcollie: the hg test seems to be working pretty well
15:26 < jcollie> is the hg repo publically available yet?
15:26 < warren> dgilmore, ack, sorry
15:26 < dgilmore> jcollie: yes it is
15:27 < dgilmore> jcollie: ping f13 for access details
15:27 < dgilmore> warren: :D
15:27 < abadger1999> jcollie: The issue was that some people wanted to try more than a straight port of what we do in cvs. And want to not have to move to a straight port now only to move to the more experimental version later.
15:28 < abadger1999> jcollie: http://www.fedoraproject.org/wiki/Infrastructure/VersionControl/dist-hg
15:28 < jcollie> have some of the new ideas been discussed anywhere?
15:28 < abadger1999> jcollie: One of the steps is talking to f13
15:28 * iWolf heads to his work meeting
15:28 < abadger1999> jcollie: It was discussed at the fedora-summit. But they deferred most of the experimentation to a fedora-scm SIG. which hasn't been formed yet.
15:29 < mmcgrath> mdomsch: thanks.
15:29 -!- glezos [n=glezosd@fedora/glezos] has joined #fedora-admin
15:29 < dgilmore> anyone have anything else to add on any of the systems stuff?
15:30 < mmcgrath> I'd like to encourage some of the officers to become more familiar with cvs-int if they get time.
15:30 < dgilmore> mmcgrath: ive done a little exploring of it this week
15:30 < abadger1999> mmcgrath: I've been poking around on it.
15:30 < dgilmore> its an ugly beast
15:30 < mmcgrath> Its just a good idea considering how farking CRAZY that box is right now. I hope to clean it up quite a bit soon.
15:30 < mmcgrath> the chroots are neat, though with xen we may want to look at a fully virtualized environment for each? who knows.
15:31 < abadger1999> h -- mmcgrath: In cvs-dist there's a git section that points to lockbox.
15:31 < abadger1999> Should that point to cvs-int now?
15:31 < mmcgrath> hmmm, not sure.
15:31 < jcollie> xen virtual servers for everyone!
15:31 * dgilmore hearts xen
15:32 < dgilmore> lmacken: your next.
15:32 < dgilmore> firewalls
15:32 < lmacken> I haven't had time to finish up the pyroman firewall deployment, and probably won't for another couple of weeks. If anyone is bored and wants to help out with it, please feel free.
15:32 < lmacken> It's only really a matter of running it, seeing what breaks, tweaking the pyroman configs, repeat. Everything should be documented in InfrastructurePrivate/Firewalls.
15:32 < lmacken> the updates system has taken top priority for me at the moment
15:33 < dgilmore> lmacken: how is that comming along?
15:33 < lmacken> and while we're ont the topic
15:33 < lmacken> I've been making lots of progress with the new updates system. This is currently my top priority, as it is also going to be used for a new Red Hat product that we are shipping soon.
15:33 < lmacken> I'm hoping to have a prototype out within a couple of weeks.
15:33 < lmacken> I haven't had time to play with the TurboGears deployment on publictest2 yet
15:34 < lmacken> so again, if anyone gets bored.. feel free to play with it :)
15:34 < mdomsch> mmcgrath, it should be there...
15:34 < mdomsch> Status: Delivered
15:34 < mdomsch> Delivered on: 11/20/2006
15:34 < mdomsch> 2:03 P.M.
15:34 < mdomsch> Delivered to: MESA, AZ, US
15:34 < mdomsch> Signed by: RODRIDGUEZ
15:35 < mmcgrath> cool
15:35 < dgilmore> mdomsch: thanks
15:35 < japj> abadger1999: on the dist-hg wiki page there is a link to disthg-utils on http://www.fedoraproject.org/wiki/Infrastructure/VersionControl/dist-hg, but this doesn't seem to work... has this server moved?
15:35 < mdomsch> np
15:35 < dgilmore> legacy buildsys is waiting on a decision from legacy as far as push scripts
15:36 < dgilmore> EPEL buildsys is up and running there will be some work for push scripts. and getting the final stuff in place
15:36 < mmcgrath> :: high five ::
15:36 < f13> ok, I'm back.
15:36 < dgilmore> any of you new guys after things to do there is plenty that can be done for EPEL
15:37 < dgilmore> ping mmcgrath or myself for some jobs
15:37 < abadger1999> japj: Probably -- ask f13, now that he's here :-)
15:37 < teknofile> dgilmore: what is EPEL?
15:37 < dgilmore> teknofile: it is Extra Package for Enterprise Linux
15:38 < mmcgrath> technofile: http://fedoraproject.org/wiki/Extras/Schedule/EnterpriseExtras
15:38 < dgilmore> teknofile: its an effort to have extras packages available for RHEL, CentOS etc
15:39 < teknofile> interesting
15:39 < teknofile> i'll read up on this afternoon
15:39 < dgilmore> mmcgrath: any progress on config management
15:39 < dgilmore> mmcgrath: ive been meaning to check out the finer details of cfengine
15:40 < mmcgrath> I don't think anyone's taken a look at glump yet. I'll send a followup email
15:40 < dgilmore> i have meant to look at it also but havent :(
15:40 < abadger1999> mmcgrath: I took a quick look
15:40 < mmcgrath> what did you think?
15:41 < abadger1999> As you say it is much different than what we currently have.
15:42 < mmcgrath> but it could be used with it, which is interesting.
15:42 < abadger1999> What do we gain from it, though?
15:42 < mmcgrath> Basically enforcement.
15:42 < abadger1999> yes.
15:42 < mmcgrath> if someone changes a file manually, it'll get changed back.
15:42 < dgilmore> mmcgrath: i do like that idea
15:42 < mmcgrath> forcing them to use the cvs system. Its not perfect.
15:42 < mmcgrath> But its better than what we have imho.
15:43 < dgilmore> right now cvs is a little out of date.
15:43 < abadger1999> That part is good. But if that was all we wanted we wouldn't need all the rest of glump.
15:43 * f13 chuckles each time he sees "glump"
15:43 < mmcgrath> dgilmore: thats a whole chunk of this project is getting cvs back up to date.
15:44 * mmcgrath seems to recall a roar of laughter on the phone during the summit.
15:44 < mmcgrath> abadger1999: "the rest of glump" ?
15:45 < abadger1999> If I recall glump constructs a script on a server and then sends it to the client to execute. Is that right?
15:46 < mmcgrath> naw, its a puller. Each client pulls the configs.
15:46 < jcollie> url for glump? google doesn't turn anything up interesting
15:46 < mmcgrath> jcollie: its used internally at Duke, they've recently gpl'd it.
15:47 < jcollie> http://linux.duke.edu/projects/mini/glump/?
15:47 < dgilmore> http://linux.duke.edu/projects/mini/glump/
15:47 < mmcgrath> oh wow, yeah thats it.
15:47 < mmcgrath> I didn't even know that was out there.
15:48 < dgilmore> anyone got anything else to add?
15:48 < dgilmore> new guys got any questions?
15:48 < abadger1999> Okay. pull vs push. But it still is putting together a script that runs on the client?
15:49 < teknofile> dgilmore: not yet; I will once I read up on EPEL tho
15:49 < mmcgrath> Its a webapp, the client just needs to know where the script is, the server sends it a manifest of what files it should get and such.
15:49 < mmcgrath> dgilmore: we should discuss epel in a bit if you have a moment, but its not really fedora-admin related.
15:50 < dgilmore> mmcgrath: sure want to do it once the meetings done?
15:50 < mmcgrath> yeah.
15:51 < abadger1999> It just seems that if we just wanted to make sure we were synced with cvs, we could just have a cron script that checked out the files
from cvs and then checked whether they were different from what's on the system.
15:52 < abadger1999> glump seemed more complex than that and I feel like I'm missing something because there must be a reason why it's done the way it's done.
15:52 < dgilmore> seems like some good work has been going on, in regards to getting the wiki behind the proxies
15:53 < mmcgrath> abadger1999: the main difference is management and backups.
15:53 < mmcgrath> our current system doesn't make any backups.
15:54 < mmcgrath> of overwritten files I mean.
15:54 < mmcgrath> plus, this is something that we can maintain with duke instead of just on our own.
15:54 < dgilmore> mmcgrath: i like the idea of glump alot
15:54 < mmcgrath> dgilmore: before we quit glezos has something he'd like to bring up
15:55 * f13 has some things to report as well.
15:55 < dgilmore> mmcgrath: now is the time
15:55 < glezos> f13, go ahead. I'll wait
15:55 < f13> ok.
15:55 < f13> Project Hosting
15:55 < f13> So, http://hosted.fedoraprojects.org exists, and there are some early testers of our "hosting" services using trac.
15:56 < f13> accounts for trac are through the Fedora account system
15:56 < f13> an admin is chosen at project creation time, they get an 'admin' tab in trac to do fine tuning of their project space
15:56 < jcollie> is that the right url? i can't look that up...
15:56 < f13> We got git and hg repos back up and functional too, but whats missing is making those repos readable by trac for the integrated source view and timeline stuff.
15:56 < abadger1999> remove the 's'
15:56 < f13> whoops
15:57 < f13> hosted.fedoraproject.org
15:57 < f13> I write up a wiki page on how to create a new project space
15:57 < f13> http://fedoraproject.org/wiki/Infrastructure/ProjectHosting/CreateNewProject
15:57 < f13> and an overall page http://fedoraproject.org/wiki/Infrastructure/ProjectHosting
15:57 < f13> I mostly did this because I needed project space for pungi (:
15:58 < f13> the other missing peices are raw webspace for code drops/yum repos/etc.. and mailing lists
15:58 < mmcgrath> :P
15:59 < f13> I think that for version1 of Feodra Hosted Projects, we can do all the setup by hand (creation of source repo, trac project space, raw webspace, and mailing list).
15:59 < f13> that allows us to launch something much sooner rather than waiting until we have a self service or otherwise integrated creation tool.
15:59 < f13> If anybody would like to help me with the few missing parts, I'd really appreciate it, because I probably won't be able to focus on it much more
15:59 < dgilmore> f13: should we put mailman on hosted
15:59 < abadger1999> f13: Does trac just need read-only local access to the repositories?
16:00 < f13> abadger1999: yes, readonly.
16:00 < abadger1999> f13: We could jsut clone the repos.
16:00 < f13> abadger1999: we could yes, things don't update in real time then, but better than nothing.
16:00 < f13> abadger1999: or rather just do scheduled rsyncs since git/hg work like that (:
16:00 < abadger1999> *grin*
16:02 < f13> so...
16:02 < f13> that about does it for Project Hosting. Its not a big priority (when compared to other things regarding core<->extras merger.
16:02 < f13> but somebody with more experience with say mailing list management and webspace management could look into
16:02 < f13> Next is VCS.
16:03 < dgilmore> f13: thanks
16:03 < f13> This took a backseat as well because of the merger. I'll ping greg/max next week regarding creating the SCM SIG where hand wavy future land can be discussed.
16:04 < f13> I'm somewhat convinced though that we'll need to do it in two stages. Stage 1 is current workflow with new SCM of choice, stage 2 is new workflow with said new SCM once new SCM has been beat up on for a bit
16:04 < f13> and new ways of doing things have been explored.
16:04 < jcollie> i'm in favor of whatever gets us off cvs sooner
16:04 < f13> I'll send announcements when the SIG gets launched, probably create a mailing list for it to keep noisy conversation away from places like maintainer's list.
16:05 < f13> I think we may actually want something like git/hg in order to do secondary arches, so that syncing down changes isn't such a huge pain in the ass.
16:05 < f13> We'll see what happens.
16:05 < f13> --Buildsystem--
16:05 < dgilmore> f13: im really looking forward to the secondary arches stuff :D
16:05 < jcollie> is bzr out of the running?
16:06 < dgilmore> jcollie: not as yet
16:06 < f13> Jeremy and I met with RH management regarding brew. It turned more into a why are we opening core and what does this really mean, but we did get to opening of brew a bit. Overall it went well, and we're meeting again on Friday to go over somethings more indepth.
16:06 < f13> so a "no" wasn't given, neither was a yes.
16:06 < mmcgrath> hmm
16:06 < teknofile> dgilmore: will you be avail in about 2hr to talk about EPEL?
16:06 < dgilmore> f13: thats all we can ask for right now i guess
16:07 < f13> jcollie: bzr isn't, but I haven't seen a bzr proof of concept show up yet.. (:
16:07 < dgilmore> teknofile: no, but i will be in about 3-4Hrs
16:07 < f13> dgilmore: I feel good about the meeting progress, and look forward to Friday
16:07 < teknofile> dgilmore: ok
16:07 < mmcgrath> teknofile: I'll be around.
16:07 < teknofile> ok, I'll have read the wiki by then
16:07 < f13> management knows that time is important and it seems they are prepared to make a decision soon, they just want to be fully aware of what decision they are making. Hurray for no snap decisions.
16:07 -!- GeroldKa [n=Gerold@fedora/geroldka] has joined #fedora-admin
16:08 < f13> Thats about all I have for infrastructure stuff.
16:08 < dgilmore> f13: thanks again
16:08 < dgilmore> mmcgrath: your floor
16:09 < mmcgrath> I don't have anything, glezos?
16:09 < glezos> um ye..
16:09 < glezos> yes
16:09 < glezos> I'll keep it short because it's already late
16:09 < glezos> L10N might have some requests in the near future. I guess now it's the best time to bring them to the surface.
16:09 < mmcgrath> k
16:10 < glezos> Some tasks are staring to form on http://fedoraproject.org/wiki/L10N/Tasks -- basically the bottom list is more concrete.
16:10 < mmcgrath> mhmm?
16:10 < glezos> Some of these need some attention from the Infrastructure team (move CVS to a new root on cvs.fp.org) and the Rel Eng team (integrate translation processes/notifications in package updates)\
16:11 < glezos> I have no idea about how things are being done or should be done so I am in no position for suggesting anything.
16:13 < glezos> There aren't a lot of people willing to really work on these issues, so the team might want some help in the tools, the transition and the general "integrate the l10n process *more* into our development/packaging model than we currently do"
16:13 < glezos> eof
16:14 -!- lyz [firstname.lastname@example.org] has quit ["Leaving"]
16:15 < glezos> So... :) If a Q should be made it would be: What do you guys think is more important/urgent and how can we do it?
16:15 -!- mcanann [email@example.com] has quit ["Chatzilla 0.9.75 [Firefox 220.127.116.11/0000000000] "
16:16 < f13> This is a really touchy subject because Red Hat itself is doing a lot of work in this area (or is supposed to) to make it easier for translators
16:17 < dgilmore> glezos: thanks
16:17 < glezos> f13, I'm sure it does. But it is a fact that translators do think we should lower the barrier of entry and increase the quality of the translations in our releases. And I must say, Fedora isn't known for it's good translation processes. :(
16:17 < f13> nod
16:18 < f13> I just want to make sure that Red Hat and Fedora Community don't go off in completely different directions.
16:18 < dgilmore> f13: that would be bad
16:19 < glezos> f13, nod
16:19 < dgilmore> can we get the Red Hat guys on board
16:19 < f13> I would hope so.
16:19 < mmcgrath> hmm
16:20 < mmcgrath> glezos: make sure you send this to the list, we'll probably discuss it more then
16:20 < f13> glezos: I'll point the translation team lead at your wiki page and ask them to work with you, if possible. Is your contact info on those pages?
16:20 < glezos> Some concrete suggestions that have been discusses are to move the pofiles to fp.org to ease the management and remove bottlenecks, and b) put string freezes and translation deadlines on the Core Schedule.
16:20 < f13> B) is easy
16:21 < f13> A)... probably not TOO difficult for the packages RH is the upstream for.
16:21 < f13> however the real thing we keep hearing is that if we want better translations in Fedora, we need to get better translations upstream
16:21 < glezos> f13, then B2 could be something like "raise a flag if a package update changes translation strings" :)
16:21 < f13> its not like Fedora is replacing tons of strings in applications
16:21 < glezos> f13, yes. But I'm not the appropriate person to coordinate these things...
16:23 < glezos> f13, the upstream thing of translations is a very complex one (Ubuntu's rosetta is very, very bad for projects) and more-or-less it could be discussed later on or anyway not in the context of the Infrastructure team.. I guess.
16:23 < dgilmore> im going to close the meeting in 30
16:23 < dgilmore> glezos: please bring it up on the mailing list
16:23 < glezos> dgilmore, I'm finished, no problem.
16:23 < glezos> Will do.
16:23 < glezos> Thanks.
16:24 < dgilmore> 20
16:24 < dgilmore> mark
16:24 < dgilmore> closed meeting thanks guys