Infrastructure/Meetings/2006-11-09

From FedoraProject

< Infrastructure | Meetings
Revision as of 14:13, 24 May 2008 by ImportUser (Talk)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Infrastructure > Meetings -> 2006-11-09


Meeting of 2006-11-09


*** Time shown in EST

15:02 < mmcgrath> Alrighty guys, ready to get started?  Who's here?
15:02 < f13> mmcgrath: I am
15:02 < lyz> hi
15:02 < jwatson> <-- here
15:02 < iWolf> I'm here.
15:02 < mmcgrath> Hello everyone
15:02 < mmcgrath> dgilmore: around?
15:02 < mmcgrath> abadger1999: ?
15:03 < mmcgrath> we'll get to it.
15:03 < abadger1999> here now.
15:03 < mmcgrath> howdy!
15:03 < mmcgrath> has any progress been made on the Package database?
15:04 < abadger1999> No.  I've been trying to work on the bzr vcs stuff and keep uptodate with dist-hg
15:04 < mmcgrath> Awesome, thats the priority right now anyway
15:04 < mmcgrath> f13, abadger1999: what do you two have to report?
15:04 < f13> nothing much new on dist-hg, other than its sitting there, ready for people to test.
15:05 < f13> Ray Strode from Red Hat asked for an account, I don't know if he did anything with it.
15:05  * paulobanon is sort of around
15:05 < f13> next week I have the cabal summit, and after that I"ll work on dist-git
15:05 < f13> this week and been all pungi all the time.
15:05 < mmcgrath> hurray pungi
15:05 < abadger1999> f13: Is dist-hg hooked into a chroot with access granted by fedora-accounts?
15:06 < f13> abadger1999: no.  accounts are local ot the box, and there is one shared plague account/cert
15:06 < f13> its very standalone
15:06 < abadger1999> I've been working out how that is done for cvs and applying it to bzr.
15:06 < f13> fedora-accounts just sends out ssh keys everywhere right?
15:06 < abadger1999> Pretty much.
15:07 < f13> k, thats how I'm using it, ssh keys
15:07 < mmcgrath> f13: yeah, it also properly distributes groups.
15:07 < f13> and using the 'fedora' group rather than say 'cvsextras'
15:07 < abadger1999> The only twist is creating a second set of ssh keys/passwd file for the chroot.
15:07 < f13> mmcgrath: nod, I figured it was a gropu and sshkey thing.
15:07 < f13> 'for the chroot' ?
15:08 < abadger1999> We put cvs, git, and mercurial on the cvs server into chroots.
15:08 < f13> ok...
15:09 < mmcgrath> I have a 'testgroup' if you want to use it sometime
15:09 < f13> abadger1999: do you have more details forthcoming?
15:09 < abadger1999> I'm still applying it to bzr and seeing what I break.
15:10 < abadger1999> f13: How about I write it up on the wiki after I get it working?
15:10 < f13> I suppose, I'm not seeing where this second set of keys plays in
15:11 < abadger1999> One set is in the main filesystem for admins to log into the box.  The second set is in the chroot for people who access the repository.
15:11 < mmcgrath> Ok, we'll go back to VCS later if thats ok owith you guys.  There's not to much else to talk about.
15:11 < mmcgrath> ?
15:11 < f13> mmcgrath: sure.
15:11 < abadger1999> Fine here.  f13 can catch me after the meeting.
15:12 < mmcgrath> k, iWolf: I'm going to contact Stacy soon about switching out proxy[3-4]  with our new dells.
15:12 < mmcgrath> You'll have your db server in no time ;-)
15:12 < iWolf> mmcgrath: sounds good!
15:12 < mmcgrath> lmacken can't make it today.
15:12 < mmcgrath> dgilmore: is the legacy stuff done as far as you know?
15:12 < mmcgrath> f13: I guess you could answer that too?
15:13 < f13> yes, the holdup is now me, and an update tool
15:13 < mmcgrath> k.
15:13 < f13> the builders can build legacy stuff, it's a matter of where to push them, so push scripts.
15:13 < mmcgrath> <nod>
15:14 < mmcgrath> So the config management stuff...
15:15 < paulobanon> gimme 5 10mins and i'll update on the proxystuff
15:15 < mmcgrath> paulobanon: k.
15:15 < mmcgrath> So I've had a chance to evaluate glump (the config management stuff that Duke uses)
15:15 < mmcgrath> Its pretty modular in design so its easy to add stuff too.
15:16 < mmcgrath> Its basically a couple of scripts, you run it regularly and pipe it through bash.
15:16 < mmcgrath> I'll send more info to the list so others can try it.
15:16 < iWolf> mmcgrath: So it looks promising?
15:16 < mmcgrath> I've had a few people contact me and who are working on the new mirror system and a few other things.
15:16 < mmcgrath> iWolf: yeah it does.
15:17 < mmcgrath> There's not much it couldn't do.
15:17 < mmcgrath> Its pretty simple in design which is good but there is a slight learning curve.
15:17 < iWolf> mmcgrath: cool, look forward to the post to the list to see more!
15:18 < mmcgrath> Just real quick.  Could everyone here verify they can access the wiki?
15:19 < mmcgrath> Other than that paulobanon and kim0 have been working on the whole proxy server caching the wiki + wiki update from 1.3 to 1.5 thing.
15:19 < mmcgrath> paulobanon: back yet?
15:19 < iWolf> mmcgrath: the main page came up for me, links to marketing and infra are timing out.
15:19 < f13> mmcgrath: wiki responds to me
15:20 < abadger1999> I can get it but it seems slow.
15:20 < mmcgrath> hmm
15:20 < abadger1999> Not sure if it's me or the wiki.
15:20 < f13> images aren't loading
15:20 < f13> oh there it goes, just very slow
15:20 < lyz> same here
15:20 < skvidal> strange wiki?
15:21 < mmcgrath> skvidal: i just had a timeout myself, followed by a refresh that came right up
15:21 < skvidal>  15:20:59 up 16 days,  7:46,  1 user,  load average: 1.15, 1.11, 1.18 on fpserv
15:21 < mmcgrath> Ok, so aside from the VCS and stuff paulo will get to in a few minutes.  Anyone have anything else to discuss?
15:22 < abadger1999> I have a different VCS thing.
15:22 < mmcgrath> abadger1999: ok?
15:22 < abadger1999> John Kraal has been contacting me about wanting to test darcs and svk.
15:22 < mmcgrath> skvidal: time wget http://fedoraproject.org/awstats/stats/
15:22 < mmcgrath> real    0m21.137s
15:22 < mmcgrath> :(
15:23 < skvidal> the machine has no load to speak of
15:23 < f13> abadger1999: cool, more options.
15:23 -!- craigt [n=cthomas@static-68-236-219-35.nwrk.east.verizon.net]  has quit ["Leaving"] 
15:23 < mmcgrath> abadger1999: who's John Kraal?
15:23 < abadger1999> geonetix?
15:24 < abadger1999> He's one of the people that answered our last request for volunteers.
15:24 < mmcgrath> got'cha.
15:24 < mmcgrath> Has he added his thoughts to the wiki?
15:24 < abadger1999> not yet.
15:24 < f13> he should probably do that, and do some local testing before doing anything w/ a xen box
15:24 < mmcgrath> if he has a solid test plan and he knows those solutions don't have blockers I'd say go ahead.
15:24 < abadger1999> Sounds good.
15:24 < mmcgrath> f13: yeah.
15:25 < abadger1999> Do we want to consider darcs at all?
15:25 < abadger1999> Because of patch commutation?
15:25 < mmcgrath> the first time I had ever heard of dracs was earlier today...
15:25 < mmcgrath> [14:22:41]  <abadger1999> John Kraal has been contacting me about wanting to test darcs and svk.   <---- Then
15:25 < abadger1999> mmcgrath ;-)
15:25 < mmcgrath> :)
15:26 < abadger1999> darcs allows you to take out old patches, reorder the order they're applied in and other neat things.
15:26 < f13> so does git
15:26 < mmcgrath> skvidal: I don't know if its the machine or not.  I'm having repeated long connection times to port 80.
15:26 < f13> git has a patchstack for that
15:26 < abadger1999> But that also means you don't have a concrete history like any other VCS
15:26 < f13>  and hg has an Mq for that.
15:26 < skvidal> it might be a network issue upstream
15:26 < mmcgrath> could be.
15:26 < skvidal> but I've got nothing here that shows anything
15:26 < skvidal> and connections from my machine at home work fine
15:27 < abadger1999> f13: I haven't used darcs yet, but my impression is it's unique to darcs.
15:27 < abadger1999> It really allows you to change the history of a file.
15:27 < mmcgrath> time echo | nc fpserv.fedoraproject.org 80^Pr^Pn^Pr^Pnreal    0m21.090s
15:27 < mmcgrath> Strange, almost exactly 21 seconds every time it hangs.
15:27 < f13> abadger1999: I don't know how useful that would actually be for a package cvs
15:27  * skvidal smells transparent proxy
15:28 < abadger1999> f13: I think it would be a liability.
15:28 < skvidal> if I  turn off the iptables does it get better?
15:28 < skvidal> try now
15:28 < iWolf> skvidal: working better for me now....
15:29 < iWolf> much snappier
15:29 < jwatson> real    0m0.366s
15:29 < abadger1999> I think part of our use of revision control is to be able to go back and pull a specific revision.  Darcs apparently makes it so you can't always get that particular revision back out.
15:29 < skvidal> then it is the multiple connections controls
15:29 < abadger1999> (in terms of what the files were at that time.)
15:30 < skvidal> iWolf: okay, try now
15:31 < iWolf> skvidal: still working well.
15:31 < skvidal> not sure that did anything, but okay
15:31 < mmcgrath> skvidal: I'm more responsive as well
15:31 < mmcgrath> And can get there from home again
15:31 < skvidal> I disabled the connection throttling
15:31 < mmcgrath> paulobanon: back?
15:32 < skvidal> but that only affects  port 80
15:32 < jwatson> time lynx -dump fpserv.fedoraproject.org  == real    0m0.898s
15:32 < skvidal> not 22
15:32 < abadger1999> Anyhow, I can tell him to go ahead and write up svk and I'll do a little more exploration of darcs before telling him that's going to be okay as well.
15:32 < f13> abadger1999: that sounds less than useful of a SCM no matter what the project.
15:32 < mdomsch> dgilmore, -devel buildroots not working quite right
15:32 < mdomsch> http://fedoraproject.org/buildgroups/development/i386/repodata/primary.xml.gz: [Errno 12]  Timeout: <urlopen error timed out> Trying other mirror.
15:32 < mdomsch> Error: failure: repodata/primary.xml.gz from groups: [Errno 256]  No more mirrors to try.
15:33  * paulobanon is back now
15:33 < paulobanon> i asked 5 10mins to be here
15:33 < mmcgrath> ok, paulobanon: give us an update
15:33 < mmcgrath> ;-)
15:33 < paulobanon> ok kim0 and i are playing with webtest.fedora.redhat.com
15:34 < paulobanon> basicly we are configuring apache + squid on the proxies
15:34 < abadger1999> f13: It seems to make managing changes between multiple developers easier... which is one of the aspects of an SCM, but not the one I value the most.
15:34 < paulobanon> right now, apache is setup and squid is kind of ready on proxy1
15:34 < paulobanon> just need to deploy it to the other proxies
15:35 < mmcgrath> paulobanon: there's acopy of the current wiki in your home dir.
15:35 < paulobanon> mmcgrath: app1 ?
15:35 < mmcgrath> yep
15:35 < paulobanon> :d nice
15:35 < paulobanon> im testing something we used back at VFUK
15:35 < abadger1999> f13: He'd also like to get some data to start testing with.
15:35 < abadger1999> f13: How did you populate dist-hg?
15:36 < f13> abadger1999: its on my wiki page
15:36 < paulobanon> a 'switch' that we can use to change the proxies from 'normal' to static or unavailable
15:36 < f13> under the Converstion section
15:36 < mmcgrath> paulobanon: awesome.  Good work.  That'll be a big help to our infrastructure.
15:36 < paulobanon> basicly you just need to  touch a file in a particular dir
15:36 < mmcgrath> In the interest of a short meeting, does anyone have anything else besides the VCS?
15:36 < mmcgrath> if not we'll get back to it.
15:36 < jwatson> I've been working on getting an fc3 installation done so that I can fix the extras nagios-plugins build on fc3
15:37 < abadger1999> It says you pulled a copy from backup.
15:37 < jwatson> does fc3 work well as a xen guest?
15:37 < mmcgrath> jwatson: thanks for that by the way.  You could also use CentOS 4.
15:37 < mmcgrath> jwatson: not without some tweaking.
15:37 < jwatson> how about centos?
15:37 < abadger1999> I'm afraid I'm too new to know where to find that.
15:37 < mmcgrath> CentOS4 would also work since thats our real target.
15:37 < mmcgrath> RHEL4/CentOS4.
15:38 < jwatson> would a centos4 install inside a chroot with an FC6 kernel be ok for building?
15:39 < jwatson> now that I say that, it it sounds like a bad idea  :)
15:39 < mmcgrath> jwatson: for building?  Should be.  We'll just be using mock.
15:39 < jwatson> Ok, I'll figure out one or the other
15:40 < mmcgrath> awesome
15:40 < mmcgrath> Anyone have anything else?
15:40 < f13> abadger1999: ah, I got a 'copy from backup' from mmcgrath
15:41 < paulobanon> later today i'll have webtest.f.rh.c/hitbox going to fp.o like apache->squid->fpserv
15:41 < mmcgrath> paulobanon: awesome.
15:41 < abadger1999> f13: k.  thx
15:41 < paulobanon> only working on proxy1 atm
15:42 < mmcgrath> If no one has anything else.  f13 abadger1999: do you want your VCS discussion on record or after the meeting?
15:42 < abadger1999> mmcgrath: How do I get a copy of the cvs tree for importing?
15:42 < f13> mmcgrath: I have no preference
15:42 < paulobanon> (im out again, i'll read the meeting log later)
15:42 < abadger1999> mmcgrath: If it's on record, I can point John Kraal to it.
15:42 < mmcgrath> abadger1999: send me an email of what you want and where and I'll just make it happen ;-)
15:43 < abadger1999> mmcgrath: Cool.
15:43 < mmcgrath> k, you guys continue.  Let me know when your done and I'll call th emeeting ;-)
15:43 < mmcgrath> or you guys can call it.
15:45 < quaid> mmcgrath: are we going to be able to resolve the (AIUI) Python version for Plone on EL 4?
15:46 < mmcgrath> quaid: yeah.  That shouldn't be a problem.  WE've discussed a couple of options but we're pretty sure RHEL5 wil be out by the time plone is ready ;-)
15:46 < warren> Oops, did meeting begin at 3?
15:46 < mmcgrath> warren: yeah.
15:47 < abadger1999> f13: Right.  So the chroot.  The current cvs server binds: /usr/sbin/chroot [REPO]  /usr/sbin/sshd -i to the various public hostnames.
15:47 < warren> sorry, I thought we adjusted it with the timezone shift like FESCO
15:47 < mmcgrath> ;-)
15:47  * warren tugged in too many directions
15:47 < abadger1999> One for git, one for cvs
15:48 < warren> BTW, need any further network changes?
15:49 < warren> I think test[1-9]  and publictest[1-9]  should provide sufficient flexibility, as long as we are adequately protecting port 22 on publictest.
15:49 < warren> I saw some logwatch reports where people began attacking port 22 on publictest
15:49 < mmcgrath> we should enforce iptables on the xen dom0 hosts
15:50 -!- craigt [n=cthomas@static-68-236-219-35.nwrk.east.verizon.net]  has joined #fedora-admin
15:51 < f13> abadger1999: ok...
15:54 < abadger1999> [REPO]  is the chroot and has a passwd file, the ssh keys of everyone allowed to access that service, the programs necessary for accessing it, and the repo data itself.
15:55 < abadger1999> f13: It should be pretty easy for you to integrate the scripts that are there with hg-dist as we're already hosting an hg server there.
15:57 -!- glezos [n=glezos@fedora/glezos]  has joined #fedora-admin
15:58 < f13> but that hg server is more for source projects, not necessarily package dist right?
15:58 < f13> or does it matter?
15:59 < f13> Actually I'm failing to see what I would need to change in my setup, other than perhaps paths
15:59 < mmcgrath> alright, this is going to go on forever.  Mind if we call the meeting?
16:00 < mmcgrath> or not its up to you.   Ijust want to make sure the logs make it onto the wiki whenever it does end.
16:00 < glezos> hi all.
16:00 < abadger1999> mmcgrath: Go ahead.
16:01 < mmcgrath> k
16:01 < mmcgrath> --------------- Meeting End ----------------