Infrastructure/Meetings/2006-11-09

= Meeting of 2006-11-09 =


 * Time shown in EST

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