From Fedora Project Wiki
< Infrastructure | Meetings
Meeting of 2006-12-07
*** Time shown in EST 15:01 < mmcgrath> We ready? 15:01 < mmcgrath> Who all is here? 15:03 * teknofile is sorta here 15:03 -!- kschreyack [n=kschreya@63.202.114.66] has joined #fedora-admin 15:03 < iWolf> here 15:03 < mmcgrath> For the new guys we follow the Schedule on http://fedoraproject.org/wiki/Infrastructure/Schedule 15:03 < mmcgrath> I'll just get started, this may be short. 15:04 < iWolf> :) 15:04 < mmcgrath> There's been plenty of discussion on the packaging database on the list this week. 15:04 < mmcgrath> abadger1999: anything to add right now? 15:04 < abadger1999> Nope. It's all on the list :-) 15:05 < mmcgrath> How about with mercurial or any of the vcs stuff? 15:05 < mmcgrath> f13: anything? 15:06 < mmcgrath> pass :-D 15:06 < teknofile> guess not 15:06 < mmcgrath> iWolf: are you still waiting on me for the new db box? 15:06 < mmcgrath> What all do we want to do with it? 15:06 < mmcgrath> in general I still don't trust it until its proven itself again. 15:07 < iWolf> mmcgrath: oh yeah, we're using the old cvs for it now? 15:07 < mmcgrath> Yeah, its just sitting there. 15:07 < mmcgrath> I figure if nothing else it'd make a good backup db server. Either way are we going the xen route with db as well? 15:07 * mmcgrath votes +1 15:07 < iWolf> mmcgrath: I am fine with Xen for it. 15:08 < mmcgrath> cool. Live migration would be very nice for the db server if we get to that point. 15:08 < iWolf> so should I put the Xen guest on the old cvs hardware or on one of the xen servers? 15:08 < mmcgrath> I'd say turn the old cvs hardware into a xen box and use it. 15:08 * dgilmore is here 15:08 < mmcgrath> dgilmore: yo 15:08 < iWolf> mmcgrath: okay. 15:09 < mmcgrath> I'm not even sure if it got hooked back up and is live or not. 15:09 < iWolf> mmcgrath: I just wrapped up a big project at work last weekend, so I hope to have some more time to dig into it again. 15:09 < iWolf> mmcgrath: did Dell come back out and look at it? 15:09 < mmcgrath> Either way what ever is left on it can be blown away. IIRC it was a total loss. 15:09 < mmcgrath> yeah, they replaced the backplane. 15:09 < mmcgrath> and the bad drives. 15:10 < iWolf> mmcgrath: should I check with Stacy to see if it was hooked back up? 15:10 < mmcgrath> Check with Stacy or mgalgoci to see if it got hooked back up. 15:10 < iWolf> mmcgrath: will do. 15:10 -!- [japj] [n=japj@japj.xs4all.nl] has joined #fedora-admin 15:10 < mmcgrath> lmacken: any firewall stuff to report? 15:11 * mmcgrath is not sure if lmacken is here. 15:11 < teknofile> mmcgrath: his e-mail to the list said he wouldn't be able to make any infra mtgs for 10 weeks 15:11 < mmcgrath> thats right, well either way he's close to done with most of that. 15:12 < mmcgrath> warren's on the xen list right now but one thing we're waiting on is the smtp server, dgilmore where's that sit? 15:13 < warren> does anyone understand how to configure the sendmail on the other boxes? 15:13 < warren> we could start by using it as outgoing SMTP on one box, make sure that works, then expand from there. 15:13 < mmcgrath> Yeah, we just need to point them to it instead of bastion. 15:13 < teknofile> sendmail or postfix? the Schedule talked about using postfix 15:13 < warren> smtp.fedora is postfix 15:13 < iWolf> warren: probably a good early test. 15:13 < mmcgrath> the other boxes are running sendmail, our primary smtp server is running postfix 15:14 < warren> Does anyone understand everything the sendmail on the other boxes is doing? 15:14 < warren> If so, we could convert those too 15:14 < warren> but let's convert them after smtp goes live and is confirmed working 15:14 < mmcgrath> They're just using bastion as the smart host for relay I think. 15:14 < warren> that's it? 15:14 < iWolf> mmcgrath: I believe that is all they do. 15:14 < teknofile> that's pretty easy to convert to postfix then 15:14 < mmcgrath> worksforme. Any volunteers alter the current configs to point to smtp instead of bastion? 15:15 < warren> which is the lowest risk host(s) to test first? 15:15 < mmcgrath> the proxy servers or the hammer servers. 15:15 < iWolf> probably something we have more than one of. 15:15 < iWolf> one of the proxies would be a good choice I think. 15:16 < warren> we also need to edit the scripts to generate postfix-suitable mail aliases 15:16 < warren> Hmm... I'll try to do this before next Thursday. 15:17 < mmcgrath> awesome, thanks warren. 15:17 < warren> I might reinstall smpt.fedora for good measure 15:17 < mmcgrath> You can do that if you want. I'm not even sure we ever got it totally working to begin with. 15:17 < iWolf> it was sending mail last week. 15:17 < mmcgrath> dgilmore: so the legacy builders, do we still need them or is legacy going away. 15:17 < iWolf> I sent a few test message through it. 15:18 < iWolf> the greylisting was not on there, but it looked like amavis and clam were. 15:18 < mmcgrath> I don't think it was recieving mail correctly (at least as far as otrs goes) 15:18 < mmcgrath> does anyone know what the news with legacy is? 15:18 < mmcgrath> f13: ping? 15:18 < iWolf> mmcgrath: there was a firewall rule that was stopping some of it. 15:18 < dgilmore> mmcgrath: no idea 15:18 < iWolf> I added the amavis port and the second postfix instance and it started sending. 15:18 < mmcgrath> ah 15:19 < mmcgrath> We'll skip legacy for now 15:19 < mmcgrath> Config Management: I see someone has packaged glump, it needs review. 15:20 < mmcgrath> Has anyone else had a chance to look at it? For or against? 15:20 < iWolf> I have looked at it briefly. 15:20 < warren> iWolf, ah ok 15:20 < warren> iWolf, I'll try to add greylisting then. 15:21 < mmcgrath> warren: cool 15:21 < dgilmore> mmcgrath: i havent looked at glump yet 15:21 < mmcgrath> iWolf: what do you think? 15:21 < dgilmore> but i want to look at it and compare it to cfengine 15:21 < abadger1999> mmcgrath: What do you think about adding something to glump to get configs from the client back to the server? 15:21 < iWolf> I think it could work well for us. The "enforcement" nature is probably something we need. 15:21 < warren> perhaps implement glump to manage one host, and see how people like it? 15:21 < warren> hands on is a great way 15:22 < mmcgrath> dgilmore: <nod> 15:23 < abadger1999> mmcgrath: Would it be easy or pretty orthogonal to what it does already? 15:23 < mmcgrath> warren: agreed 15:23 < mmcgrath> abadger1999: to implement it on one host you mean? 15:23 < warren> glump for smtp maybe? 15:23 < abadger1999> No, passing configs back 15:24 < warren> hmm... no better to use glump on something that we *know* works first =) 15:24 < mmcgrath> warren: heh 15:24 < abadger1999> warren: *grin* 15:24 -!- GeroldKa [n=Gerold@fedora/geroldka] has joined #fedora-admin 15:24 < warren> just pick a host and implement glump on it? 15:25 < mmcgrath> Yeah, we could look close at cfengine too. Who here has experience with it? 15:25 < mmcgrath> dgilmore? 15:25 < GeroldKa> hi all 15:25 < dgilmore> mmcgrath: i have not used it. 15:25 < mmcgrath> hello 15:25 < [japj] > hi 15:25 < dgilmore> but i figure nows a good time to learn 15:25 < abadger1999> mmcgrath: I ask because glump's enforcement is a stick to make people do the right thing. Would be good to think about a carrot as well (making it easy to test a config and then send it back to the server.) 15:25 < mmcgrath> Yeah, we should take a look. 15:26 < iWolf> abadger1999: that would be a nice feature as well... 15:26 < dgilmore> abadger1999: make your change test it and scp it accross 15:26 < mmcgrath> When you say send it back you mean pulling the client config from the servers? 15:27 < abadger1999> dgilmore: True. But then you have to remember paths, and etc. And what if we ant to do version control on theserver side? 15:27 < dgilmore> abadger1999: sure 15:27 < mmcgrath> Should we continue this on the list? 15:27 < abadger1999> Maybe to write that I should come up with a list of nice things and then we can write a script to handle it. 15:27 < warren> mgalgoci tells me that they're pretty happy with cfengine 15:27 < iWolf> mmcgrath: +1 15:27 -!- japj [n=japj@japj.xs4all.nl] has quit [Connection timed out] 15:27 < abadger1999> mmcgrath: +1 15:27 -!- [japj] is now known as japj 15:28 < mmcgrath> abadger1999: want to start a 'requirements' type email with all of the 'nice things' you come up with ;-) ? 15:29 < abadger1999> mmcgrath: I'll do that and stick it on the wiki too. 15:29 < mmcgrath> Awesome. 15:29 < mmcgrath> Ok, next item is metrics. 15:29 < mmcgrath> This round of metrics went fairly well. We're bout to hit 600,000 users 15:30 < iWolf> nice! 15:30 < dgilmore> :D 15:30 < mmcgrath> We need to figure out what we're going to do for FC7. I'll re-start that conversation on the FAB 15:30 < mmcgrath> https://admin.fedoraproject.org/cacti/graph.php?rra_id=all&local_graph_id=59 <- Graph 15:30 < mmcgrath> for those that don't know 15:30 < mmcgrath> https://admin.fedoraproject.org/cacti/graph.php?rra_id=all&local_graph_id=60 <- rawhide 15:30 -!- jcollie [n=jcollie@dsl-ppp239.isunet.net] has joined #fedora-admin 15:30 < mmcgrath> Anywho, more on that later. 15:31 < mmcgrath> postfix we talked about 15:31 < mmcgrath> hardware reporting tool is kind of in limbo at the moment. 15:31 < mmcgrath> lmacken's not here to discuss the updates system 15:31 < mmcgrath> The caching proxies have come a long way but paulo and kim0 aren't here. 15:31 < mmcgrath> Basically we've updated Moin to the newest version, its about ready for testing. 15:32 < mmcgrath> and f13's still working on "Project Hosting" which could turn into a huge thing if its popular. 15:32 < mmcgrath> So thats all We've got for the priority 1 and 2 stuff. 15:32 < mmcgrath> Anyone have anything they'd like to add? 15:32 < japj> any word on brew? 15:33 < warren> japj, progress on convincing people, we're still waiting 15:34 < mmcgrath> Anyone have anything else? I'll end the meeting in 20 15:35 < mmcgrath> On a side note there is an Infrastructure position available as was announced on the Fedora Announce list! Anyoneinterested is welcome to apply! 15:35 < mmcgrath> Ok, MEETING END ---------------