16:00:12 #startmeeting Server SIG Weekly Meeting (2015-12-01) 16:00:12 Meeting started Tue Dec 1 16:00:12 2015 UTC. The chair is sgallagh. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:12 Useful Commands: #action #agreed #halp #info #idea #link #topic. 16:00:12 The meeting name has been set to 'server_sig_weekly_meeting_(2015-12-01)' 16:00:12 #meetingname ServerSIG 16:00:12 #chair sgallagh mizmo nirik stefw adamw simo danofsatx mhayden jds2001 16:00:13 #topic roll call 16:00:13 The meeting name has been set to 'serversig' 16:00:13 Current chairs: adamw danofsatx jds2001 mhayden mizmo nirik sgallagh simo stefw 16:00:26 morning 16:00:29 .hello sgallagh 16:00:30 sgallagh: sgallagh 'Stephen Gallagher' 16:01:31 .hello mhayden 16:01:31 mhayden: mhayden 'Major Hayden' 16:03:51 * nirik guesses everyone else is in a turkey coma. 16:05:11 Tryptophan poisoning was the cited excuse in the rolekit meeting this morning 16:05:30 I'll wait a few more minutes to see if people filter in 16:06:21 hah 16:07:11 * jsmith is tied up in other meetings :-( 16:08:38 ahoy 16:08:42 * adamw was making pancakes 16:08:48 top tip: marmalade and nutella 16:09:04 .hello adamwill 16:09:05 adamw: adamwill 'Adam Williamson' 16:09:14 adamw: Sounds delightful, but we have nut allergies in my household :( 16:10:09 sgallagh: damnit, nutella is worth the odd restricted airway. 16:11:26 OK, it's been ten minutes and while we don't really have quorum, I don't think we're debating anything problematic. 16:11:37 #info No quorum, but proceeding anyway 16:11:41 #topic Agenda 16:11:45 I just got here, sorry 16:11:45 #info Agenda Topic: Server/Cloud Interaction 16:11:45 #info Agenda Topic: Trimming the DVD 16:11:53 #info Quorum achieved. 16:12:07 \o/ 16:12:11 ooh, achievement! 16:12:54 i like achieving things 16:13:14 mhayden: What does it feel like? I've long since forgotten 16:13:23 Joking aside: any other agenda items? 16:13:27 it's kind of like drinking 16:13:48 but not nearly as entertaining 16:14:26 OK, moving on 16:14:30 #topic Server/Cloud Interaction 16:14:46 mhayden: jds2001 couldn't make it today, so I guess you're on the hook for status here. 16:14:53 so jds2001 and i are setting up some time to plan this thing out 16:14:58 /me pries the bus off of mhayden 16:15:10 i've had some initial discussions with dusty mabe on this topic and he's all ears 16:15:38 and i think part of the confusion in the user community is around cattle vs. pets in virtualized envieonments 16:15:46 for example, some folks prefer Fedora Server in their VM's 16:15:46 ok 16:15:53 while others prefer Fedora Cloud 16:16:09 and some people are staunchly fighting for one or the other while other folks don't really care strongly 16:16:51 but, seeing that a cloud instance at AWS/RAX/elsewhere is essentially a server, the interaction between both groups should be good 16:17:23 so jds2001 and i will sync up on those touch points and find a way to collaborate and keep the users informed on which workloads might work better on Fedora Server/Cloud 16:17:36 Right, so we should be advertising the differences based on use-case rather than ideology, I think 16:17:43 Right, exactly 16:17:48 full disclosure, i work for rackspace 16:18:05 and our customers typically want a server-ish experience, especially those with dedicated envs 16:18:14 /me grabs his torch and pitchfork :-P 16:18:16 so the idea would be to show when you should use each, rather than trying to converge them into one thing? 16:18:25 * danofsatx grabs the rope 16:18:30 nirik: i think so -- there are good use cases for each 16:18:55 and the cloud folks are doing some interesting work with atomic/containers/docker/rocket/container_tech_of_the_week 16:18:57 yeah. They are pretty similar, but yeah, may not be able to converge them without sacrifices. 16:20:17 So the subtext here is that we should consider whether we're going to produce a full Server image for the various services or if we're going to explore moving the adopt-your-cattle script into cloud-init 16:20:55 hah, possibly 16:21:58 #info General agreement that Cloud and Server should remain individual offerings 16:22:20 #info Marketing focus should be on workloads/use-cases 16:22:46 more to come from jds2001 and i 16:23:40 #action mhayden and jds2001 to continue investigation here 16:23:45 Thanks for the update 16:23:54 #topic Trimming the DVD 16:24:18 While many of us in the US are busy "trimming the tree", we in the Server SIG should consider... 16:24:36 nice segue. 16:24:40 Thanks 16:24:43 ha 16:25:05 Last week we addressed the default installation 16:25:08 related, there's a bunch of old junk in comps... when/if I get time I am planning on cleaning it out. 16:25:33 This week, I think we should address which pieces of the optional bits on the DVD we don't really care about. 16:25:55 how long did you work on that one, sgallagh? 16:26:12 There's a lot of content there that's present because of the old Fedora tradition of "if there's room and someone might need it, stick it on the DVD" 16:26:47 But in reality, the set on the DVD *never* perfectly matches anyone's preferred set of packages. 16:26:54 right 16:27:05 So real-world deployments always have to use either the mirror network or a local mirror to supplement it anway 16:27:34 should the DVD keep getting produced? (just tossing that out there as food for thought) 16:27:40 So my thought is that we should trim a lot of the excess off of the DVD and get it down basically to just the default install plus optional hardware enablement and role support files. 16:28:48 mhayden: That's a fair question. I personally still see some value in having something that can be burned to physical media, but I'm open to arguments that it's a waste. 16:28:51 i think we already went down the 'should we get rid of the DVD' path, didn't we? 16:28:53 mhayden: yes, it should. 16:29:16 In my experience, having the DVD makes it easier to do testing at least, since I don't have to set up a local mirror to produce new VMs 16:29:16 I think there's some cases where it's useful, but I wish we had a way to tell for how many people... ;( 16:29:25 i can see it being valuable for systems that are not on a great internet connection 16:29:48 or on any internet connection (see the many military installations in town for reference) 16:29:48 mhayden: I'm trying to minimize that argument. 16:30:11 Mostly because it still needs to be on a decent intranet connection at least to do anything truly useful 16:30:21 my use for the DVD in the past has been to explode it into a directory structure for PXE installs 16:30:32 danofsatx: I think those are the cases where a local mirror should be assumed 16:30:52 mhayden: Incidentally, would you mind writing up a HOWTO blog or something on that 16:31:00 That would be great documentation to have. 16:31:01 agreed, but not always. I had to build my own in order to get one. 16:31:24 sgallagh: I think it's still in the install guide? or did it get dropped? 16:31:29 danofsatx: I didn't mean assumed to already be present. Just assumed that such a thing was necessary in that environment 16:31:42 nirik: Oh, is it? I guess I missed that section. 16:32:02 https://docs.fedoraproject.org/en-US/Fedora/23/html-single/Installation_Guide/index.html#pxe-overview 16:32:19 You're right. Mea culpa. 16:32:26 I should really read our docs one of these days. 16:33:27 OK, so https://git.fedorahosted.org/cgit/spin-kickstarts.git/tree/fedora-install-server.ks includes the set of package groups we put on the DVD. 16:34:00 Actually, starts at https://git.fedorahosted.org/cgit/spin-kickstarts.git/tree/fedora-install-server.ks#n87 16:35:32 My proposal is to basically drop everything from line 87 down except for @sql-server, @freeipa-server, @smb-server and @virtualization-headless 16:36:12 Oh, and @web-server 16:37:04 Thoughts? Arguments? Rotten tomatoes? 16:37:10 we'll want to keep ansible-node. 16:37:57 adamw: That's going to become a dep of the fedora-release-server package 16:38:04 * nirik is ok with that... we will want to make sure to get feedback from people in the alpha/beta cycles about it. 16:38:04 Or, no it's not 16:38:05 Sorry 16:38:29 adamw: Yeah, I'll make sure that's present explicitly. 16:39:27 nirik: Do you know at what point we'll start getting test composes, BTW? 16:39:51 well, we have rawhide now... if dracut would stop breaking everything 16:40:00 sgallagh: we usually start at branch point. 16:40:19 nirik: Right, but we're not producing nightly artifacts except the boot.iso, right? 16:40:23 which is feb 2nd. 16:40:25 but there's still plans to get daily full composes, just not sure how far along that is. 16:40:34 ok 16:40:39 boot iso and live and appliances. 16:40:52 but hopefully we can add trees and dvd images and everything else. 16:41:17 danofsatx, mhayden: any opinion on this plan? 16:41:48 If we're generally in favor of this, I'll get the spin-kickstart changes in today. 16:42:06 * adamw is fine with it. 16:42:14 * nirik is ok with it 16:42:14 i'm okay with it 16:42:20 i'm curious to see how much the disk usage changes 16:43:38 mhayden: Yeah, I'm not really sure. I suspect that the lion's share of the usage is still held by FreeIPA and its dependencies. 16:43:46 (Particularly Dogtag) 16:43:50 oh yeah 16:43:55 there be dragons ;) 16:44:09 Dogtag at least is working on eliminating the full java requirement and getting back to java-headless 16:44:12 Which should help some 16:44:25 i unfortunately need to scoot 16:44:30 but i'm okay with the change proposed 16:44:51 OK, then I'll go ahead and have at it, I think 16:45:43 Of course, the flip side of the FreeIPA issue is that we *may* in Fedora 24 be converting over to using FreeIPA as a container instead of RPMs. So we might be able to get that off the disk too. 16:45:47 More on that at a future meeting. 16:47:16 #action sgallagh to remove most of the package groups from the Server DVD except for the web-server, FreeIPA, Samba, Virtualization, Ansible bits and hardware enablement. 16:47:53 I'm also hopeful that FreeIPA will finish shedding its Python 2 deps this cycle, but that might stick around until F25. 16:48:33 #topic Open Floor 16:48:42 Any other topics for the remaining 12 minutes? 16:48:43 +1 for plan 16:48:50 * danofsatx late to the party again 16:48:57 * nirik has no items 16:50:21 OK, thanks for coming, folks. 16:50:32 I appreciate your Service. 16:51:02 #endmeeting