15:00:01 #startmeeting Server SIG Weekly Meeting (2015-09-01) 15:00:01 Meeting started Tue Sep 1 15:00:01 2015 UTC. The chair is sgallagh. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:01 Useful Commands: #action #agreed #halp #info #idea #link #topic. 15:00:01 #meetingname ServerSIG 15:00:01 The meeting name has been set to 'serversig' 15:00:01 #chair sgallagh mizmo nirik stefw adamw simo tuanta mitr danofsatx 15:00:01 #topic roll call 15:00:02 Current chairs: adamw danofsatx mitr mizmo nirik sgallagh simo stefw tuanta 15:00:05 .hello sgallagh 15:00:06 sgallagh: sgallagh 'Stephen Gallagher' 15:00:10 ahoy to the oy hoy 15:00:12 .hello dmossor 15:00:14 morning. 15:00:16 danofsatx: dmossor 'Dan Mossor' 15:00:32 .hello duffy 15:00:33 mizmo: duffy 'Máirín Duffy' 15:00:42 * danofsatx is having a split personality day 15:01:20 /me used to think he was schizophrenic. Turns out I was just imagining things... 15:02:16 .hello stefw 15:02:18 stefw: stefw 'Stef Walter' 15:02:48 I'm dividing attention between moving data, setting up an IPA replica, researching Cyber Securty curriculum for a new degree program at my uni, and kinda-sort paying attention to the LAS podcast. 15:02:53 oh yeah, and this meeting. 15:03:09 .hello simo 15:03:09 simo: simo 'Simo Sorce' 15:03:57 OK, let's get started. Agenda: 15:04:05 #topic Agenda 15:04:05 #info Agenda Item: Blocking Media List 15:04:05 #info Agenda Item: Default NTP service 15:04:05 #info Agenda Item: Websites Update 15:04:28 Does anyone else have a topic to add to the agenda? 15:05:13 nothing here....lets go. 15:05:23 #topic Blocking Media List 15:05:48 I sent out a proposed response 15:05:52 #link https://lists.fedoraproject.org/pipermail/server/2015-September/002023.html 15:06:06 * danofsatx saw it 15:06:19 +1 to proposed list, seems fine to me. 15:06:46 oh yeah, +1. 15:06:56 +1 from me as well 15:07:03 +1 15:07:12 The only piece that I thought we might want to discuss is the ARM image. I could see an argument for blocking on it... or not. 15:07:36 I don't have a problem with it blocking 15:07:44 sgallagh: that's arm64 right ? 15:07:59 Well, hardware with which to test it is not readily available 15:08:01 simo: No, that's 32-bit ARM 15:08:06 ah 15:08:12 simo: armhfp is 32-bit 15:08:24 why would we block in server for arm32 when we are droppin intel32 ? 15:08:26 i'd at least want someone to be committed to *using* it for us to block on it. 15:08:38 right 15:08:42 adamw: I am using it, and will be using it :-) 15:08:51 * adamw dumps all the work on jsmith 15:08:52 +1! 15:08:53 simo: ARM actually still has active kernel support. i686 doesn't 15:08:55 adamw: But I'm not sure that I'm enough of a use-case to justify blocking :-) 15:09:14 simo: 32-bit ARM is still the primary case for now. it's hard and very expensive to buy 64-bit ARM hardware. 15:09:14 jsmith: If you're willing to commit to testing it for release validation... 15:09:18 simo: There are enough active contributors in the ARM SIG that work gets done :-) 15:09:24 sgallagh: Sure... I can do that 15:09:25 of course, you can't run much of a server on a 32-bit ARM box, really. 15:09:26 sgallagh: sure, but we have very little if no userspace testing for the servers bits 15:09:51 sure, it's good for small stuff like dns/dhcp, etc 15:09:53 testing freeipa is enough of a pain without also having to do it on frigging arm 15:09:54 adamw: that's my point 15:09:57 *server* 15:10:06 but yeah, not sure about a freeipa server or a db server. 15:10:10 adamw: My servers may be small, but they're mine! 15:10:17 can we use something like the provider that's donating space for ARM compiliation? 15:10:20 I run PostgreSQL on my arm boxes all the time :-) 15:10:22 * danofsatx may be way off base 15:10:22 you can also use qemu, or internal hosts for arm server testing 15:10:25 nirik: I actually have an ARM-powered MariaDB server running at home, but... meh. 15:10:30 so unless someone's very clearly committed to doing it, i'd say -1 to ARM. i'm not realistically going to say i'd manage to get ARM testing done every milestone. sometimes i might be able to help. 15:10:48 ok 15:11:05 if we have a volunter to take care of ARM with server bits I think it may be ok 15:11:10 How 'bout I make a proposal... 15:11:16 Proposal: Approve the proposal from the aforementioned email with one revision: we will ship the ARM image as non-blocking. 15:11:18 note that we are just talking on whether it should be blocker right 15:11:24 right 15:11:27 or are we talking about ship/no-ship ? 15:11:38 just blocking/not-blocking, i think. i'm fine with shipping it. 15:11:43 * nirik waits for jsmith's proposal 15:11:49 heh...fedora-arm meeting is happening next door in meeting-2 15:11:49 simo: See the proposal I just made and vote or request a change to it :) 15:11:50 same here 15:11:59 fine with shipping, prefer non-blocking 15:12:10 sgallagh: +1 to your proposal 15:12:45 sgallagh: +1 to non-blocking for F23, but I'd like to change to blocking for F24 15:12:52 +1 to sgallagh proposal 15:12:58 +1 15:13:28 jsmith: Come back with at least one other person willing to commit to supporting that, please. (I don't want to take on that responsibility without covering the hit-by-a-bus case) 15:13:29 jsmith: I'd make it blocking once we have readily available arm64 hw 15:13:45 ARM64 is going to be a whole new can of worms. 15:13:50 right 15:13:55 (They're related by name only, really) 15:13:55 simo: Well, AArch64 is a bigger project -- it's still secondary arch 15:14:00 but at least it can be a primary target for server 15:14:09 simo: It should only be blocking once it's primary arch 15:14:18 sgallagh: well the cpu architecture is not wholly different 15:14:22 simo: Doesn't make sense to block on it before then, and that's going to be a ways out 15:14:26 Right, there's discussion about changing the way we define primary and secondary (and merging the various Koji instances) 15:14:27 sgallagh: the devices are though .. 15:14:29 right, but we might be changing how we do that. I think the conversation is out of scope here. ;) 15:14:36 nirik: +1 15:14:36 jsmith: ok 15:14:55 OK, so I count +5 to my proposal (including myself) 15:15:09 * adamw is with sgallagh: i'm fine with blocking on arm when we have people willing to commit to testing/using it. 15:15:11 mizmo: Do you foresee any issues with websites here? 15:15:29 well, not only testing/using it... but also fixing it. ;) 15:15:45 sgallagh, wrt adding arm as an arch? 15:15:46 MINOR DETAILS 15:15:54 nirik: Well, the ARM SIG is pretty responsive, once someone notices a bug 15:16:17 mizmo: We already have it 15:16:20 https://arm.fedoraproject.org/ 15:16:36 anyhow, if folks want it non blocking thats ok with me... 15:16:38 as sgallagh said, we are working on trying to enable things like the ServerWG promoting aarch64 as a primary arch for it while it remains secondary for WorkstationWG 15:16:43 sgallagh, so how would websites be affected? 15:16:50 is it that one is 64 bit? 15:16:58 so there's 2? 15:17:06 mizmo: I don't know. I was asking if changing from blocking to non-blocking of that image would be relevant to you 15:17:14 Right now, we aren't shipping the 64-bit version 15:17:26 64-bit isn't on the table for F23. 15:17:28 So set that aside. 15:17:45 In F22 we shipped the ARM32 disk image on the arm.fp.o page 15:17:47 so the arm image for server is no longer blocking? 15:18:09 mizmo: That's what we're voting on (with +5 in favor resulting in it passing unless strong arguments change someone's mind) 15:18:26 i think the website just wont display it if its not available so i don't think it's a huge deal. the one bit of suckage is that fedora minimal would be the only server image we could offer then 15:18:52 why would that suck ? 15:18:58 the image will still be built and shipped (unless generation is somehow broken), aiui 15:19:05 simo, cuz the website woul dhave a category for a bucket of one item :) 15:19:11 which would look weird 15:19:11 :) 15:19:13 but we might want websites to play down its importance, i guess 15:19:20 and that is the pain part :) 15:19:27 adamw: I think the point is that if it's built but jsmith's testing reveals serious breakage, we'd voluntarily not ship it too 15:19:29 let's uncathegorize the unbuckets! With unicorns 15:19:35 oh, fair enough. sure. 15:19:44 any major changes to the layou tof the page means i have to email arm's lawyers for another approval 15:19:52 * jsmith sighs... 15:20:09 mizmo: meh, due to arm trademark there ? 15:20:12 yep 15:20:16 lovely 15:20:17 ok 15:20:23 Mostly I am just crossing all my "T"s to make sure the right people know what's going on. 15:20:25 we'll cross that bridge when we get there 15:20:33 yep, we'll do what we gotta do 15:20:37 (or if as it may be) 15:20:59 sgallagh, its good to have a heads up that some changes might be needed there 15:22:27 To summarize the ARM discussion: We intend to ship the 32-bit ARM Server image if it passes the validation tests, but if it fails them we will not block the release of Fedora. Further, if it fails substantial portions of the release, we will not advertise and ship the ARM image. 15:22:52 Accurate? 15:23:12 WORKSFORME 15:23:14 Accurate. +1 15:23:26 ack 15:23:52 #info We intend to ship the 32-bit ARM Server image if it passes the validation tests, but if it fails them we will not block the release of Fedora. Further, if it fails substantial portions of the release, we will not advertise and ship the ARM image. 15:24:44 #agreed We approved the proposal for the deliverables described in the agenda email, with the aforementioned change in the ARM decision. (+5, 0, -0) 15:24:51 #topic Default NTP service 15:25:11 #undo 15:25:11 Removing item from minutes: 15:25:24 Actually, I just realized that one of those +1s was jsmith who is not a WG member. 15:25:37 mizmo: Would you care to vote? :) 15:25:42 #undo 15:25:42 Removing item from minutes: AGREED by sgallagh at 15:24:44 : We approved the proposal for the deliverables described in the agenda email, with the aforementioned change in the ARM decision. (+5, 0, -0) 15:27:24 Or stefw? 15:27:50 Proposal was: Approve the proposal from the aforementioned email with one revision: we will ship the ARM image as non-blocking. 15:27:50 * danofsatx will vote twice if he has to 15:28:21 +1 15:28:31 Thanks 15:28:36 #agreed We approved the proposal for the deliverables described in the agenda email, with the aforementioned change in the ARM decision. (+5, 0, -0) 15:28:41 #topic Default NTP service 15:28:57 OK, so there are two questions here: 15:29:18 1) Should we ship with a time synchronization service enabled by default? 15:29:18 2) Which one? 15:30:03 My smart-alec answers would be "yes" and "yes" :-) 15:30:08 So let's start with 1). Proposal: All Fedora Server installations should have time-synchronization against Fedora's timeservers enabled by default. 15:30:17 .fire jsmith 15:30:17 adamw fires jsmith 15:30:36 On a more serious note, I think NTP should be enabled by default 15:30:43 what, they don't already? sure. 15:31:07 adamw: I suspect it's been an oversight more than anything else, but we should at least assert a preference 15:31:12 "fedoras time servers"? 15:31:26 you mean the fedora.pool.ntp.org pool? or ? 15:31:34 nirik: I was just going to ask that... I'm assuming it's the fedora pool at ntp.org 15:31:35 i was assuming that was the intent, yeah. 15:31:38 nirik: Yes, those 15:31:45 ok, then +1. 15:32:02 fedora.pool.ntp.org 15:32:34 +1 15:33:15 I so wish HW vendors would put reasonably accurate clocks in their chips, but apparently that is never gonna happen, so NTP is the next best thing 15:33:17 danofsatx, mizmo, stefw: ? 15:33:42 /me wants an atomic clock on every PC and a chicken in every pot... 15:33:55 +1 on time sync service enabled by default 15:33:57 +1 15:34:23 #agreed All Fedora Server installations should have time-synchronization against Fedora's timeservers enabled by default. (+6, 0, -0) 15:34:28 (that included my implicit +1) 15:34:59 OK, so now the harder part of the question: We have three options for how to accomplish this, with varying levels of maturity and capability. 15:35:02 do we have a good choice on which to enable by default? 15:35:02 sgallagh: you want ATOMIC clocks? but think of the children! you're a monster 15:35:03 sorry cubing 15:35:20 sgallagh: I propose we use chronyd 15:35:29 it is the current default in Fedora and is mature enough 15:35:43 has 'some' advantages over ntpd apparently 15:35:51 * danofsatx concurs with chrony 15:36:03 Before we get too far; has everyone read up on the email thread? 15:36:21 I'm not sure I want to try to summarize all of the points there. 15:36:38 * danofsatx jsut read it this morning 15:36:41 * stefw has read it yesterday 15:36:47 * nirik has read it 15:36:56 ok 15:37:00 adamw five minutes from now has TOTALLY read it 15:37:03 (*reads*) 15:37:39 My only original reason for going with timesyncd was because we already had it and I was unaware of its limitations compared to ntpd/chronyd. 15:37:48 * simo read 'nuf 15:38:14 sgallagh: I am totally for the new kids on the block, except when I am not :) 15:38:15 Given that no one came to its defense on the mailing list, I'm prepared to knock it off the list of contenders, leaving us with the NTP vs. chronyd question. Sensible? 15:38:25 on general principles i'd be in favour of sticking with what the project as a whole has been using for years, i.e. chronyd 15:38:37 * nirik is slightly in favor of chronyd also 15:38:38 it's fine for products to differ when they have a good reason to, but if we don't, we shouldn't just pick one out of a hat. 15:38:39 Yeah, I feel the same 15:38:50 Yeah, I'm leaning that way as well, particularly with mlichvar's promises to help add any features we need 15:38:50 * simo nods 15:39:18 So, does anyone want to make a strong non-chronyd case 15:39:19 ? 15:39:38 15:39:41 Otherwise, I guess we'll call that consensus. 15:39:58 Proposal: stick with chrony 15:39:59 +1 15:40:03 +1 15:40:28 +1 15:40:33 (Where by "stick with" we mean "install it and enable it by default" 15:41:08 +1 15:41:13 danofsatx, mizmo? 15:41:55 so i have a q 15:41:58 sgallagh: we decided to install and enable with the previous proposal, this one is about which one we go with :) 15:41:59 sorry if its a dumb one 15:42:01 Go ahead 15:42:09 simo: Fair enough 15:42:11 if freeipa needs ntpd, does using chrony by default cause any issues / redundancies 15:42:19 mizmo: no issues 15:42:22 mizmo: No, FreeIPA long ago handled that 15:42:26 but you haven;t read the thread 15:42:30 gotcha :) 15:42:31 It just installs ntpd and disables chronyd 15:42:39 ah okay 15:42:42 +1 then 15:42:54 mizmo: we may end up changing freeipa to just use chronyd 15:42:55 +1 15:43:25 there is no inherent issue with, we just didn't get around to add configuration for it instead/as well as ntpd 15:43:29 #agreed Fedora Server will use chronyd by default for time synchronization services (+6, 0, -0) 15:43:40 #topic Websites Update 15:43:51 mizmo: You have the floor. (Careful, it's slippery) 15:44:20 lol 15:44:23 okay so two things 15:44:34 #1 the majority of text updates we came up with last meeting are live on the prod site now 15:44:45 #info the majority of text updates we came up with last meeting are live on the prod site now 15:44:52 #2 i haven't heard back from junland yet about his quote, if i dont hear back do we have a backup plan for a new quote? 15:45:14 (thats the only part of the text updates we came up with that hasn't been updated) 15:45:26 mizmo: Ask one of our friends or family to speak the quote for us? :) 15:45:37 Fedora Server: it's great, like bacon. only I don't eat bacon. But like people have told me bacon is great! 15:45:40 there, a new quote for you 15:45:41 authenticity! 15:45:47 no spoon feeding! 15:45:49 mizmo: We also have the option to use bracket replacement 15:46:01 Like in newspapers when substituting a more complete noun. 15:46:02 .fire adamw for not eating bacon 15:46:02 adamw fires adamw for not eating bacon 15:46:03 sgallagh, usually that is with approval of the speaker as well tho 15:46:09 for a product testimonial anyway 15:46:13 OK 15:46:17 bacon is so 4 years ago 15:46:33 newspapers - well they do what they want :) 15:47:48 well how about this 15:47:53 ill see if i hear from him by next meeting 15:47:57 mizmo: Perhaps send an email out to the users@ list and ask for people to tell us what cool stuff they're doing with each of the Editions? 15:48:03 if not next meeting we'll talk next meeting about finding another spokesperson? 15:48:04 (And/or on Magazine) 15:48:11 two weeks seems reasonable right 15:48:21 sgallagh, that seems like a good plan too 15:48:24 ill wait until next week 15:48:25 We could do this anyway; it never hurts to have extra quotes 15:48:34 has anyone heard from john as of late? 15:48:39 Worst case, we add a "Testimonials" page 15:48:49 user@ or server@ - since we're, like, you know, discussing the server page 15:49:14 danofsatx: Well, not all of our users are subscribed to the server@ list, which is largely a developer audience 15:49:18 oh yes, the users@fpo list is.... not a happy place would stick to server 15:49:39 * mizmo tried asking for 'user' feedback on the new website on users@fpo two websites ago..... never again 15:50:02 Sure, but maybe Magazine or opensource.com would provide some useful content. 15:50:02 point taken. 15:50:05 there's a lot of crumedgions on there. 15:50:19 magazine or perhaps social media? g+ ? 15:50:26 yeh all much better venues 15:50:28 Most of the people who talk on server@ are engaged in building it 15:50:29 cool thanks guys 15:50:56 maybe what ill do is post to magazine / social media and have a link to the magazine post sent to server for people to refer friends to 15:51:07 mizmo: +1 15:51:22 well i'm good 15:51:35 anything else on agenda? 15:52:01 #action mizmo to post to magazine/social media asking for cool uses of Fedora Server (and other Editions) 15:52:26 #topic Open Floor 15:52:35 Floor is open, mind the drop. 15:53:16 * danofsatx haz nothing 15:53:33 * simo minds the gap in his mind 15:53:33 oh god now im thinking about the game of thrones kid and his pit 15:54:01 http://acephalous.typepad.com/.a/6a00d8341c2df453ef017c33637be3970b-500wi 15:54:12 "open floor" 15:54:49 I think that we've come to the edge of rational discussion. 15:55:05 (And then retreated back into our more familiar irrationality) 15:55:12 Thanks for coming, folks. 15:55:12 * danofsatx looks for rationality in the backscroll 15:55:14 hunger 15:55:18 #endmeeting