16:03:13 #startmeeting Server Working Group Weekly Meeting (2014-11-18) 16:03:13 Meeting started Tue Nov 18 16:03:13 2014 UTC. The chair is sgallagh. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:03:13 Useful Commands: #action #agreed #halp #info #idea #link #topic. 16:03:17 #chair sgallagh mizmo nirik davidstrauss stefw adamw simo tuanta mitr 16:03:17 Current chairs: adamw davidstrauss mitr mizmo nirik sgallagh simo stefw tuanta 16:03:19 #topic roll call 16:03:27 * stefw is sort of here 16:03:33 * nirik is here still 16:03:34 Hello 16:03:36 * junland is here. 16:03:57 * danofsatx-work is split with kde meeting in 1 16:04:05 er, in base - this is 1 16:04:24 Side-note: it would be nice if zodbot would let us correct the meeting title without ending and starting again. 16:05:06 patches welcome. ;) 16:05:12 understood 16:05:34 I count quorum, so let's get started 16:05:44 #topic Agenda 16:06:02 I have two items from the mailing list and two I thought of just before the meeting started: 16:06:11 #info Agenda Item: Approval of the Final Criteria 16:06:17 #info Agenda Item: Status check as we enter Final Freeze 16:06:26 #info Agenda Item: Fedora 22 Planning - GUI 16:06:30 #info Agenda Item: Fedora 22 Planning - Roles 16:06:41 Anyone have other topics for this week? 16:06:56 * junland has nothing in terms of agenda. 16:08:20 ok, then let's start 16:08:26 #topic Approval of the Final Criteria 16:08:33 They look ok to me... +1 16:08:49 #link https://lists.fedoraproject.org/pipermail/server/2014-November/001551.html 16:09:00 +1 16:09:10 /me was hoping to get adamw's input as well, but he doesn't seem to be around 16:09:29 he's on PTO thru tomorrow. 16:09:49 simo: Given that you represent the team responsible for meeting most of these criteria, I'm especially interested in your vote. 16:09:56 I think he's good with it. he wrote it, after all 16:10:08 danofsatx-work: ? 16:10:23 sgallagh: re-reading 16:10:24 adamw's view on the criteria....sorry 16:10:46 danofsatx-work: I wrote the ones we're talking about right now 16:10:53 oh wait, you wrote those. sorry 16:10:58 * danofsatx-work needs more coffee 16:11:17 sgallagh: I think you may want to add "the domain controller" must be able to server kerberos tickets ? (can be tested with kinit and kvno) 16:11:56 otherwise the only things that need to work are LDAP and DNS, and those are not sufficient for a domain join of freeipa-client (sssd) to work 16:12:05 "* Enrolled clients must be capable of authenticating against a valid user account using SSSD." 16:12:16 that is vague 16:12:21 Intentionally :) 16:12:27 you could use the ldap backend and do password based auth 16:12:44 Any reason we should not make sure krb5 works in the criteria ? 16:12:57 (sorry for missing it early I thought it was there) 16:13:06 Well, I'll strengthen that sentence to say that SSSD mus be usable with the 'ipa' backend 16:13:14 ok that will do 16:13:24 I didn't want to put any specific requirements on exact Kerberos features as part of the criteria 16:13:51 Our concern here should be for our ability to perform common operations 16:13:56 nothing exact in my wording, just the basic stuff 16:14:16 "* Enrolled clients must be capable of authenticating against a valid user account using SSSD's 'ipa' provider." 16:14:34 right getting krb tickets is part of common operations, but the line you throw captures it now 16:14:39 (which, interestingly, covers both Kerberos and LDAP, thanks to the migration mode) 16:14:55 * Users must be capable of performing password-less single-sign-on 16:14:55 between two enrolled clients using GSSAPI. 16:15:01 ah I somehow missed this one 16:15:09 which also checks getting tickets for other hosts 16:15:17 ok I think it all looks good then 16:15:20 #info Adjusted text: "* Enrolled clients must be capable of authenticating against a valid user account using SSSD's 'ipa' provider." 16:16:11 I'll interpret that as a +1 and give my own as well. 16:17:08 stefw: mind chiming in? We're technically a vote short of approval. 16:17:35 * stefw catches up 16:18:32 that wording looks good, and matches the intent of the criteria 16:18:39 so +1 to the adjusted text 16:19:30 #agreed Draft Criteria approved and will be submitted to QA for inclusion. (+5, 0, -0) 16:19:39 #topic Status check as we enter Final Freeze 16:20:08 ok we have 4 critical FreeIPA bugs 16:20:09 So as of about 16 hours ago, Fedora 21 is in Final Freeze. 16:20:22 simo: :( 16:20:25 I have a patches for one of them posted upstream and being reviewed 16:20:59 nirik: sorry found out only Thursday and the bug was nasty, took me 3 days to find all ramifications and build a solution 16:21:02 simo: are any of them blockers? or they should be freeze exceptions or? 16:21:21 good question 16:21:31 I think there was one considered blocker 16:21:58 * nirik doesn't see any freeipa ones on https://qa.fedoraproject.org/blockerbugs/milestone/21/final/buglist 16:22:00 the bug I fixed could be considered a blocker as I am changing (backwards incompatible) a control encoding as the original implementation was ... wrong 16:22:01 Just so we're clear, a blocker is any bug that violates one of the criteria we just voted on. 16:22:05 if there are any, please propose them asap. 16:22:14 If it doesn't do that, it's at *best* a Freeze Exception candidate. 16:22:18 but it has a minimal impact so it could conceivably be handled as a zero day 16:22:19 better to know sooner where we stand. 16:22:40 sgallagh: we have the TLS issue still I think 16:22:49 so that may be seen as violating the criteria 16:23:26 simo: One at a time; I'm not sure which one you're saying violated the criteria 16:23:57 nirik: FYI, the openldap one on that list has FreeIPA impact 16:24:04 Which is why I proposed it 16:24:25 ok. 16:25:33 sgallagh: the openldap one 16:26:07 simo: Right, that one I already proposed 16:27:36 Let's not turn this into a blocker review meeting. 16:27:59 We don't need to, just try and look them over and propose any that hit critera. ;) 16:28:05 simo: Can you commit to owning this and making sure bugs get requested for Blocker/FE ASAP? 16:29:05 sgallagh: I am not familiar with Blocker procedure 16:29:09 any pointers ? 16:29:20 simo: https://qa.fedoraproject.org/blockerbugs/propose_bug 16:29:23 ty 16:29:41 There are notes there to help you find criteria to match against 16:30:33 ok, other Server-relevant issues. 16:30:57 I checked in with stefw and the Cockpit team; we're going to need to request a Freeze Exception to clean up some branding stuff. 16:31:13 We're pretty much locked down as far as code-changes, so it shouldn't introduce any risk. 16:31:13 indeed, i haven't been able to get ahold of andreasn 16:31:14 good to know. 16:31:21 if it's a blocker tomorrow, i'll just try and do it myself 16:32:04 stefw: I'm not considering it a blocker; it doesn't affect any functionality and *can* be fixed with an update. It'll just not have the right appearance. 16:32:09 ok 16:32:50 Any objections? 16:33:06 nope. let folks know and we can test. :) 16:33:11 #info simo notes that there are several potential blocker issues in FreeIPA. 16:33:12 nope 16:33:26 none from the peanut gallery 16:33:26 #action simo to propose blockers and freeze exceptions for FreeIPA issues 16:33:53 #info Cockpit needs a branding update. We will propose it as a Freeze Exception and not a blocker. 16:35:28 Status on rolekit: ready to go 16:35:42 Thanks to danofsatx-work, we got karma to put it into stable just in time for the Freeze 16:36:32 yay me ;) 16:37:04 We still have ongoing branding stuff happening in the installers 16:37:07 I managed to get it done while sitting in one class listening to a lecture and prepping for a presentation in the next class. multitasking at my finest ;) 16:37:15 heh 16:37:17 But I *think* all we're waiting for now is spot to build a new fedora-logos package 16:37:32 haha 16:37:53 #info rolekit is good to go for F21 Final 16:38:25 #info Branding issues for all of the products are still being sorted, but should be in their final phases now. 16:38:52 I *think* we're also in good shape with the websites folks. 16:39:20 I worked with stickster to come up with some advertising copy for the features we're offering 16:39:28 yeah, hope so 16:39:50 #info Fedora Server brochure page looking good; remaining piece is in the Websites team's hands. 16:40:05 do you have the link handy for the draft? 16:40:34 I don't think it's deployed anywhere yet, but it's in the websites git... 16:40:51 danofsatx-work: No, stickster was showing me screenshots and I expect they're expired now 16:40:56 oh, ok 16:41:23 OK, that's all the status I have for now. Anyone else have anything to add? 16:41:30 https://git.fedorahosted.org/cgit/fedora-web.git/tree/getfedora.org/data/content/server/index.html?h=getfedora 16:41:41 were we discussing future roles? 16:41:42 somewhere around there and related files. ;) 16:42:21 danofsatx-work: We have more topics 16:42:27 oh, duh 16:42:32 * danofsatx-work goes to find more coffee 16:42:37 I should have said "anyone else have anything to add to the status check"? 16:42:51 * nirik has nothing more for status 16:43:16 #topic Fedora 22 Planning - GUI 16:43:38 This topic is largely a placeholder, but the topic came up in discussion today. 16:43:39 do we need to actually bless some specific GUI? 16:43:49 not today 16:43:54 Now that we're in Final Freeze, discussing future plans is a good idea 16:44:18 well, I mean ever... ;) I suppose we would if we want to ship one on the dvd... 16:44:24 nirik: I think we want to have a customized Server GUI installable, yes. 16:44:26 Yes, we need a plan. 16:44:32 KDE team seems willing to give a "low resource" specific DE a shot for server. 16:44:43 what would be be customized in this gui? 16:44:55 Do you think we could have the netbook workspace? 16:44:58 Something that may not be a complete Workstation/GNOME/KDE default, but stripped down for server needs 16:45:00 not have the window decorations and fancy graphics 16:45:19 this is very confusing to me 16:45:21 Reduced overhead and limited default apps, I think 16:45:23 a window manager, based on X11 or Wayland, to serve GUI applications like installers and such 16:45:25 i thought Cockpit is meant to be the server GUI 16:45:36 stefw: It is, but there are a couple cases Cockpit can't solve 16:45:39 people can obviously install whatever they want 16:45:41 Cockpit can't install, say, Oracle Database. 16:45:46 (Sorry, this was covered prior to your joining) 16:46:20 stefw: As danofsatx-work notes: there are a lot of third-party applications (especially in Oracle's family) that require a local GUI to install and don't forward over SSH very well 16:46:25 But I'll be working a draft and post it in the list 16:46:27 scratch that - DB is a text installer. Other Oracle apps, however, are GUI based. 16:46:43 yeah, sure, and people can just 'yum groupinstall @gnome' 16:46:47 right? 16:46:55 or kde or xfce or lxde or ratpoison, etc. 16:46:58 if they want to do legacy things, they can do wild stuff if they want 16:47:18 I'm not sure I'd call that "legacy", but yes this is the current state 16:47:26 stefw: Beyond that Oracle issue, I think it is reasonable to provide enough tools to diagnose and fix a cockpit failure. GUI tools, that is, because other kinds don’t count ☺ I’m hoping that a file manager, text editor, log viewer, and a privilege escalation method to edit files in /etc would be all that is actually needed. As for anything else, you’re right, there’s (yum install) for that 16:47:26 i think that X11 on a server is legacy 16:47:39 stefw: I wouldn't want the whole of gnome just for installing stupid oracle 16:47:55 stefw: besides gnome works really bad on non-gl graphics cards 16:47:56 if oracle has set things up to require gtk + X11 in order to install then so beit 16:47:59 but we don't need to bless it 16:48:07 stefw: I _absolutely_ want to focus on cockpit as _the_ GUI as opposed to anything GNOME-based, but really the CLI is far more legacy than X11. 16:48:14 stefw: So is there any alts for X11? 16:48:21 stefw: It's not just Oracle (though they are the 800lb gorilla) 16:48:27 * nirik is happy to read proposals, but not sure this is a great use of our resources. ;) 16:48:28 junland: Wayland ;) 16:48:42 stefw: would cockpit be able to export a VNC session with a stripped down DE (And I mean really stripped down to the bare bones) 16:48:45 danoofsatx-work:got it 16:49:02 it would be interesting to use something like an html5 vnc in cockpit 16:49:05 simo: what would that VNC session be used for? Only the Oracle-like installations? 16:49:06 but that's really just a tiny part of the issue 16:49:19 mitr: stuff that needs a UI, yes 16:49:22 the real issue is that you have to have this entire massive stack on your server to install some legacy app that does not work on headless servers. 16:49:32 you *can* totall ydo this 16:49:36 with whatever DE you want 16:49:40 right 16:49:42 but i don't think we should bless it in anyway 16:49:54 but providing a small footprint DE would be valuable I think 16:50:03 don't we? what about XFCE? 16:50:12 what about it ? 16:50:16 but junland's orignal proposal wasn't for remote management, it was for local management for small-business deployments (read: not data center) 16:50:21 it's a small DE 16:50:36 For the record, I'm not very keen on us elevating a new release-blocking desktop either. 16:50:38 if it fits the bill fine by me 16:50:39 We also have to look at Pedestal servers not just rack mounts 16:50:46 perhaps we could gather more concrete use cases on the list? it might be that just documenting how to install things could meet these needs? 16:50:56 cockpit can have tools added to it 16:51:24 nirik: I'll provide cases. 16:51:24 ie, here's how to install $desktop to a fedora-server install: 16:51:29 stefw: What is the recovery path to cockpit being “broken”? (Someone incorrectly editing firewall/network config, or running out of disk space perhaps) 16:51:40 mitr: SSH 16:51:43 mitr the shell 16:51:43 FWIW I would love to have FF as the root runnign app that shows just cockpit, which can optionally run a DE in VNC in localhost ... just to have mindblown :) 16:51:48 identical to the cases for X11 being broken 16:52:01 sgallagh/stefw: CLI doesn’t count for the users we are building cockpit for 16:52:03 simo: You scare me sometimes ;-) 16:52:07 simo, yes that would be cool for servers with a monitor 16:52:27 mitr, yeah any GUI can break down 16:52:30 stefw: more importantly gives the same interface to the user whether there is a monitor or not 16:52:31 whether xorg or cockpit 16:52:40 Though... I kind of wonder if we could bludgeon FirefoxOS in there... 16:52:40 Web De? 16:52:40 DE* 16:52:46 I”m not saying that a local GUI needs to be _the_ solution, but it seems the easiest one I can think of (as opposed to a "magic” rescue ISo one can boot into to fix “everything”) 16:52:54 sgallagh: no, I am pretty sure not. ;) 16:53:13 mitr: It also needs to be a option for IT admins... 16:53:48 anyway, i think server admins can solve this today. I see it a legacy work around 16:53:53 and they can solve it with one command 16:53:57 junland: Most/all that IT admins need to do should eventually be possible to do trough cockpit. It seems to me long-term we only need the Oracle and the disaster recovery cases 16:54:01 mitr: how about a systemd target for rescue? but it's hard to know all the cases it would have to rescue you from. ;( 16:54:11 nirik: Yeah, that’ 16:54:16 s the issue. 16:54:34 Giving users tools to do arbitrary debugging seems far easier than writing a self-healing system (unfortunately?) 16:54:41 and gui/de is going to be more complex than just a command line, so in the end there will be cases where that won't help you either. 16:54:51 True 16:54:56 mitr: Got it. I just want everyone to be on a agreement / disagreement for me creating a proposal. 16:55:40 junland: I think there will inevitably be discussion after this meeting and after you post a proposal. That is fine and good :) 16:55:45 Yes 16:55:47 junland: always happy to see proposals, but thats not saying we will want to do that. ;) 16:56:05 junland: Yeah, please put together a proposal and we'll see if it meshes with our strategy. 16:56:15 This was mostly intended to bring the topic back to mind 16:56:38 Got, just gonna shut gun this one and put my thoughts together for a proposal. 16:56:54 Sounds good. Thanks very much. 16:57:09 We're almost out of time today, so let's defer the Roles discussion until next week. 16:57:25 #topic Fedora 22 Planning - Roles 16:57:30 #info Deferred until next week 16:57:36 #topic Open Floor 16:57:46 Anything for Open Floor? 16:57:58 * nirik has nothing. 16:58:01 * junland has nothing 16:58:08 * mitr has nothing 16:58:39 OK, then let's call it a day. 16:58:45 Thanks for participating, folks! 16:58:56 thanks for hosting, sgallagh 16:59:04 #endmeeting