14:31:02 #startmeeting Fedora QA meeting 14:31:02 Meeting started Mon Aug 12 14:31:02 2013 UTC. The chair is adamw. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:31:02 Useful Commands: #action #agreed #halp #info #idea #link #topic. 14:31:05 #meetingname fedora-qa 14:31:05 The meeting name has been set to 'fedora-qa' 14:31:11 #topic Roll Call 14:31:20 hi 14:31:41 ahoy hoy 14:31:45 * handsome_pirate waves 14:31:49 gday adamw 14:31:49 * tflink is present in meat space and virtually! 14:31:49 * Cerlyn is here 14:31:55 * jskladan tips his hat 14:31:57 * pbrobinson is here 14:31:57 * akshayvyas is here 14:32:00 * pschindl is here 14:32:03 * pwhalen is here 14:32:03 so in person we have: dgilmore, pbrobinson, rflink, jsmith, handsome_pirate, sgallagh and jskladan 14:32:22 pschindl: I don't see you around! 14:32:35 also tflink 14:33:10 we can set up a hangout for those of you outside if you like, but we're not all in vision of the camera and the mics probably wouldn't pick things up anyway 14:33:43 tflink is gonna set up the hangout 14:34:01 and chris tyler and andrew oatley-willis just joined us (more arm folks) 14:34:14 * kparal doesn't have the plugin 14:34:20 Hangout thingy is getting set up 14:35:08 * akshayvyas is takin RH333 class :D 14:35:13 kparal: there'll you be a youtube version which should work with html5 14:35:46 meeting wiki page link: https://fedoraproject.org/wiki/QA/Meetings/20130812 14:35:56 open it up as it has a detailed agenda on it 14:37:00 kashyap_: https://fedoraproject.org/wiki/QA/Meetings/20130812 14:37:05 adamw, Thanks. 14:37:22 #topic ARM release criteria / test matrix adjustments 14:37:35 so i'll try and do irc <-> real life shuttling in some kind of ad hoc way as we go along 14:37:45 and we're still setting up the hangout 14:37:51 but let's get moving :) 14:38:22 and viking-ice just arrived! 14:39:35 so I would like to go through the questions on the agenda in order if that's okay 14:39:50 #topic ARM: Which images will we build for which milestones? 14:39:55 adamw: i really don't think we should ;) 14:40:00 DAMN YOU GILMORE 14:40:20 * jskladan aaaand, my wifi is back up 14:41:04 plan is to build all ARM images for all milestones (same set as F19, plus Desktop) 14:41:49 #info plan is to build all ARM images for all milestones (same set as F19, plus Desktop) - dgilmore 14:41:55 minus Desktop? 14:42:03 GNOME doesn't work on ARM, does it? 14:42:04 Minimal, Desktop, XFCE, LXDE, SoaS, Mate, KDE 14:42:14 kparal: GNOME may work on a very small number of systems 14:42:15 no there will be desktop 14:42:20 that leads into the next question... 14:42:21 kparal: F19 there was no desktop, we are adding it 14:42:26 Live hangout is at http://youtu.be/s6MJ5wqVvBc 14:42:28 it will work with llvmpipe but it won't be fast 14:42:38 pwhalen: the Beaglebone Blacks support Desktop? 14:42:53 kparal, it will eventually 14:43:03 #topic ARM: Which of the images will be release-blocking? 14:44:05 Minimal is most critical from ARM perspective 14:44:23 xfce is most useful DE from arm POV 14:44:30 Xfce is the most important desktop for ARM users (pbrobinso) 14:44:55 xfce has been our release blocking desktop 14:45:35 KDE is current release blocking desktop that works well 14:46:06 minimal and KDE tested seperatelly 14:46:38 minimal initial setup is done via serial consol 14:46:44 s/consol/console 14:46:46 proposed #agreed release blocking image set for ARM will consist of minimal and KDE images for F20 14:46:53 KDE is graphical 14:46:58 +1 14:47:40 +1 14:47:41 dgilmore: did that llvmpipe arm issue ever get figured out? 14:48:47 +1 14:48:48 +1 14:48:51 +1 14:48:53 +1 14:48:56 halfline: it did but it is really slow 14:48:56 +1 14:49:10 +1 from me 14:49:18 great 14:49:21 #agreed release blocking image set for ARM will consist of minimal and KDE images for F20 14:49:35 halfline: gnome works. its just slow. 14:49:51 halfline: so for 20 im not sure we want to block on gnome. 14:50:07 #info vfat image will work on all hardware, so doing the vfat tests is the most important thing 14:50:23 #topic ARM: What requirements do we want to apply to the images? 14:50:31 ah okay, yea if it's chugging along at 2fps or something, that's not really something worth doing 14:50:52 Discussion on whether vfat images or ext4 images are more important to be tested 14:50:55 so it's sounding like we can pretty much apply the criteria as they stand to the minimal and KDE images; we don't want to restrict the scope for any ARM special case stuff. right? 14:51:56 #info some criteria are just irrelevant: installer and Xen/cloud stuff 14:52:09 #info dual boot with windows is irrelevant 14:52:18 https://fedoraproject.org/wiki/Fedora_Release_Criteria?rd=QA:ReleaseCriteria 14:52:28 Release criteria ^^ 14:53:11 # link Fedora 20 Alpha criteria -- https://fedoraproject.org/wiki/Fedora_20_Alpha_Release_Criteria 14:53:14 * kparal notes the video is ~1 min behind 14:53:55 kparal: serves you right for not being here :-P 14:54:01 Anaconda can be run via netboot 14:54:17 * kparal lowers his head and feels ashamed 14:54:20 Possible installer image for F20 (netboot) 14:54:29 i dont know any 1 by face ...... guys pls introduce 14:54:31 heh 14:55:11 In general the ARM testing I can do is limited since I'm stuck looking at OLPC remixes anyway 14:55:13 TECHNICAL INTERLUDE PLEASE STAND BY 14:55:44 Do not change the channel, do not adjust your set. 14:55:48 question we're dealing with now is: you can run anaconda on ARM but how important is this? do we really want to block beta if all the direct-deplyoment images work, but there's an ARM-specific bug in anaconda? 14:56:00 what frequency is this chat being broadcasted on? 14:56:40 adamw, Doesn't that depend on the severity of this ARM-specific bug? 14:56:50 Live hangout is at http://youtu.be/s6MJ5wqVvBc 14:56:56 fossjon: On the static between the stations 14:57:35 hangout volume is better now too 14:57:44 :) thanks jsmith && sgallagh 14:57:47 sounds like the answer is 'yes' 14:57:59 fossjon: You're welcome :-) 14:58:05 adamw: the discussion just started here :) 14:58:06 Cerlyn: is it too quiet? I can boost the volume on the mic 14:58:17 tflink: the volume is fine for me 14:59:25 #info calxeda is an important platform that requires anaconda for deployment, so we do want to block on anaconda at least for a standard singel disk partitioning setup 14:59:45 tflink: its good....dont boost i am using hadphones :P 14:59:48 Current discussing whether or not to support partitioning 15:00:14 No raid is officially supported 15:01:51 is there anyone else who needs introducing? 15:02:13 #topic ARM: Do we need to add or revise any test cases to cover this? 15:02:31 https://fedoraproject.org/wiki/Test_Results:Fedora_19_Final_RC3_Install 15:02:43 so this is the stuff we're looking at for this question (which we may decide to kind of punt on and figure out later) 15:06:56 https://fedoraproject.org/wiki/Test_Results:Fedora_19_Final_RC3_Base 15:07:41 https://fedoraproject.org/wiki/QA:Base_validation_results_template 15:08:04 So, current discussion is on which test cases are aplicable 15:08:24 Some do not apply to arm, such as DVD tests 15:09:22 #info looks like https://fedoraproject.org/wiki/QA:Testcase_base_initial_setup will need adjustment 15:09:27 It might be a good idea to look through the test cases during the arm meeting 15:09:58 timezone, language, user, root password, date/time (actually important on ARM systems without RTC) 15:10:28 Date/time even with Chrony may be slow coming up depending on how network is brought up 15:11:14 #info on a quick overview it looks like what we have to do is edit initial-setup test case, check for install test cases that simply don't apply, but other than that we're good 15:11:46 #topic ARM: How should we express the requirements in the criteria and validation test pages? 15:11:57 we have actually just done this, which is awesome. 15:12:39 totally 15:12:41 #info we're mostly wrapped up on ARM but mattdm has popped out 15:12:49 so, let's use this time for Any Other ARM Business 15:12:53 #topic ARM: Any other business? 15:13:16 we're gonna take a second to do a QA group picture 15:15:06 anyone can think of anything we need to query for ARM? 15:16:38 So, it may be a good idea for QA folks to show up at weekly arm meeting Wednesday 15:16:38 #info ARM folks will try to do some dry runs through the validation process before we hit TC1 to catch any missing coverage 15:17:39 #topic Visible Cloud release criteria / test matrix adjustments 15:18:23 so for anyone who's not aware: this is the https://fedoraproject.org/wiki/Changes/VisibleCloud Change (what was a feature) 15:18:41 basically it's about making the cloud images one of our most prominent download items, on a level with the live images 15:18:49 for us, this has the effect of making cloud pretty much a 'primary arch' 15:19:02 so we have to go through much the same process and questions as we just went through for ARM 15:19:18 #topic Cloud: What precisely are the configurations we are requested to test? 15:20:30 party foul: use of the s word 15:20:37 the guy in the green t-shirt is matthew miller, cloud guru 15:21:18 #info you can test cloud images locally 15:21:49 #info we have a fedora infrastructure openstack environment, and mattdm believes that working in that cloud env is a must 15:21:54 adamw: where r u :) 15:22:08 i jsut waved 15:22:20 .fire adamw 15:22:20 adamw fires adamw 15:22:26 akshayvyas: bottom right 15:22:31 .hire adamw 15:22:31 adamw hires adamw 15:22:43 !!!!!! 15:23:10 kparal: yeah and whoz in the GEEk tee 15:23:29 me 15:23:30 akshayvyas: red_alert 15:23:35 #info working on a local test system is NOT a release requirement - it's just a convenience for testing. we can ship if that doesn't work. 15:23:36 oh, he changed his nick 15:23:36 Group photo is at http://imgur.com/dUABWRT 15:23:38 :) 15:23:46 (Feel free to move it, etc.) 15:24:20 #info working on EC2 is also a release requirement 15:24:38 red_trela: cool 15:26:11 oh, just so folks know - we've budgeted 2 hours for this meeting live 15:26:57 #info fedora infra openstack cloud and ec2 are the two environments which matt sees as being 'release requirements' 15:27:16 #topic ARM: Do we have the necessary accounts, systems, knowledge to do this testing? 15:27:49 .fire adamw 15:27:49 adamw fires adamw 15:27:52 #topic Cloud: Do we have the necessary accounts, systems, knowledge to do this testing? 15:29:12 so the question here is: can we let a sufficient number of testers get the access they need to the fedora infra cloud and some kind of ec2 account in order to do this testing? 15:30:48 #info we can have cloud folks manually grant access to the fedora infra cloud to sufficiently trusted people, and use trystack to 'gate' that (if you want to be an openstack tester, you can start out on trystack and we'll give you privs later) 15:30:54 people can always set up their own OpenStack, too ;) 15:30:57 this is a bit of an ugly process hack for now, but we can refine it later 15:31:00 point! 15:31:11 adamw: do we hav any estimation of testers 15:31:11 btw, remote people, please do fire in points or questions, it's not a read-only meeting :) 15:31:41 so we have at least 5 people who are currently able to test this 15:31:50 i don't know if we have an estimation of people who may be interested 15:33:14 #info there is a community account for EC2 which some people have access to: Robyn is the gatekeeper 15:33:16 dont know much about cloud but can we create a single account that can be accessed by any one during testing 15:33:31 akshayvyas: the problem with that is it's very easily abused 15:33:48 akshayvyas: so i think the answer is 'no'; we need to have per-person accounts so we have accountability 15:34:04 adamw: yeah ....point 15:34:43 matt is going to talk about cloud-init for a bit 15:35:05 #info cloud-init is sounding like the major bit of the process that is likely to be breaking: the rest of the cloud stuff is very similar to a typical minimal deployment 15:35:26 it takes care of the initial configuration that is specific to cloud environments 15:36:58 #info we are going to want to develop some more specific requirements around cloud-init configuration magic, but that's for post-f20 15:37:37 #topic Cloud: What precise functionality do we require from the tested configurations? 15:37:52 or... 15:38:08 #topic Cloud: What precise functionality do we require from the tested configurations that is particular to cloud images? 15:39:40 #info ssh key injection, logging into fedora account with ssh key, resizing of root filesystem 15:40:12 #info there is an idea of pulling updates from s3 rather than public mirrors to save people's external transfer costs, but that is turned off for now as it was costing RH a lot of money to populate the s3 server 15:41:52 #agreed we care about "installation" (deployment) and base matrices for cloud. we do not care about desktop. 15:42:20 .fire pbrobinson 15:42:20 adamw fires pbrobinson 15:43:06 * Cerlyn hires pbrobinson to work for OLPC again 15:43:10 #agreed badge proposal: use of 'the s word' in a QA meeting 15:43:26 #topic Cloud: What changes are needed to the release criteria and test matrices to reflect this? 15:43:37 Making sure you have proper control of the firewall might be more important for cloud than it is for a desktop, as you need the remote control. 15:45:05 good point 15:45:20 Cerlyn security groups -- firewall managed outside of the system. had some people suggest that no iptables was the way to go 15:45:30 note that we are currently _not_ shipping firewalld 15:45:50 #info overview plan: we need the same 'direct deployment' tests as we need for ARM, the initial-setup test is a no-go for cloud, and we add a (set of) cloud-init test(s) which basically cover initial-setup territory for cloud 15:47:04 mattdm, Don't we? I think I could swear I disabled firewalld and enabled standard iptables get OpenStack running on F19. But I could double check :) 15:47:21 s/get/to get 15:47:41 kashyap_: you're speaking of the host, mattdm of the cloud image 15:47:58 red_trela, Oops, yeah. Ignore me mattdm 15:48:06 * kashyap_ kicks himself 15:48:10 .fire kashyap_ 15:48:10 adamw fires kashyap_ 15:48:25 oh, I can make adamw fire people :D 15:48:47 red_trela, That too without /his/ consent.. 15:48:57 #info regularly updated cloud images are something that's of interest but probably down the road a ways 15:48:58 .fire adamw 15:48:58 adamw fires adamw 15:49:01 :) 15:49:06 .hire adamw 15:49:06 adamw hires adamw 15:49:08 .fire tflink 15:49:09 adamw fires tflink 15:49:18 .hire tflink 15:49:18 adamw hires tflink 15:49:27 .hire everyone 15:49:27 adamw hires everyone 15:51:00 #action adamw to talk to mattdm and rbergeron about getting people access to the testing platforms 15:51:55 #info viking-ice and gholms may be interested in helping with testing and need access 15:52:04 anyone out in IRC land interested in becoming cloud test ninjas? 15:52:05 we will pay you in badges 15:52:23 will test on my own OpenStack infrastructures :) 15:52:28 adamw, Also might be good to send a "Call for volunteers" to the list 15:52:29 awesome thanks red 15:52:33 +1 15:52:40 #action adamw to send call for cloud testers to the list 15:52:47 red_trela: awesome, need to get feedback loops setup 15:52:53 +1 dgilmore 15:53:04 adamw: both lists, test and cloud :) 15:53:14 dgilmore: what's that? 15:53:32 red_trela: easy ways for you to provide feedback, so we know its working 15:53:44 red_trela: often the only feedback we get is when its broken 15:53:45 dgilmore: rhbz? and the test matrices? :) 15:53:52 we dont know when it works 15:54:20 it works whenever I don't complain ;) *good at complaining* 15:54:26 red_trela: test matricies yes, but otherways to say hey i tested and its awesome 15:54:36 dgilmore: right 15:55:11 red_trela: right but its good for us to know yes its great 15:55:14 adamw: I think there are some web services where we can have "a hangout" completely in HTML5, if that's what you meant by accessibility 15:56:47 kparal: not precisely what I meant, but an html5-based system is much more likely to be possible to make accessible than a flash-based one 15:57:03 kparal: but what we really meant is a11y accessibility - people with hearing / vision / speaking impairment etc 15:57:12 Hangouts historically have been limited to 10 participants (excluding those viewing on Youtube) 15:57:15 yup 15:57:27 #topic Open floor 15:57:28 voice certainly raises the bar for non-native speakers 15:58:42 #info tflink will try setting up a mumble server for us to use for stuff and see how that goes 15:58:46 kparal: that's a good point too 15:59:00 it also raises the bar for working from a coffee shop 15:59:05 headset! 15:59:11 mattdm +200 15:59:18 or public transport with a mobile phone :) 15:59:23 .fire kparal 15:59:23 adamw fires kparal 15:59:27 Fedora at one point had a SIP server but no one used it so it was shutdown 15:59:34 Cerlyn: right, we were just talking about that 15:59:44 Cerlyn: the problem with that was it was so hard to keep it working for two weeks at a time 16:00:05 so tflink is hopeful that mumble will actually be fire-and-forget and reliable enough that people will actually use it. we can try it, what the hell 16:00:34 so do we have any other business aside from video/voice stuff? 16:00:40 if you haven't seen badges yet, go see it 16:00:47 https://badges.fedoraproject.org/ 16:00:53 it is freaking AWESOME 16:01:00 and you get lots of badges for testing 16:01:05 (Trying to reply on IRC when the video feed is delayed is fun) 16:01:14 adamw: sure it's awesome when you dominate the ladder! :-) 16:01:16 oh, to make badges 'kick in' you have to do one thing of that 'type' - so if you do one bodhi feedback it will pick up all your historic ones 16:01:18 So we have backward incompatible systemd cgroup changes in rawhide which will break units that take advantage of the old controller stuff 16:01:19 kparal: ;) 16:01:34 https://fedoraproject.org/wiki/User:Johannbg/Systemd/cgroup-changes 16:02:35 #info this shouldn't break anything too crazily but we should keep an eye on it for alpha 16:02:44 #info So we have backward incompatible systemd cgroup changes in rawhide which will break units that take advantage of the old controller stuff 16:03:07 Cerlyn: yeah, sorry about that :( i've been trying to transfer the most important stuff to IRC as it happens 16:06:10 i think we're trailing off into "interesting discussion" at this point 16:06:19 yes 16:06:28 perhaps a link to the page ;) 16:06:30 so last call for any other business, or else i'll close down the irc meeting log, but we can leave the hangout running for people who want to follow this :) 16:06:34 Viking-Ice: which page? 16:06:38 oh the core os thing> 16:06:50 http://coreos.com/ 16:06:52 no the systemd cgroup changes 16:07:00 oh! didn't you link it? 16:07:00 for meeting notes 16:07:12 adamw: i dont think we have extra things 16:08:39 Viking-Ice: you don't need to do a #url or #link rthing, we figured out, any URL pasted just goes into the automatic summary anyway 16:08:51 hurray 16:09:02 don't tell that to the spammers, though :) 16:09:45 hehe 16:10:53 alright, thanks for coming folks! 16:10:55 #endmeeting