[00:29] SergioMeneses: hi, I have let the laptop team know that you are not available that week for the classroom sessions. If it becomes impossible, then we can have a chat with the classroom team and arrange that for a different week. [00:32] phillw, but we have more members in the group... I'm not the only one [00:32] :) === _salem is now known as salem_ === fginther|afk is now known as fginther === salem_ is now known as _salem === _salem is now known as salem_ === salem_ is now known as _salem [04:37] Good morning === elfy_ is now known as forestpiskie [07:05] good morning [09:39] has everyone added themselves to https://wiki.ubuntu.com/QATeam/WhoWeAre [09:41] * pitti adds himself [09:43] yay, one more person [09:44] now if only kirkland and roaxsoax would stop having lives, my work items for S cycle would be complete [09:47] pitti, if only you'd put yourself in the right spot ( +1 goes above +2) [09:47] * Noskcaj fixes [09:47] Noskcaj: well, jibel's +2 is really the same TZ that I'm in [09:48] Noskcaj: perhaps fix jibel's to say UTC+1 ? DST applies to all of us, after all [09:48] ok, sounds better [09:48] (and CEST → CET) [09:48] IMHO it's a bit pointless to keep mentioning DST in every other row [09:49] i thin kit's all fixed [09:49] *think it's [09:50] Noskcaj: not quite, CET = UTC+1, CEST = UTC+2, I'll fix [09:51] ok, ty [09:51] * Noskcaj doesn't understand European or Us timezones [09:51] Noskcaj: you don't have DST? [09:51] oh, nevermind [09:51] pitti, i do, but i mean i don't know the abbreviations etc. === cking_ is now known as cking === _salem is now known as salem_ === salem_ is now known as _salem === _salem is now known as salem_ === hggdh_ is now known as hggdh === mmrazik is now known as mmrazik|afk [15:20] smart scopes is here if your on saucy :-) [15:21] sweet :D [15:23] balloons, nice! [16:03] hello ... is there someone specified what exactly is done here: http://reports.qa.ubuntu.com/memory/arch/armhf/ ? [16:03] e.g. at what time do we measure for "Fresh Install - touch" ? [16:03] and what does "Daily Upgrade - desktop" mean? [16:04] who knows anything about that? [16:04] plars: ? [16:05] asac: at one time, we were doing these tests on ubuntu desktop image running on nexus 7. For a long time, preseeding didn't work right so we were doing day-day upgrades [16:06] ah i see [16:06] asac: you can ignore those now though, as the desktop nexus7 images are no longer an issue. I've asked to have it removed or at least archived from the view [16:06] it just means: this is not a fresh install, but something we upgraded as we didnt know any better [16:06] but its still the same measurement point. [16:06] asac: also, be aware, there is a problem with jenkins publishing right now, so all the results you are seeing are old [16:06] so what is measured here? [16:06] right after login? [16:06] jobs are still running, but not getting published properly [16:07] what does that mean? [16:07] after boot, it was auto-login though, so effectively the same [16:07] multiple runs though [16:07] but log in to desktop? [16:07] or just headless? [16:07] yes, it's logged into the desktop [16:07] ok. [16:08] why did our stddev increase so much recently? [16:08] (assuming the black line is that) [16:09] i guess the big spiked are really infrastructure bustages or soemthing [16:09] but seems we settled on afar higher value than 1 month ago [16:10] plars: also, do we wait after logging in until the system has settled? how are we doing that? [16:11] asac: at the moment, we just give it 5 min, then we run smem every minute, 30 times for each run [16:12] asac: also there are some things that get disabled [16:12] asac: like the cron job for apt-update-xapian-index [16:12] ic [16:12] only applicable on desktop of course [16:12] ok [16:12] but you work on revamping that in the way tvoss proposed in the doc? [16:12] or is that already worked in? [16:12] yes, we've been talking to tvoss about that [16:12] are those changes underway? [16:13] yes, javier is working on that piece [16:13] felt reasonable and hopefyully would mkake the number more useful [16:13] when do you think will that land? [16:13] no idea at the moment, there's a lot to be done there, and last I talked to javier he was running into some issues with autopilot on it [16:14] so we may have to come up with a different approach if those cannot be resolved [16:14] some of the events we need to wait for are ui events though, so we hoped it would be a good fit [16:14] stop :) [16:14] i dont undersatnd how autopilot comes into picture [16:14] is that part of the use casese we wanted to measure [16:14] * asac opens the doc [16:15] asac: because some of the things that were requested require driving the ui [16:15] my understanding is wee want to measure these with reboots in between X times: [16:15] 1. Phone booted [16:15] 2. Browser started [16:15] 3. Browser finished loading [16:15] plars: ah ok... autopilot is used to drive UI for those [16:16] exactly [16:16] but we talk about the same list, correct? [16:16] yes [16:16] ok so we have two things: [16:16] 1. move to different approach on how we measure (with reboots in between) [16:16] autopilot seemed like the most direct approach, but if it doesn't work out, other possibilities will be investigated [16:16] 2. land use case measures from doc [16:16] can we have 1. before 2.? [16:18] i believe 1. can be delivered pretty soon, while 2. might take a bit as you need to model the test cases in autopilot [16:18] asac: what do you mean by moving to a different approach? [16:18] have you checked if they have similar autopilot tests? [16:18] i guess there might be something close enough to avoid redoing that [16:18] well, you still need to make sure that you measure at *exactly* the same place [16:18] plars: i cant remember exactly. tvoss said you discussed changing how you measure to get better results [16:19] i believed it was about not measuring 20 times aftrer one boot [16:19] but rather booting 20 times [16:19] and measure after certain events [16:20] if I recall correctly from the meeting, even rebooting 20 times was something he felt like was overkill, and maybe worth trying once to see if it was consistent, but also agreed that if we aren't seeing large fluctuations from one day to the next, we are effectually doing the same thing [16:20] plars: the current chart shows high stddevs [16:20] those numbers are not very good [16:20] and will not allow very tight steering of our memory barriers [16:23] asac: yes, I've looked at those [16:23] right. ok lets chat on monday when tvoss is back [16:23] the biggest thing at the moment is that we have a few processes, mainly one in particular [16:24] hang on, the name escapes me... [16:25] hud-service [16:25] that pops on at the beginning [16:25] we could wait a bit longer and wait for that to go away [16:25] my understanding is that it's just active 1. right after boot and 2. when there's activity [16:26] maybe we can turn services that we dont want to measure off? [16:26] "quiesce" the system? [16:26] plars, asac: trying to catch up, but tvoss wants something orthogonal to the current test we have [16:26] I was looking into that late last week. delaying isn't ideal, but it's also not much better to check for the presence of arbitrary processes [16:26] there could always be "something new" [16:27] doanac`: dont try, we are far faster in writing than you can think :) [16:27] :) [16:27] sorry its friday afternoon [16:27] doanac`: indeed, I don't think we want to change what we have, so much as we want to add to it [16:27] doanac`: really? can you give me your story? [16:27] asac: I know! :) [16:27] let me find the link [16:28] https://docs.google.com/a/canonical.com/document/d/1XjgLa59wFhdqErXb4rHs3FoojsDbMyf0oQa0Km2W-qY/edit [16:28] I think waiting an extra 5 min would do it [16:28] as far as this problem goes [16:28] so its measuring memory at certain times, but also measuring the process itself not just the whole system [16:29] ie - how much memory does browser-app take [16:29] hud service seems to disappear by the 4th run on the results I've looked at [16:29] so there are more distinct things you measure [16:29] and compare [16:29] which means 5min delay, +4 once/min runs [16:29] so that would be 2 min buffer [16:30] i guess you still have to wait for the system to quiece though since that's one of the memory metrics that are listed [16:31] for now, I think it makes sense to change the delay at least, which should tighten up the stddev quite a bit I believe [16:31] plars: we should impleemnt real triggers [16:31] not do arbitrary sleeps [16:31] just as a general thing [16:31] :() [16:31] asac: true, so you want the system to tell us when it's quiesced? [16:31] and i am sure doanac` agrees [16:31] exactly [16:32] that's what I was saying earlier too [16:32] maybe a voting daemon [16:32] but how does it know? [16:32] multiple components saying that they are ready [16:32] and once all are done, fire off [16:32] if you know of a better way than waiting or watching processes that is [16:32] requires work [16:32] yes [16:32] look at all the stuff that starts [16:32] ensure it can tell you when its feinished [16:32] (will be custom solutions for each thing for sure) [16:32] and see [16:33] also figure if we can shut down stuff [16:33] but thats not really good. we want full experience [16:33] asac: want to talk to the touch foundations folks and see if they have something like this that could be worked in? [16:33] phonedations? [16:33] asac: or is that something we should talk to tvoss about? [16:33] well, you guys could do analysis of what is noisy [16:34] then we can work together to ensure you can get triggers [16:34] asac: it's hard for me to say that word :) [16:34] yeah [16:34] lets really check monday [16:34] asac: I can tell you what's noisy now, I just did [16:34] asac: it's that one process at the moment [16:34] tvoss is on vacation :) ... so we can blame him anyway until then [16:34] asac: but I can't guarantee it will always be that way in the future [16:34] haha, ok [16:34] and have a good weekend [16:35] plars: so 1% stddev is good [16:35] 2% is acceptable [16:35] everything beyond isnt really that nice [16:36] its a reproducible machine after all ... should be very very low [16:36] if we do it right [16:36] doanac`, asac: for the moment, I'm going to extend the quiesce delay at least and see where that gets us. It's certainly wrong to have some runs with hud-service active and others without it [16:36] asac: indeed === plars is now known as plars-afk [16:44] yeah lets see [16:45] do extra runs over weekend so we see a pattern === plars-afk is now known as plars [17:31] balloons after my update today i've lost totem. Is it planned to be removed? Curious since we have just merged the totem test [17:32] DanChapman, lol..not to my knowledge. I'm guessing you got a partial upgrade [17:33] that means not everything hit the archive at the same time.. happens in the dev release. You have to be careful how you upgrade and make sure it doesn't want to do something wacky like remove half your desktop packages [17:34] ahh right i see. Its the only thing that went as far as i can see. Will be more careful :D [17:35] :-) [17:35] it's simple enough to re-install.. make sure you still have ubuntu-desktop meta package too [18:30] asac: we will land the first test next week [18:31] asac: and soon after it'll start reporting [19:08] Letozaf_, hello again ;-) [19:08] balloons, Hello :) [19:08] I just saw your MP.. time to finish this! [19:08] balloons, yes [19:08] balloons, pitty that drop shadow does not work :p [19:10] Letozaf_, trying to convince me to hack on it a bit more :-p [19:10] ? [19:11] balloons, :) only if you think you can solve the problem, but I was already preparing to file a bug :p [19:11] indeed.. we'll let it go to the bug [19:11] balloons, ok :-) [19:12] it's in.. I'll push to the production branch too [19:12] whew.. excellent [19:14] balloons, good ! [19:36] balloons, I reported but number 1188781 [19:36] bug #1188781 [19:36] bug 1188781 in Autopilot "autopilot fails to detect screenshot attribute" [Undecided,New] https://launchpad.net/bugs/1188781 [19:36] balloons, sorry bug [19:36] not but [19:37] balloons, yes right :p [19:38] Letozaf_, perfect.. we should try and make sure we open bugs when we get stuck or otherwise leave an open feature [19:38] at any rate, excellent work [19:38] balloons, ok, thanks :p [19:39] so we cleared through everything :-) [19:39] not bad for a little over a month eh? Remember the brokenness in April :-) [19:39] balloons, yes looks like [19:40] balloons, yes great job [19:40] balloons, we have a lot of tests working now [19:41] Letozaf_, yes everything in the production branch works.. and will always work [19:41] the qa team is working on integrating that so they'll be running everyday [19:41] it should happen this month [19:41] so yay! [19:41] balloons, cool! nice to know it YAY!!!! [19:41] so enjoy your weekend.. lots of things to be happy about.. [19:42] balloons, sure and thanks for your help :) [19:42] balloons, hope you enjoy your week-end too [19:43] balloons, hopefully it will stop raining here in Itlay this week-end, I am really sick of rain :p [19:43] sun came out again here Letozaf_ [19:43] we had a tropical storm dumping rain and wind for several days here [19:43] but it's gone today finally [19:44] balloons, also here, today was quite fine, but until yesterday it kept on raining and it was cold [19:45] wild.. is that normal for summertime? [19:46] balloons, noooo!!! this year is really odd, well it's still spring, but this time of the year shouldn't be so cold and rainy [19:46] still spring? lol I suppose [19:47] it's June! [19:47] balloons, well officially summer starts the 21st of June :p [19:47] where I used to live last year it was like 8 c at this time while I was visiting.. very crazy cold for no reason [19:47] sorry.. I was visiting last year, the place I grew up [19:48] balloons, I think it's better in Florida, weather wise [19:48] yea, I think so :-p [19:49] but people who want "perfect" weather tend to hit the california coast. You can have 16 C year round [19:49] or in san diego 23 c year round :-) [19:49] balloons, thats great! [19:49] balloons, next time I decide to go and live somewhere else I will go to S.Diego [19:49] I prefer the variation I get here.. but in general no cold for me thanks :-) [19:50] Letozaf_, lolololol [19:50] balloons, neither do I like cold [19:50] South Africa and California I think are very similar climate and geographywise [19:50] balloons, but this year weather was really bad up to now, hope things will change [19:50] not sure how long you lived in SA [19:50] balloons, yeah South Africa has great weather [19:51] balloons, I left SA when I was only 10 I was born there [19:51] balloons, so I lived there 10 years [19:51] lol. just a kid :-) But old enough to remember a bit perhaps [19:51] if you have better memory than me [19:51] I've no idea what I did at 10 [19:52] balloons, well I remember all the nice things, my friends, school, my house the good weather [19:52] that's the best way to remember :-) [19:52] balloons, yes [19:53] balloons, I remember I used to hate Africaans, infact do not remember much [19:58] Letozaf_, hmm.. More than me. [19:58] balloons, :) [20:10] hello Noskcaj [20:10] hey balloons [20:11] i think i've got everyone onto thw WhoWeAre page [20:12] nice.. how's the testdrive and using dev release page coming? [20:37] balloons, the page is finished except for screenshots. [20:37] and i need kirkland and roaxsoax to stop having lives outside of ubuntu [20:38] Noskcaj, lol [20:38] it's hard sometimes [20:38] it's winter for you but summer for everyone else [20:39] on a different topic, i have a topic for vUDS, but my timezone doesn't allow me to do anything about it. Get Magnet links in Ubuntu [20:42] Noskcaj, I would open a blueprint and share it.. Sadly if you can't but there it's a little tough [20:42] i'll try [20:42] back to testdrive: i also need howard to live in a place that doesn't have a huge exam period, like australia [20:43] lol.. he's in a compatible timezone though! [20:43] it could be much worse [20:43] one more week and he's done [20:52] good point === salem_ is now known as _salem