/srv/irclogs.ubuntu.com/2015/07/13/#ubuntu-ci-eng.txt

=== chihchun_afk is now known as chihchun
sil2100Browser restart needed08:30
ogra_jibel, sil2100 bug 147389808:57
ubot5bug 1473898 in ubuntu-system-settings (Ubuntu) "custom channel version in "OS Build Details" shows wrong version info" [Undecided,New] https://launchpad.net/bugs/147389808:57
* seb128 hits ogra_ with a duplicate bug report sticker09:06
* ogra_ puts on the sticker and wears it with pride09:08
seb128hehe09:08
seb128ogra_, joke aside we did discuss that issue on IRC several times, you knew it was reported no?09:09
seb128oh well, the question in the bug stand09:09
seb128is s-i providing a (dbus) api we can query to get the info09:09
ogra_i definitely forgot about it, sorry09:09
seb128no worry09:09
ogra_i think it does ... that a barry question though09:09
ogra_*that's09:09
seb128http://manpages.ubuntu.com/manpages/vivid/man8/system-image-dbus.8.html09:10
seb128see        Information() section09:10
seb128it has build number, device name, channel name09:11
seb128maybe the  version_detail one?09:11
ogra_yes09:11
ogra_but that needs to be parsed09:11
seb128why can't things be easy? ;-)09:11
ogra_system-image-clie does that (if you compare ots output with the raw channel.ini content)09:11
ogra_*-cli09:12
seb128k09:12
seb128so 3 options09:12
ogra_i guess you can steal the parser from there09:12
seb1281. we re-do the parsing09:12
ogra_(or at least a prototype)09:12
seb1282. we ask s-i to provide the parsed/formatted info09:12
seb1283. we system() call the s-i-cli command09:12
seb128I like 2. better09:12
ogra_as long as someone works on 2 :)09:13
seb128I'm not going to duplicate parsing code09:13
seb128so if somebody want to see that bug fixed they can nag the s-i team to have 2. implemented09:13
ogra_seb128, that bug *was* fixed09:14
ogra_it is a regression09:14
seb128?09:14
seb128I doubt it09:14
ogra_settings clearly showed the right thing for quite a while09:14
seb128no it didn't09:14
ogra_some time last week i noticed that my krillin still shows the right stuff while my arale showeed the 14... number09:14
seb128that code didn't change since cwayne reported it, which is before vivid09:14
ogra_and on the weekend i noticed that krillin switched the format too09:15
seb128k, maybe /custom/build_id has the right thing09:15
seb128had09:15
ogra_sure, *it* changed somewhere :)09:15
seb128yeah, just not on settings09:15
seb128likely somebody changed the content of those files on disk then09:15
ogra_could be the way the stamp is created (why we read it is beyond me though(09:15
seb128we read it because it's the only info we have and it's not documented09:16
* ogra_ only noticed the changes in the UI09:16
seb128so whoever did the setting code assumed that reading that was better than nothing09:16
ogra_well, s-i should be our only source for that info09:16
seb128yeah, but s-i doesn't publish that info09:17
ogra_it does, you just dont want to parse it :P09:17
ogra_(which is correct)09:17
seb128are you sure it does?09:18
seb128"              ยท version_detail - A string containing a comma-separated list of09:18
seb128                key-value   pairs   providing   additional  component  version09:18
seb128                details, e.g. "ubuntu=123,mako=456,custom=789"."09:18
ogra_yes09:18
seb128I'm not even sure those pairs contain what you are asking for09:18
ogra_they do09:18
seb128and if it does, is the format documented somewhere?09:18
ogra_look in channel.ini09:18
ogra_yes, in the system-image docs on the wiki somewhere09:18
ogra_version_detail: ubuntu=20150710,device=20150709-8965e37,custom=20150709-814-6-40,version=5809:19
ogra_thats what i have in channel.ini09:19
ogra_20150709-814-6-40 is the proper version for the custom tarball09:19
seb128k, fair enough09:19
Laneya{sv} anyone?09:24
seb128you would think...09:27
pete-woodstrainguards: silo plz kthx? :)09:28
sil2100pete-woods: on it in a min :)09:34
Mirvpete-woods: I can, if sil2100 takes a min :)09:34
Mirvpete-woods: silo 04609:35
pete-woodsMirv: awesome, thanks!09:35
=== chihchun is now known as chihchun_afk
infinitypstolowski: FWIW, the kernel team's response to this is that we're fools for expecting good usec resolution in the first place. :P10:18
pstolowskiinfinity, heh, yeah, I wouldn't expect 1 usec precision even though stat offers tv_nsec, but surely it should be different every second, no?10:20
infinitypstolowski: So, it gets more interesting.  Apparently, the mtime bumps at HZ resolution, according to Andy's math.  Sometimes, it's spot on, sometimes not quite, depending on platform.10:23
infinitypstolowski: So, that could be 4 times per second or 10 or whatever, depending on kernel config.10:24
infinitypstolowski: Anyhow, as I said in my mail, your bug is that you're checking for .tv_nsec, and if it exists, you're using it bare.10:24
infinitypstolowski: You need to use tv_sec *and* tv_nsec to have any hope of knowing if time progressed.10:24
infinitypstolowski: Cause it looks like the odds are very real that on a low-res kernel, tv_nsec can actually duplicate from second to second.10:25
pstolowskiinfinity, ok. thanks for investigating that! will fix and restest!10:25
infinitypstolowski: And that has a typo. ;)10:46
infinity+                    last_write_time_nsec_ = st.st_mtim.tv_nsec;10:46
infinity+                    last_write_time_nsec_ = st.st_mtim.tv_sec;10:47
pstolowskiinfinity, oh jeeez, thanks for keeping an eye on it :/10:47
infinitypstolowski: I was just curious to see your fix, but you get a review for free out of it.10:47
infinitypstolowski: Oh, the other thing I'd suggest on top of the code change (which is good), would be to alter the testsuite to sleep for slightly longer than 1s, ie: 1.111:00
infinitypstolowski: That would make it more resilient when running on ext3, since a jittery clock can make a system with no subsecond resolution give you the same second before and after a 'sleep 1'.11:01
pstolowskiinfinity, good idea, yeah11:01
infinitypstolowski: A 1.1s sleep would be guaranteed to cross a 1s boundary, unless the system is an Amiga from 1991.11:01
infinitypstolowski: (it also fixes your nsec timestamps, but that's moot, since you needed to fix your code to actually be sane anyway).11:02
* sil2100 off to lunch12:13
=== _salem is now known as salem_
cwaynesil2100: heya13:13
sil2100cwayne: hey!13:20
cwaynesil2100: question: I've got an update for the euronews scope that's just branding updates, shall I go through the ci-train, or JFDI?13:22
sil2100cwayne: how do those branding updates look? What's changing?13:29
cwaynesil2100: just like the scope headers/colors13:31
sil2100cwayne: I would say this can just go in as it is, but let's double-confirm with jibel13:34
renatuhey guys I am facing this bug #1461476 on OTA5 image13:39
renatuthis is very critical in my opinion, since my phone never gets locked13:39
renatuhttps://bugs.launchpad.net/ubuntu/+source/unity-system-compositor/+bug/146147613:39
sil2100jibel, davmor2: ^ ?13:40
davmor2sil2100: I've not seen that on arale or krillin, renatu is there a way to test for it that doesn't require plugging in a usb lead13:42
pmcgowanrenatu, do you know how you got it in that state?13:43
renatupmcgowan, I do not know13:43
sil2100My arale works fine here13:44
pmcgowankgunn, and I were trying to reproduce on vivid last week and could not13:44
renatupmcgowan, I remember that I was using a long delay something like 5min13:44
renatuI updated my phone some few times13:45
renatuusing over the air update13:45
ogra_renatu, thats a wily bug13:45
renatuand now it is not working. I try to change to 1 min but still not working13:45
ogra_wily bugs are generally low prio13:46
pmcgowanrenatu, what does powerd-cli list say13:46
renatuogra_, I am using 15.04 r(58)13:46
ogra_renatu, then i'd file a new bug if i were you13:46
pmcgowanogra_, it sounds like the same bug to me13:46
renatuogra_, in the bug comments there is more people with the same bug on vivid13:47
ogra_with over a month apart on twoi different releases ?13:47
ogra_well, whatever you think :)13:47
pmcgowanthe usc was recently syncd I believe13:47
pmcgowanrenatu, can you run that powerd command13:47
ogra_ah, into both ?13:48
pmcgowanneed to check with kgunn13:48
renatupmcgowan, http://paste.ubuntu.com/11872517/13:48
=== chihchun_afk is now known as chihchun
seb128ogra_, see what happens when you say it's fine to have bugs in wily ... things get copied later to vivid and then the bug hit users...13:50
pmcgowanrenatu, ok that doesnt tell me anything13:50
ogra_seb128, i dont say it is fine to have bugs in wily ... i warn people that they are treated lower prio than vivid13:50
ogra_(and vivid is definitely more important with wily never seeing a real phone )13:51
seb128right13:51
seb128I'm just saying that overlooking wily bugs bite us back on vivid13:51
renatupmcgowan, anything else that I can use to debug?13:52
ogra_if the same things land simultaneously ... for sure13:52
pmcgowanrenatu, I am grabbing someone for u-s-c13:52
=== karni is now known as karni-brb
kgunnpmcgowan: what channel ?13:54
kgunnpmcgowan: sorry, was otp, i'm about to set a silo up with a potential fix13:55
kgunnfor the screen blank13:55
pmcgowankgunn, screen blank?13:55
kgunni've discovered at least, fresh flash, just go to setting, change value to something, and it won't blank13:55
pmcgowankgunn, is it only on a new flash then?13:56
kgunnpmcgowan: seems so, somehow it seems to "recover" i presume through punching the powerbutton/user activity13:56
pmcgowankgunn, I dont really understand how the MR fixes it13:57
pmcgowanrenatu, was that the first bot after a flash for you?13:57
pmcgowanboot13:57
renatupmcgowan, no13:57
kgunnpmcgowan: first boot was only in terms of consistency13:58
renatupmcgowan, I have rebooted several times13:58
kgunni can't claim it truly self heals13:58
kgunnfirst boot = first boot after flash13:58
pmcgowankgunn, MR says we get invalid dbus events14:00
=== karni-brb is now known as karni
pmcgowanbut why is that happening14:01
alf_pmcgowan: here14:02
pmcgowanalf_, thanks, renato is seeing that bug on the ota14:02
pmcgowanalf_, how can he show its the same issue?14:02
pmcgowanalf_, and do we know why we get invalid events on dbus?14:03
kgunnor what's the diff between valid and invalid events14:03
pmcgowanmzanetti, can we get silos 18 and 35 signed off14:04
mzanettipmcgowan, both for OTA-5?14:04
pmcgowanmzanetti, possibly, debating cracking it to get them in14:04
mzanettiack14:04
mzanettidobey, hey, about this ^^14:05
mzanettiare you on QA duty today?14:05
alf_pmcgowan: kgunn: The events are not invalid per se, perhaps a better word is "uninteresting" for USC. USC is listening for clients disconnecting from dbus, so that if such a client disconnects USC can clear any keep-display-on requests it may have issued and not explicitly cleared.14:05
mzanettierm... sorry dobey, I meat davmor214:05
mzanettidavmor2, I have two silos, both for unity8, both need QA signoff14:06
mzanettidavmor2, should I merge them into one?14:06
dobeyoh ok :)14:06
alf_pmcgowan: kgunn: There was an error in our USC code that retriggered the inactivity timeout when we got such disconnection events from clients that hadn't registered any keep-display-on requests.14:06
davmor2mzanetti: rvr and I will be hitting silos shortly if he hasn't started already, what's up?14:06
pmcgowanalf_, any idea why this only happens rarely?14:06
mzanettidavmor2, so the thing is, we either merge them into one, or we need to QA, land, rebuild the other, QA, land14:07
mzanettidavmor2, what's your preferred approach?14:07
davmor2mzanetti: keep them separate, we only open the gates again this morning we were just finishing off the arale testing before moving onto silos,  are these suddenly urgent or something not been following the conversation sorry14:08
mzanettidavmor2, I think this is for OTA-5 still14:08
alf_pmcgowan: kgunn: That's what I have managed to reproduce at least... it's about the timing of these events. If e.g. the timeout is 2:00 and such an event comes at 1:50 then the timer is reset for 3:50 etc14:09
jibelmzanetti, which silos?14:09
mzanettijibel, 18 & 3514:09
kgunnpmcgowan: fwiw, it's not rare 1st boot post flash14:09
kgunnit's 100% for me14:09
alf_pmcgowan: kgunn: we could also have a different problem in addition to that one though14:10
jibelpmcgowan, the lockscreen in landscape lands in OTA5 finally14:10
jibel?14:10
alf_kgunn: which USC version are is included in OTA5?14:10
kgunnalf_: i'd have to look14:10
=== sil2100 changed the topic of #ubuntu-ci-eng to: Need a silo or CI Train support? ping trainguards | Need help with something else? ping cihelp | Train Dashboard: http://bit.ly/1mDv1FS | QA Signoffs: http://bit.ly/1qMAKYd | Known Issues: http://s-jenkin.ubuntu-ci:8080/ needs to be restarted, it's hung. Landing gates still closed
jibelmzanetti, davmor2 if both must go into OTA5 I'd rather merge them and do only 1 landing.14:12
mzanettiI'd say too. on it14:12
alf_kgunn: because I also found a related problem on USC trunk today (https://bugs.launchpad.net/unity-system-compositor/+bug/1473979), but I don't think the offending code has been released in any package14:12
davmor2jibel: fair enough14:12
kgunnalf_: unity-system-compositor - 0.0.5+15.04.20150506.1-0ubuntu114:13
kgunnthat's what's in ppa overlay14:13
seb128alf_, kgunn, https://launchpad.net/~ci-train-ppa-service/+archive/ubuntu/stable-phone-overlay/+sourcepub/5077581/+listing-archive-extra14:13
alf_kgunn: ok, that doesn't have the issue I just mentioned14:13
sil2100mzanetti: combining sounds nice, assigning14:15
davmor2mzanetti: so line 77 will be the unified silo for both right?14:15
sil2100mzanetti: assigned14:15
mzanettidavmor2, thanks14:15
mzanettisil2100, thank14:15
mzanettidavmor2, yes :D14:15
mzanettibuilding14:16
mzanettisil2100, I did not reuse the other in case we decide to only include one of them... But I promise I won't request any more silos now and use the current ones up asap14:17
kgunnalf_: so to be clear, i grabbed14:19
kgunnhttps://code.launchpad.net/~afrantzis/unity-system-compositor/fix-1461476-display-off-ubuntu/+merge/26412814:19
kgunnso we can test in a silo, and land if it solves14:19
dbarthhi trainguards; i'm trying to get a silo for line 6114:20
kgunn(i also created a wily silo as well with the other twin of that mp)14:20
dbarthfor reference, we uploaded a new rev. to the debian changelog (recommended by sil2100) to make that compatible with the CI sw14:20
alf_kgunn: Ack. Which channel are you using to flash the phone (to ensure we are in sync)?14:22
alf_kgunn: (Since you can reproduce it consistently)14:22
kgunnalf_: ubuntu-touch/rc-proposed/*14:22
kgunnany device14:23
alf_kgunn: and bq-aquaris.en or bq-aquaris.en-proposed?14:24
kgunnalf_: either...yeah, i think en-proposed just has custom tarball on it14:25
kgunnextras :)14:25
kgunncrap14:25
kgunnoh, it's just the wily one...14:26
alf_kgunn: yeah, that needs 0.14 mir14:26
kgunnalf_: ah...duh...you told me that, sorry i'm not coffee'd enough yet14:27
kgunnalf_: do we want anpok to make it part of mir0.14 ? or just land it later ?14:27
alf_kgunn: If it can make it into 0.14 then great14:29
alf_kgunn: I think we have time to pull it in14:29
Mirvdbarth: landing-052 for line 6115:06
alf_kgunn: let me know if the fix works or doesn't work for you15:18
popeysil2100: dbarth anyone reported issues between content hub and webapps?15:19
popeyit impossible to tweet a photo from gallery via. content hub15:20
popeybrowser dies during transfer15:20
kgunnalf_: thanks!15:20
popeyarale rc proposed 5815:20
kgunnwil do now15:20
sil2100hm, didn't see that reported15:25
sil2100davmor2, jibel: did you notice anything like that?15:25
dbarthpopey: on krillin?15:26
dbarthpopey: if you have too many apps opened, the oom killer might have you:/15:27
dbarthMirv: thanks!15:28
popeyarale15:28
popeydbarth: no apps open15:29
ogra_popey, well, there is that bug from sturmflut about arale memory management being totally off15:29
popeyit is unusable15:29
ogra_so you might hit oom even if there is plenty of ram15:29
popeywhich bug?15:30
davmor2sil2100: yes it is a failed test15:33
ogra_popey, if i could fine the # i would have given it here :P15:34
ogra_*find15:34
popeydbarth: 5 apparmor denials in twitter15:34
popeysorry for being terse. on mx4 with no pc15:34
ogra_popey, bug 146807715:35
* ogra_ tickles ubot215:36
cwaynejibel: hiya, would you like me to submit a scope update to QA first for a simple branding update, or just upload it to the store? sil2100 mentioned uploading it directly is fine but the check with you first :)15:37
popeyogra_: would i see oom in dmesg?15:40
ogra_syslog i think15:40
kgunnpmcgowan: alf's fix is looking good for me16:00
kgunnu-s-c screen blank16:00
alf_kgunn: \o/16:00
pmcgowancool16:01
kgunnpmcgowan: do i need to mark that tested? e.g. will we put into ota5 ?16:01
seb128alf_, why did you say earlier that the vivid version didn't have the issue?16:01
pmcgowankgunn, indeed16:01
kgunnseb128: 2 diff issues....16:01
kgunn1 issue exist only in wily16:01
seb128k16:01
kgunnthe other exsits everywhere16:01
mzanettijibel, davmor2: silo 44 is built and tested, ready for qa.16:04
=== chihchun is now known as chihchun_afk
pmcgowanjibel, davmor2 silo 49 is also ready16:27
=== dpm is now known as dpm-afk
slangasekrobru: meeting?18:00
robruslangasek: yeah I'm in there...18:01
slangasekrobru: no you aren't ;)18:01
sil2100We can't see you18:01
robruslangasek: in landing-meeting? google says I'm the first one18:01
sil2100landing-team18:02
sil2100mzanetti: whoops, I blew something up in unity8 trunl18:20
sil2100*trunk18:20
sil2100Fixing it now, apologies for that...18:20
mzanettisil2100, no worries :)18:23
sil2100Damn, stupid me...18:23
sil2100;/18:23
sil2100One wrong button press and everything went to hell18:24
mzanettihehe, what happened?18:24
mzanettisil2100, did you merge a silo instead of discarding it?18:25
sil2100Yeah, I checked 'force' instead of 'free only'18:26
sil2100;)18:26
* sil2100 does a push --overwrite18:27
* mzanetti looks away18:27
davmor2ssh sssssh it's oh so quiet, it's oh so still, it's oh so lovely....until, BIG BANG CRASH WALLOP sil2100 presses the wrong button18:27
sil2100mzanetti: could you check in a moment if it looks sane now?18:27
sil2100davmor2: ;p18:28
mzanettisil2100, looks ok. the history is a bit funny. the "Merged branch lp:~aacid/unity8/revert_session_screenshotter" comment seems to be in a LP translation commit instead of albert's one18:29
mzanettinot sure if that's how the train merges tho18:29
mzanettia pull on my previous trunk shows the correct modified files18:30
sil2100hm, I think it might be just leftover from the previous state18:30
mzanettilooks we're good18:30
popeycan someone try and play vid on http://pad.lv/147408118:31
popeyam on poor wifi and dunno if my video worked18:31
mzanetti"This video is private. "18:32
popeyballs18:32
popeyta18:32
mzanettinp18:32
popeynow?18:33
mzanettiyep, working18:33
popeyta18:33
robrubrbrunch18:36
davmor2sil2100: any second now18:47
davmor2sil2100: ^18:48
davmor2pmcgowan: ^18:48
pmcgowanwoot18:48
pmcgowanthanks18:48
sil2100o/18:52
sil2100Publishing18:52
sil2100Forced the publish18:53
pmcgowansil2100, building?19:26
davmor2sil2100, or robru: once the image is built can you please ping ToyKeeper who will carryout testing on it, many thanks19:26
robrudavmor2: hm, doesn't seem there's an imgbot at the moment19:27
robrusil2100: when are you building the image?19:27
=== bfiller_ is now known as bfiller_afk
pmcgowanI haven't seen imgbot for some time19:30
ToyKeeperHmm, well, I suppose I can just keep checking.20:06
ToyKeeperI assume it'll be krillin 66 and arale 59?20:07
sil2100popey, robru: yes20:23
sil2100It's building, but slow20:23
popeyyes ?20:23
sil2100uh20:23
sil2100popey: sorry!20:23
sil2100I meant pmcgowan ^20:23
sil2100;)20:23
* popey goes back to beer20:23
ToyKeeperAll you p<tab>s are the same, right?  ;P20:23
popeylulz20:24
* pmcgowan wishes for beer20:25
* ogra_ hands pmcgowan a Fucking Hell20:27
pmcgowanoh my20:27
ogra_http://www.younilife.com/site-uploads/2014/03/fucking-hell-bear.jpg20:27
ogra_(from the bavarian city "fucking" ... and its a light beer ... (hell in german))20:28
ogra_:)20:28
pmcgowanogra_, you had me going20:30
popeyi did wonder if ogra_ had lost it20:31
ogra_lol20:31
ogra_i only recently learned about that beer :)20:31
popeybut then remembered he lost it years ago20:31
ogra_lol, so true ...20:32
mzanetticihelp: this looks like a temporary issues to me: https://jenkins.qa.ubuntu.com/job/wily-adt-unity-scope-click/ARCH=amd64,label=adt/80/console20:49
mzanetticihelp: also only failed on one arch. can we try restarting it for getting silo 44 landed?20:50
cjwatsonmzanetti: can't retry on only one arch there, unfortunately (that'll be fixed by the new infrastructure I believe), but I've kicked off a new build20:52
mzanetticjwatson, thanks20:52
fginthercjwatson, thanks20:53
sil2100;/23:03
sil2100Why is the importer taking so long?23:08
sil2100It's running all the time23:09
=== salem_ is now known as _salem
sil2100ogra_, slangasek: is it normal that import-images runs for 2 hours already?23:23
slangaseksil2100: not normal, but it's not impossible...23:37
slangasekhmm why is it using xz instead of pxz23:38

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!