mandelmorning all!08:17
JamesTaitGood morning all! :)08:28
=== eu is now known as Guest67155
gatoxgood morning!11:29
ralsina_good morning!12:09
alecumood gorning!12:10
gatoxralsina_, alecu hi12:13
gatoxalecu, i'm waiting for charles because i found a weird problem with the syncmenu12:13
ralsina_gatox: "weird".... do I want to know?12:14
alecugatox: what happened?12:15
gatoxralsina_, is not something that we were using so far..... but the issue i'm working on requires that..... and it seems to be a problem12:15
gatoxalecu, i'm preparing a script independent from u1-client now to test it12:15
gatoxbut the problem is12:15
gatoxthat when i execute the call to set_paused to change the ON|OFF button in the menu from u1-client to keep the menu state synced with syncdaemon...... it seems that it enters in a loop and keep changing the signal all the time eating all my cpu12:16
mandelralsina_, gatox hello o/12:18
gatoxmandel, hi12:18
mandelralsina_, I landed my first branch in nux.. a rather pathetic one: https://code.launchpad.net/~mandel/nux/set-password-mode-correct/+merge/12642612:18
mandelralsina_, explains what I mean when I say it sucks..12:18
ralsina_mandel: no comments12:19
ralsina_mandel: I *do* have comments, but I will keep them here, in this little box.12:19
mandelralsina_, which I'm sure I share..12:19
ralsina_mandel: I mean... this basically means noone ever *used* a password box, right?12:19
mandelralsina_, exactly, we are the first ones12:20
mandelralsina_, and tabbing also broke our staff on the unity side which I already fixed in the preview12:20
ralsina_mandel: awesome12:20
mandelralsina_, at least I'm learning the code base well looking for stupid things like this12:21
ralsina_mandel: hey, that glass is half full! :-)12:21
mandelralsina_, is the only possible way to look at this12:22
gatoxalecu, you can test it with this script if you want: http://paste.ubuntu.com/1228368/12:22
mandelralsina_, I'm also nearly done with the link ui element, problem is, I have no design input regarding its focus looks etc..12:22
mandelralsina_, so I'll have to spend some time on that as soon as it is done12:23
gatoxalecu, you can see there is not any kind of loop calls there....... and if you press the ON|OFF in the menu the function is called just once...... but if you press the other button..... it gets stuck calling the "change_sync_status" function12:23
ralsina_mandel: http://what-if.xkcd.com/6/12:23
ralsina_mandel: oh, you reminded me what I have to do now :-)12:24
gatoxralsina_, alecu if it's okay for you.... i'll move on to the shares tab issues now..... and wait for charles to answer my ping......12:24
gatoxhe always answer....... but he is away right now12:24
ralsina_gatox: sure12:25
mandelralsina_, lol12:26
=== ralsina_ is now known as ralsina
gatoxah dobey ..... i needed to ask you something12:41
gatoxdobey, do we need a ui freeze for this? https://code.launchpad.net/~diegosarmentero/ubuntuone-client/menu-separator/+merge/12600612:41
gatoxdobey, ralsina told me to ask you12:41
dobeyto land in trunk? no. to land in stable-4-0 and ship to ubuntu next week, technically yes.12:41
ralsinagatox: see, toldya12:42
ralsinadobey: we can probably slip it by since it's not visible in the default install12:42
dobeybut as there are no docs for that UI, and it's not visible by default, and there are no string changes, we can probably slip it in12:42
ralsinadobey, gatox: but it's easier to do post-release in a microreleae12:42
ralsinaOMG, I am channeling my inner dobey!12:43
ralsinadobey: BTW, feeling better?12:43
dobeyand ui freeze or not, no reason to block it landing in trunk…12:44
ralsinadobey: I am reluctant to have trunk and stable-4 diverge right now12:48
ralsinadobey: unless there is a very good reason12:48
mandellunch time here12:48
=== mandel is now known as mandel|lunch
dobeyralsina: they alreayd have, for u1client anyway12:53
* ralsina goes diff12:55
ralsinadobey, gatox: ok, in that case, let's put that in trunk and try to sneak it12:55
ralsinalet's make lemonade out of the no-sync-menu lemon12:55
dobeyanyone know why this failure would happen? -> https://launchpadlibrarian.net/117300280/buildlog_ubuntu-quantal-armhf.ubuntuone-client_3.99.92-0ubuntu2_FAILEDTOBUILD.txt.gz13:04
ralsinadobey: looking...13:07
dobeyugh, my ping sucks on speedtest.net13:07
dobeybut yay for more speed13:08
ralsinadobey: well, there is a test failing...13:08
dobeyright, but why would that test only fail on armhf?13:08
ralsinadobey: the test code seems to have nothing architecture-relevant13:08
ralsinano bit twiddling...13:08
dobeyralsina: right, though possibly a dependency might and it might have changed recently13:09
ralsinadobey: in which case it's not going to be debuggable, really13:09
ralsinadobey: what would be a armhf? I have an arm box here I could get ubuntu on...13:10
dobeyi have no idea. i guess it's the newer binary format, and it supports a7+ chips or something like that13:11
ralsinatempting if I had unlimited time13:11
ralsinadobey: how bad is this failure practically?13:11
dobeyi don't know. i'm tempted to just not run the tests on arm13:12
ralsinadobey: could be a race condition if that builder was being insanely slower than anything else13:12
dobeyyeah i think i need to ping doko. this was also only in the test rebuild of the archive, which i think does things a little differently than normal builders do13:13
dobeythe package curently in the archive seems to have built fine on all archs13:13
dobeyi wonder if we can get arm enabled for our PPAs13:15
ralsinadobey: good idea13:15
ralsinadobey: we could even get an arm jenkins if I do put ubuntu on that thing :-)13:16
dobeyi wonder if we have any arm instances available in the 'stack13:16
=== yofel_ is now known as yofel
ralsinahey, we have windows dailies again, yay13:18
briancurtinralsina: yeah i noticed they were down for almost 10 days i think :/ apparently the jenkins user token on the server was out of date (or something like that)13:20
ralsinabriancurtin: well, there were failing tests, too, which you fixed :-)13:20
briancurtinif there were 25 hours in a day, i would like to find a way to make it more noisy when it fails or is down. i wish i could hook it up to my doorbell13:21
dobeythere's an arduino for that13:22
dobeybut then you'll be checking the door every 5 minutes13:23
briancurtini still have those mini police lights that ive been meaning to hook up to an adruino. there's a video of a guy who hooked them up to his web server status tools13:23
ralsinabriancurtin: well, if your IRC client has customizable alerts, you can make it arbitrarily noisy :-)13:24
ralsinabriancurtin: specially with something like an android IRC client13:24
dobeyhook it up to a wub machine13:24
dobeywhen there's an error, your floor will drop13:24
briancurtinralsina: yeah now that it's up and running i'll catch the updates in -internal and think about making something out of them13:25
dobeyralsina: so armel/armhf builds should be enabled for our nightlies/beta/stable PPAs now :)13:41
ralsinadobey: cool!13:43
dobeyralsina: http://askubuntu.com/questions/192467/frozen-dialogue-window-in-ubuntu-one <- have you seen that before?13:44
ralsinadobey: looking13:44
ralsinadobey: no, have not seen it13:45
dobeyralsina: the issue you fixed previously was about the file picker dialog not coming up at all?13:46
ralsinadobey: don't really remember13:46
dobeyok, i recall there was an issue with that part of the UX, and you fixed something for it. thought this might be the same/related issue13:47
dobeyralsina: were you reviewing gatox's menu-separator branch?13:48
=== slank` is now known as slank
=== slank is now known as Guest84818
dobeyralsina: also, nice reply. "Any progress?" "Yes."13:53
=== m_conley_away is now known as m_conley
dobeyyay unicode bugs :(14:01
ralsinadobey: I have a +1 for that branch somewhere, just didn't apply it14:05
dobeyralsina: ah, would be a good branch to land to trigger a nightlies build :)14:06
mmccmorning, folks14:08
gatoxmmcc, hi14:09
=== mandel|lunch is now known as mandel
mmccmandel, I was looking at the operation queue in the daemon last night and I have a question - were you intending to have it spawn multiple threads to send events to the clients? (this isn't about the one dispatcher thread, it's the operation queue on which you put an NSBlockOperation for each event14:14
mmccthe default is for it to use as many threads as it can, and to run them concurrently -- so events are not delivered in the order we get them from the system14:15
mandelmmcc, yes, the idea is that it is dispatching to multiple clients the same event, right? maybe there is a bug in that idea. I wanted to dispatch to client 1 and client 2 at the same time14:16
mandelmmcc, although, now that I think about it, that should never happen14:17
mmcchmm, ok. yeah, need to look at that14:17
mmccyeah, probably won't happen.14:17
mmccright now, it uses up to ~512 threads on my machine concurrently sending events to the same client. seems like we should deliver them in the order that they come from the system - in theory a delete could move ahead of a create, or other bad stuff. especially so on a heavily loaded system14:19
mandelmmcc, yes, that sound like a bad idea14:21
mandelmmcc, also, 512 threads is a crazy number for such little work14:21
mandelalecu, the ui is back to work and the link now have the underscore, I'll be cleaning the code a little and will probably talk with them on who to land this in a ppa, how is the scope work going?14:24
dobeyhrmm, i really need to install my new hard drive and get quantal installed on it, for my workstation14:25
dobeyand figure out what to do about my unusable RAID14:25
mmccmandel: definitely. I had some extra logging in there which was slowing down each individual event, so I think that's why it kept spawning threads for new events. anyway, it's going to get serialized somewhere talking to the socket, so the queue should be serial too14:25
mmccmandel: I wonder how that socket API handles many threads calling it. we don't lock it, and it has a timeout. chances are some events are dropped when there are lots of them like that14:26
alecumandel: I'm stuck in the middle of cleanup, and fighting with what seems an issue either with me or the vala compiler :-(14:27
mandelmmcc, it would be nice to know what happens when the twisted method does not fetch events fast enough too14:27
mandelalecu, ok, let me know when you have something so I can test everything together14:28
* mandel logsout to try to use his unity branch14:28
mmccmandel: it looks like if the socket buffer fills because the consumer isn't reading fast enough, eventually a send will timeout and that'll make the daemon add the client to the dead list14:30
Laneyah, I see in the music streaming status that you get timeouts with a large library14:30
Laneyis there any actual point in contacting support?14:30
mandelmmcc, hm.. that is a problem, right? we don't want to be removed from the daemon.. or if we are, we need to send a local rescan event in the sd and reconnect to the fsevents daemon14:33
mmccmandel: yeah, I'll file a bug to fix that. need to test & propose a couple other tweaks first14:35
mandelmmcc, sounds like a decent amount of work to be done14:38
dobeyjoshuahoover, beuno: ^^ can you answer Laney ?14:39
beunoLaney, we're testing out a fix for that right now. How much music do you have?14:40
Laneyup to 1632414:40
joshuahooverLaney: beuno is the guy leading the charge to fix the timeouts so you're talking to the right guy :)14:41
Laneythere's probably some album art and stuff in that count though14:41
mmccmandel: yeah, handling heavy load might require a bit of extra thought. the stuff I'm doing now was luckily small obvious changes with a big impact :)14:41
Laneybeuno: about 87G ...14:41
mandelmmcc, awesome! I have some pause between c++ and c++ did you take a look at the review I did with the needs fixing?14:41
beunoLaney, what email do you use for SSO?14:41
LaneyI just told it to sync ~/Music14:41
mmccmandel: for instance, NSThreads need their own autorelease pool, so we were leaking basically every object we ever allocated. :\ Wish I'd noticed that earlier14:42
mandelmmcc, ouch!14:42
mmccmandel: no, I didn't see the review. what branch was it?14:42
mandelmmcc, I can't remember, you posted it very late your night a few days ago14:42
mmccmandel: yeah, urbanape reported the daemon using 15GB of memory14:43
Laneybeuno: ^14:43
urbanapesince killing it yesterday, it's grown steadily to 7GB now14:43
Laneysorry, I'm bad with pinging people14:43
mmcche was running out of disk! :)14:43
beunoLaney, cool, give me a few minutes14:43
mandelmmcc, wow!14:44
mmccurbanape: see above, definitely fixed in the next build, we will no longer leak everything. I'll get it to you stat14:44
urbanapeGlad to help14:44
mmccurbanape: many thanks for pointing it out! :)14:44
mandelurbanape, yes, nice catch!14:45
urbanapeThe benefits of a tiny SSD.14:46
beunoLaney, pm'd14:51
dobeyralsina: apply that +1 already :)14:56
Laneyah, sweet sweet streaming music14:58
* Laney gets scared and leaves15:00
gatoxralsina, alecu ?15:01
alecume (writes notes)15:01
gatoxFound a problem with the SyncMenu implementation when you want to trigger a state change from python. Fixed Bug #105620115:02
gatoxPropose branch for u1-cp shares tab. Keep working in Shares tab related issues.15:02
ubot5Launchpad bug 1056201 in Ubuntu One Control Panel "Share Tab try to reshare alreaady shared files" [Undecided,Confirmed] https://launchpad.net/bugs/105620115:02
gatoxdobey, go15:02
dobeyDONE: bug triage, sick day, poked to enable armel/armhf on our PPAs15:02
dobeyTODO: bug triage, fix bugs, prep for final release on monday15:02
dobeyBLCK: None15:02
dobeymandel: go15:02
mandelDONE: (All is preview work)Fixed bug in nux that was preventing text entries from hidding the input when using SetPasswordMode(true). Fix the mechanism to find the currently selected ui element in the preview to return the text entry. Got password input back to work. Finished styling of secundary actions (forgotten password, change payment).15:02
mandelTODO: talk with design to get a style to be used when we hover over secundary actions and when they are activated. Start cleaning code and splitting it to be sent to unity trunk.15:02
mandelBLOCKED: no15:02
mandelCOMMENTS: on a very slow connection for no known reason (I'm at the office).15:02
mandelmmcc, please15:02
mmccDONE: tried&failed to repro sync bug, fixed memory leaks in daemon15:02
mmccTODO: wrap up memory leak fix, look at op-queue and load handling15:02
mmccBLOQ: no15:02
mmccNEXT: briancurtin15:02
briancurtinDONE: poked jenkins into running, review, looking into vista issue from support15:03
briancurtinTODO: hopefully get more details on vista issue, run cloud-to-computer branch to manually test it15:03
briancurtinNEXT: ?15:03
alecuDONE: a bit of cleanup, got stuck with vala sync and closures misbehaving15:04
alecuTODO: wrap up vala lens work15:04
alecuBLOCKED: yes, with vala15:04
alecuNEXT: ralsina15:04
gatoxalecu, ralsina i already talk with charles, he sees the problem too and is working on that...... now i have this other branch for review if you have a moment: https://code.launchpad.net/~diegosarmentero/ubuntuone-control-panel/not-share-again/+merge/126474 (plus the others for u1-client)15:13
* mandel log out to try a new unity version15:14
mmccgatox, will the test in that branch work on windows with those hard coded '/'s in the paths?15:15
gatoxmmcc, where? for the last branch?15:15
mmccgatox, this one: https://code.launchpad.net/~diegosarmentero/ubuntuone-control-panel/not-share-again/+merge/12647415:15
gatoxmmcc, the / are only for the tests15:16
mmccgatox: ok, just checking. I remembered the paths caused a problem in previous tests15:18
gatoxmmcc, ok..... thx for checking..... but yes, i tested this branch on windows too and it's all green15:18
mmccgatox: cool, sorry to bug you then15:19
gatoxmmcc, nop! no problem!15:20
* gatox lunch + bank15:22
dobeyralsina: do you have that +1 for https://code.launchpad.net/~diegosarmentero/ubuntuone-client/menu-separator/+merge/126006 ? :)15:23
dobeyand i think it's time to get some lunch here. bbiab15:23
=== gatox is now known as gatox_lunch
mmccmandel, I replied & fixed up wrt your comments here https://code.launchpad.net/~mikemc/ubuntuone-control-panel/launchdaemon/+merge/12484715:27
mmccmandel: also, thanks for the thorough review15:28
mandelmmcc, awesome, looking again15:39
ralsinaHello again, sorry, unexpected lunch15:47
ralsinadobey, gatox_lunch: global +1 on the separator15:51
mmcchey, has anyone running os x 10.8 tried setting up a new SSO account ? Lisette is seeing errors when displaying the captcha, and I'm not - I'm wondering if it's the OS version15:55
mandelmmcc, I did back when I was working full time with it and had no issues16:00
mmccmandel: ok, thanks. also, looks like she's running 10.7 anyway. it fails loading PIL, but I can't imagine why16:03
mandelmmcc, hm..I had a similar problem back then.. with it not finding the libjpg in the system16:04
mandelmmcc, but this happened to me at the very very beginning of the project16:05
ralsinamandel, mmcc: it may be better to bundle our own ibjpeg :-(16:06
mmccmandel: this is a different bug. IIRC, your bug was Qt not loading its image plugins correctly, and the fix was to add PIL. this bug is that PIL doesn't work for lisette: https://bugs.launchpad.net/ubuntu-sso-client/+bug/105053416:06
ubot5Ubuntu bug 1050534 in Ubuntu Single Sign On Client "darwin: sso occasionally fails to show captcha and hangs" [High,New]16:06
mmccit can't find the _imaging.so c extension module, but I had her list the contents of the lib-dynload directory, and it's there16:07
mmcchmm, maybe this is the bug mandel saw long ago? does _imaging.so depend on libjpg…?looking16:08
mandelmmcc, I think so, I saw that but because I did easy_install pil and the lib was not there16:09
mandelmmcc, so it will fail to find *.so I cant remember witch16:10
mmccok, so _imaging.so does depend on libjpg, but we're not bundling it. so now I should figure out why it works on my system :)16:13
mandelmmcc, he, good :)16:14
mandelok, EOD for me16:16
mandelmmcc, I already gave you the +1 but left it for you to approve16:16
mandellaters o/16:16
mmccok mandel thanks16:17
mmccanyone have any insight on this unhelpful KeyError in the cloud-to-computer folders panel? https://pastebin.canonical.com/75365/16:18
mmccthis is another one from lisette, who is an excellent (unlucky) tester16:18
ralsinammcc: hmmm16:19
ralsinammcc: let me think that one 5'16:19
mmccthanks ralsina16:20
ralsinammcc: may I mention that I hate @handle_errors?16:20
ralsinammcc: is she getting a dialog about merging UDFs?16:23
ralsinammcc: ok, the log is useless, no idea where in that function it's crashing16:24
bender_rodriguezI'm using this channel for testing purposes, please ignore me.16:25
lisettteralsina: yes i did get a merge notification when trying to sync my cloud Music folder16:25
ralsinalisettte: ok16:28
ralsinammcc: then the only possble place to cause that is the call to self.backend.change_volume_settings16:28
mmccralsina: check. lisette, can you paste the syncdaemon log? I'm wondering what that stringbelongs to16:29
lisetttemmcc: which one? i got 816:32
mmcclisettte: the most recent one, probably.16:33
mmccshould be just ~/Library/Caches/ubuntuone/log/syncdaemon.log16:33
lisetttemmcc: not syncdaemon-exceptions.log?16:34
mmcclisettte: no, just the regular one16:37
dobeyarm builds are building16:46
aleculunch and errands for me16:53
dobeyman, the armel builds are going to take forever16:55
dobeybut armhf seems happy so far at least16:56
dobeyping sucks, but eh16:58
ralsinadobey: 1stworldproblems16:58
dobeyand the bw would be much faster, but for some reason they still won't let me buy any service faster than the 75/35 :-/16:59
kirklandbought a couple of albums on the u1 web interface, got an email receipt that I've been charged, and that OOPS16:59
dobeybeuno: ^^17:00
ralsinadobey: I am at my wife's office and get 92ms/2.65Mbps/0.49Mbps17:00
kirklandhiya beuno17:00
ralsinadobey: at home I get around 2x each one17:00
dobeyralsina: clearly you should go home :)17:01
ralsinadobey: including ping :-)17:01
dobeyoh, you mean 2x slower?17:01
chaselivingstonkirkland: what email address is associated w/ your u1 account?17:01
ralsinadobey: 2x longer ping, 2x more bandwidth17:01
dobeyralsina: is it a satellite dish?17:01
kirklandchaselivingston: -> PM17:01
ralsinadobey: haha, satellites are 330ms ;-)17:01
ralsinadobey: ok, correction, the ping is only 30ms because speedtest's server is hosted in my home's ISP17:02
ralsinadobey: but ping to for example is 90ms there are 50 here17:02
dobeyyou should tell google to upgrade buenos aires with gigabit everywhere17:03
=== gatox_lunch is now known as gatox
* gatox hates banks17:04
ralsinadobey: I suspect even google doesn't have quite that much money17:04
ralsinaconsidering buenos aires has roughly 25 million meters of street17:04
dobeyralsina: google has street view of the great barrier reef. I think they can spend $15 mil to run cable through a city :)17:06
ralsina  dobey: it's more like 150 mil17:12
ralsina(I am guessing like crazy here ;-)17:12
ralsinabut yes, the street view of the great barrier is awesome17:13
dobeywell, i'm guessing that buying ~25mil meteres of cable will come with a very nice bulk discount :)17:14
Chipacadobey: but then you need to import it into argentina17:16
dobeyeven at $1/m it still comes out to $25 mil just for the cable. and installation costs will probably come out cheaper. of course, all the other equipment will also be pretty costly. but probably total cost would come out much less than a couple fighter jets, and depreciated over 5 years.17:17
dobeyChipaca: i hear politicians like money. :)17:17
ralsinaJust the connectors for joining the cable are probably several million dollars.17:21
ralsinadobey: yes, you need the little thingies to connect the fiber optics to other things :-)17:22
ralsinadobey: and in this case, you need a few millions of those17:22
dobeyoh, fiber17:22
ralsinaeven if they did fiber to UTF switches per block, that's still about ... 200K switches17:23
ralsinawhich is (more wild guessing) about 3 days of the world's switch production17:23
dobeyfiber is so passe. neutrino network ftw.17:23
ralsinadobey: the problem is the million-gallon neutrino receiver ;-)17:24
dobeyralsina: great thing is, it can be anywhere on the earth :)17:24
ralsinaAlso, 25 megameters of optic fiber with a 2mm plastic jacket are 1.6 million cubic meters of fiber17:25
ralsina(decimal point may be slightly misplaced ;-)17:25
ralsinaThe right number is 157K cubic meters of fiber, or a cube 54 meters on a side17:29
mmcchmm, I need some reviews: a short one for alecu (or just anyone not ralsina) here: https://code.launchpad.net/~mikemc/ubuntuone-client/fix-default-fs-monitor/+merge/12529117:48
gatoxmmcc, alecu this branch has been updated: https://code.launchpad.net/~diegosarmentero/ubuntuone-client/ubuntuone-client-timer/+merge/12549917:48
gatoxmmcc, i'll take that one17:48
mmcccould use a second review on this one: https://code.launchpad.net/~mikemc/ubuntuone-control-panel/remote-folders-fix/+merge/12603717:48
mmccgatox, thanks17:48
mmccand this really short one needs reviews also: https://code.launchpad.net/~mikemc/ubuntuone-client/darwin-platform-platform/+merge/12634517:49
mmccback in ~15, need to get coffee. tea will only get you so far17:50
beunokirkland, hi!  I'll be right with you17:59
beunokirkland, did you by any chance cancel your U1 account at some point?18:00
chaselivingstonbeuno: we got him straightened out already :)18:00
beunoah, awesome18:01
gatoxmmcc, +1 in the first one and globally approved..... and the other one was globally approved already18:13
mmccgatox: great, thanks18:14
mmccgatox, on your client-timer branch, the DummySyncMenu needs an update_transfers function, because you pass that to aggregator.register_listener in aggregator.py. I wasn't clear in my comment earlier but I was thinking you'd just s/start_timer/update_transfers in the dummy class and its test18:14
gatoxmmcc, ahhh ok18:15
ralsinaoh, joy, launchpad 502 error from jenkins. Let's give it another chance.18:18
gatoxmmcc, done18:18
ubot5Launchpad bug 502 in Launchpad itself "Duplicate tasks shouldn't be permitted" [Medium,Fix released] https://launchpad.net/bugs/50218:18
ralsinaoh, ubot5 you are so amusing18:18
dobeyit's gangnam style18:19
mmccgatox: thanks, +118:23
gatoxdobey, LOL18:23
mmccgatox, is that the last branch you have pending for the sync_menu stuff?18:24
gatoxmmcc, nop..... y have another one18:24
gatoxmmcc, this one: https://code.launchpad.net/~diegosarmentero/ubuntuone-client/ubuntuone-client-menuorder/+merge/12576818:24
mmccgatox, ok, cool. don't think I can review that without Q though, right?18:28
mmccbut at least I can propose my other branch that fixes the common.py dummy syncmenu now18:28
gatoxmmcc, nap18:28
* mmcc could really use a nap18:28
gatoxmmcc, ah yes! let me know when you propose that....... i can do the review18:28
dobeywonder what the easiest way to set up raid 1+0 on ubuntu server is18:31
ralsinadobey: never done it on ubuntu :-(18:33
ralsinadobey: usually, that's not bootable, so you end up setting up /boot as a 4-way mirrored raid18:35
mmcclaunchpad timeout for everyone or just me?18:36
ralsinammcc: it timed out for me and jenkins a few minutes ago, then worked18:36
gatoxmmcc, working now18:37
dobeyralsina: well for these 4 disks, they're just extra data partitions, not the OS drive. I'll be setting up just plain raid mirroring for the OS drive on a couple of small SSDs18:37
ralsinadobey: ok, then just install and build the raid manually18:38
ralsinadobey: it's a few mdadms away :-)18:38
ralsinadobey: IIRC it's just creating md0 and md1 as raid1 and then a raid0 over them (or viceversa)18:39
ralsinathen mounting that last one18:39
dobeywell, i want to have 2 of them in raid 0. and the other two being the raid1 mirror of that first set18:40
dobeyso some redundancy, and ~4TB of storage18:40
ralsinadobey: sure, 1+0 is exactly the same as 0+118:41
ralsinadobey: so, just do it via mdadm after install, should take 5 minutes18:41
dobeyjust want to make sure i don't accidentally destroy my server in the process :)18:42
ralsinawell, that can *always* happen ;-)18:42
ralsinajust type v e r y c a r e ff u l l y ;-)18:42
* dobey ponders just marking all these bugs invalid18:51
mmccargh, launchpad still timing out on resubmitting a merge proposal…18:52
dobeyi wonder if mdadm accepts UUIDs19:08
ralsinadobey: yes, option -u19:17
dobeyralsina: man page says that's only for assemble mode19:23
ralsinadobey: what mode do you need uuids for?19:23
dobeyralsina: i want to avoid using /dev/sd[a-z] at all19:24
ralsinadobey: ok, so you use uuids for assemble mode19:24
dobeyralsina: my enclosure is JBOD and whenever i power it off and back on, the disk positions change19:24
ralsinathen you have new devices that have their own UUIDs19:24
dobeyralsina: --create rquires devices19:24
ralsinadobey: ok, but create is only once19:25
ralsinadobey: and then the device names don't matter because you assemble the device using uuids19:25
* ralsina is slightly foggy on mdadm finesse19:25
dobeyi don't think i can do what i want :(19:27
chaselivingstonping mmcc: file sync keeps disconnecting for me, any ideas why?19:34
mmccchaselivingston: not off the top of my head, can you send syncdaemon.logs?19:35
chaselivingstonmmcc: of course :)19:36
chaselivingstonmmcc: https://pastebin.canonical.com/75392/19:36
dobeybrb, need to take a little break19:36
mmccchaselivingston: hmm, that's all download progress updates. which I think maybe we should stop logging :)19:39
chaselivingstonmmcc: maybe so :)19:39
mmccdo you have  a syncdaemon.log.* from around the time when you got disconnected?19:40
chaselivingstonmmcc: let me look19:40
chaselivingstonmmcc: just emailed a zip your way w/ some logs19:42
mmccchaselivingston: thanks19:42
chaselivingstonmmcc: np19:43
mmccchaselivingston: it looks like you're just legitimately losing the network connection. are you on a flaky network or something?19:50
chaselivingstonmmcc: no, everything else is good...19:50
chaselivingstonmmcc: or at least seems to be19:50
mmcchmm. here's a paste of what I'm looking at -- https://pastebin.canonical.com/75396/ <--- does anyone else recognize this SSL error that chaselivingston is seeing?19:55
alecugatox: I've added a couple of small fixes needed in your branch: https://code.launchpad.net/~diegosarmentero/ubuntuone-client/ubuntuone-client-timer/+merge/12549920:12
* gatox looking20:12
alecugatox: but I've already approved it, since they are very small20:12
gatoxalecu, yes copy/paste problems mostly..... fixing20:13
alecugatox: the "listener" name change is because they don't specify what they are listening to.20:13
mmccanyone want to review a setup-mac branch? https://code.launchpad.net/~mikemc/ubuntuone-windows-installer/byo-libjpeg/+merge/12653820:21
mmccthis should fix lisette's issue with the captcha20:21
alecugatox: the menuorder branch looks mostly good20:21
mmccit bundles a copy of libjpeg20:21
ralsinammcc: got it20:21
alecugatox: I'm running the tests on both20:21
gatoxalecu, timer fixed.... waiting for launchpad to rescan it to approve20:21
ralsinammcc: if you have the qt image plugins working, then maybe it's a better idea not to use PIL20:22
ralsinammcc: OTOH, that still means bunding libjpeg so notinteresting :-(20:22
mmccralsina: heh. well, we're already bundling libqjpeg… which doesn't appear to need libjpeg20:23
ralsinammcc: hmmm.... well, then maybe it is a bit interesting ;-)20:23
mmccso if we wanted to get rid of PIL we could… but I'd want to be sure it worked on windows too…20:23
mmccand I haven't done anything with the Qt plugins on windows20:23
briancurtinmy hands could use a break from mashing the keyboard, be back in a bit (i'm not intentionally running away from PIL/Qt)20:25
ralsinammcc: on windows we already bundle libjpeg20:25
ralsinammcc: IIRC we are using PIL only in !≃ linux ... checking...20:25
mmccralsina: no, it looks like PIL is always being used20:26
ralsinaok, then let's leave that for some distant future if your current fix works20:26
mmccbut on linux qt plugins should just work, right?20:26
ralsinammcc: yes, on linux they do20:27
ralsinaon windows they don't20:27
ralsinaon macs I don't know20:27
mmccon macs they work if you put them in the place Qt expects them (and we do, now)20:27
mmccI think on windows it's the same issue, Qt is just looking for them somewhere that isn't well documented20:27
mmccand you need to have an empty qt.conf somewhere else that isn't welld ocumented, to poke it to look in the other dark place20:28
ralsinammcc: yes, it's one of those things where we just ran out of time to debug20:28
mmccyep. I got that impression from Manuel's colorful "HACK" comment :)20:28
ralsinaand that I would love if we could document and publish in a google-friendly place, because noone else knows quite how to do it, or they are not telling20:29
alecugatox:     85:  [C0111, ShareLinksTestCase.test_file_already_shared.fake_method] Missing docstring20:29
alecugatox: in the u1cp branch20:29
ralsinaWhen I solve one of those, I do this: http://lateral.netmanagers.com.ar/weblog/posts/BB997.html20:30
alecugatox: this looks dangerous:         file_path = unicode(file_path)20:32
mmccralsina: I'll make a todo to write that up. I wrote something close to a HOWTO on the python-mac mailing list a little while ago, I'll dig it up20:32
alecugatox: what's the encoding that's coming?20:32
gatoxalecu, QString to unicode20:33
mmccbtw ralsina, I think on windows it's just QCoreApplication::applicationDirPath()\plugins\imageformats\<<imagesgohere>>20:33
mmccer, <<pluginsgohere>> :)20:33
alecugatox: is that any safer?20:33
mmccalthough there might need to be an empty qt.conf in (QCoreApplication::applicationDirPath())\qt.conf to trigger it looking there20:34
gatoxalecu, yes..... move everything to use the pyqt api2 :P...... i never had problems with that...20:34
ralsinammcc: I got it working that way for sqlite, but not for image plugins http://lateral.netmanagers.com.ar/weblog/posts/BB955.html20:34
ralsinammcc: *could* be because it also means bundling libjpeg. Or not :-)20:34
mmccralsina: I don't think so, my copy of libqjpeg doesn't link to libjpeg20:35
alecugatox: also, a few lines below, there's this: full_path = os.path.expanduser(file_path)20:35
* briancurtin back20:35
ralsinammcc: hmmm... that could be because of how your qt was build20:35
gatoxalecu, yes20:35
ralsinammcc: "system libjpeg" versus "included libjpeg"20:36
gatoxalecu, what's the problem with that?20:36
alecugatox: I think we should be using the platform independent code that dobey did20:36
alecugatox: the os module is not very safe vis unicode20:36
ralsinammcc: but anyway +120:36
mmccralsina: ok, thanks20:36
alecugatox: regarding any encoding for that matter.20:36
ralsinaalecu, gatox: use os_helper wherever needed & possible20:36
alecudobey: does dirspec have anything similar to os.path.expanduser?20:37
gatoxalecu, mmmh... we will probably should replace a couple of expanduser then..... i'll do it in that branch......20:37
ralsinaalecu, gatox: dirspec.utils.user_home20:39
gatoxralsina, alecu ack.... i'll replace all the occurrences of expanduser with that.... or something related to that20:39
ralsinagatox: I think it always returns bytes, but you'd have to double check in each context if that's a good thing20:40
dobeyoh right, you could use user_home i guess20:41
mmccralsina, I think in here: http://lateral.netmanagers.com.ar/weblog/posts/BB955.html it might work if you did this: http://paste.ubuntu.com/1229251/20:42
dobeyalecu: ubuntuone-client has an expanduser wrapper within itself as well i think20:42
ralsinammcc: quite possible.20:42
dobeyuser_home doesn't really solve the unicode issues20:42
ralsinammcc: one of these days I will boot windows, and try it ;-)20:42
briancurtinanyone know why u1cp would get "No module named pkg_resources" on windows? i cant start a CP from source, not sure why when i can type "import pkg_resouces" fine in the interpreter20:43
mmccbriancurtin: hmmm. that sounds kind of familiar, let me grep my notes20:43
dobeyyeah, that20:45
dobeyyay inappropriately titled modules20:45
briancurtinmmcc: im about to rebuild the buildout without using "--distribute" to see if that changes anything20:45
ralsinaseems to be exactly the same thing as the one in dirspec except with more asserts20:45
ralsinawe should remove it20:45
mmccbriancurtin: no luck in my notes… where is the interpreter finding it? and how are you running the CP bin from source?20:46
mmccI'm not 100% sure I ever got it to run from source on windows, but I thought I had20:46
dobeyralsina: eh? there is no expand_user in dirspec though20:47
briancurtinmmcc: it finds it in the distribute egg. running "python bin\ubuntuone-control-panel-qt" (after other terminals have SSO and SD running)20:47
briancurtin(with SSO and U1C on the PYTHONPATH)20:47
dobeyralsina: though yes, we need to consolidate all that stuff, when we do the python3 work20:47
mmccbriancurtin: hmm. not sure I have any better ideas than rebuilding…20:48
gatoxralsina, alecu i have another u1-cp branch for review: https://code.launchpad.net/~diegosarmentero/ubuntuone-control-panel/mouse-events/+merge/12654420:48
gatoxand now..... EOD for me!20:49
ralsinadobey: this is basically a large wrapper around a .replace('~', user_home) and we have this kind of things way too spread20:49
briancurtinmmcc: it's quick enough to rebuild, hopefully it works20:49
dobeyralsina: right20:49
ralsinagatox: looking20:49
dobeyralsina: i don't think the solution is to do .replace(user_home) everywhere, but rather we probably need to put expand_user in dirspec.utils instead (fixing it to work on 2 and 3), and have everything use that.20:50
ralsinadobey: right20:50
ralsinaor move our os_helper into a separate project and use that everywhere20:50
ralsinasince it's a generally useful library20:50
dobeyour code needs plenty of reorg :-/20:51
ralsinadobey: less than it used to need!20:51
ralsinadobey: baby steps ;-)20:51
dobeyyes, i know. i did a lot of that work :)20:52
ralsinadobey: I think the problem is that we are trying to do those "turn this into a library" early in the cycle, and we really should do it *late*. And then, in the early cycle, just remove it from the old places and use the new library20:53
ralsinathat way we can multitask a little20:53
gatoxok..... eod here!! see you tomorrow people! bye!20:56
briancurtinmmcc: ha, rebuilding didnt work. fun times ahead.21:01
=== mmcc_ is now known as mmcc
ralsinaEOD for me, as usual will be around later21:02
briancurtincrap, this pkg_resources problem is now blocking two things :/21:22
=== m_conley is now known as m_conley_away
briancurtini have to head out of here, see everyone tomorrow21:57
dobeysame here, later all21:59
=== Guest67155 is now known as adorilson
mmcchmmm, was stress-testing syncdaemon and the fsevents daemon and I've gotten to a point where there are 200 waiting uploads and nothing currently transferring23:55
mmccas shown by u1sdtool --waiting and u1sdtool --current-transfers, respectively23:55
mmccthis seems like a bad place to get stuck…23:56
mmccthere are 68349 offloaded queue items according to the latest MARK log entry, too23:56
mmccand the network is up, it just stopped activating things to run at some point23:59

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