davecheneycherylj: ping01:03
davecheney10:03 < davecheney> cherylj: what's the status of 1.25.4 ?01:03
davecheney10:04 < davecheney> i'm building up a lot of chagnes on master that should be backported to 1.2501:03
thumperdavecheney: I think you should prepare the changes for 1.25 branch01:04
thumperthe question becomes whether to do one or many01:04
thumperdavecheney: I hear zfs works real well with lxd01:04
davecheneythumper: should I land what I have now?01:07
davecheneywhen is 1.25.4 coming out ?01:07
thumperdavecheney: I think some of these connection type issues are needing to be fixed in 1.25 too01:07
thumperso yes, I think you should be landing these changes on 1.25 too01:07
axwanastasiamac: thanks for review, please see my response to your issue before I go ahead01:08
perrito666sweeet master is bless01:08
axwperrito666: I hope you kicked off all your merges before you announced that01:09
perrito666axw: no pending merges from me :)01:09
axwperrito666: :p01:09
alexisbdavecheney, cherylj is out for the evening01:12
=== alexisb is now known as alexisb-afk
mupBug #1491399 changed: TestInstallMongodServiceExists fails on centos <centos> <ci> <precise> <regression> <test-failure> <juju-core:Fix Released> <juju-core 1.25:Triaged> <https://launchpad.net/bugs/1491399>01:17
mupBug #1494726 changed: TestGetUsesCache different mime-type centos <centos> <ci> <test-failure> <unit-tests> <juju-core:Fix Released by cherylj> <juju-core 1.25:Fix Released by cherylj> <https://launchpad.net/bugs/1494726>01:17
mupBug #1491399 opened: TestInstallMongodServiceExists fails on centos <centos> <ci> <precise> <regression> <test-failure> <juju-core:Fix Released> <juju-core 1.25:Triaged> <https://launchpad.net/bugs/1491399>01:20
mupBug #1494726 opened: TestGetUsesCache different mime-type centos <centos> <ci> <test-failure> <unit-tests> <juju-core:Fix Released by cherylj> <juju-core 1.25:Fix Released by cherylj> <https://launchpad.net/bugs/1494726>01:20
mupBug #1491399 changed: TestInstallMongodServiceExists fails on centos <centos> <ci> <precise> <regression> <test-failure> <juju-core:Fix Released> <juju-core 1.25:Triaged> <https://launchpad.net/bugs/1491399>01:23
mupBug #1494726 changed: TestGetUsesCache different mime-type centos <centos> <ci> <test-failure> <unit-tests> <juju-core:Fix Released by cherylj> <juju-core 1.25:Fix Released by cherylj> <https://launchpad.net/bugs/1494726>01:23
davecheneyalexisb-afk: ahh, thanks01:25
davecheneythumper: do we have a error.Causes checker ?01:27
davecheneyie, rather than writing01:27
davecheneyerrors.Cause(err), gc.Equals, ....)01:27
davecheneycan wr write01:27
thumperI don't think so...01:28
davecheneyerr, gc.Causes, expected01:28
davecheneyi thought you were talking about adding htat way back when01:28
thumperthere is jc.Satisfies01:28
davecheneythat checks type01:28
thumperI'm not sure if that looks at the cause01:28
davecheneywhich isn't close enough01:28
thumperpretty sure we don't01:28
thumperbut could be useful01:29
davecheneyanyway, i have got my branch to the point that we dont' have to add all that 'request error: ' prefix nonsense01:29
davecheneywhich is the whole reason params.ClientError existed -- to filter that out01:29
davecheneynow we have *rpc.RequestError as a type, properly annotated all the way back out to the caller01:29
davecheney-               c.Assert(err, gc.ErrorMatches, "permission denied")01:33
davecheney+               c.Assert(errors.Cause(err), gc.Equals, &rpc.RequestError{01:33
davecheney+                       Message: "permission denied",01:33
davecheney+                       Code:    "unauthorized access",01:33
davecheney+               })01:33
thumperbut begs for some new checkers01:34
thumperc.Assert(err, blah.IsUnauthorized)01:34
davecheneylets see how big the diff is01:36
davecheneyi dont want to change more than I have to to backport this to 1.201:36
davecheneyi dont want to change more than I have to to backport this to 1.2501:36
davecheneyyeah, this attempt is generating a much smaller diff01:51
mupBug # changed: 1497355, 1508585, 1510951, 1514877, 1534201, 1536425, 1536426, 1539656, 1542336, 1543178, 1543189, 154319301:53
mupBug # opened: 1497355, 1508585, 1510951, 1514877, 1534201, 1536425, 1536426, 1539656, 1542336, 1543178, 1543189, 154319301:56
=== natefinch-afk is now known as natefinch
mupBug # changed: 1497355, 1508585, 1510951, 1514877, 1534201, 1536425, 1536426, 1539656, 1542336, 1543178, 1543189, 154319301:59
mupBug #1538868 changed: kill-controller failed: sudo: switch: command not found <ci> <kill-controller> <local-provider> <regression> <juju-core:Invalid> <juju-core api-command-rename:Fix Released by cherylj> <https://launchpad.net/bugs/1538868>02:02
mupBug #1541473 changed: ensure-availability brings down the state-server <ci> <ensure-availability> <regression> <tech-debt> <juju-core:Fix Released by cherylj> <juju-core 1.25:Triaged> <https://launchpad.net/bugs/1541473>02:02
mupBug #1543216 changed: EOF uploading charm <charm> <ci> <deploy> <intermittent-failure> <tech-debt> <juju-core:Fix Released by cherylj> <https://launchpad.net/bugs/1543216>02:02
tych0cherylj: hrm. still hasn't run :(02:04
tych0cherylj: is there a queue somewhere i can look at?02:04
mupBug #1538868 opened: kill-controller failed: sudo: switch: command not found <ci> <kill-controller> <local-provider> <regression> <juju-core:Invalid> <juju-core api-command-rename:Fix Released by cherylj> <https://launchpad.net/bugs/1538868>02:08
mupBug #1541473 opened: ensure-availability brings down the state-server <ci> <ensure-availability> <regression> <tech-debt> <juju-core:Fix Released by cherylj> <juju-core 1.25:Triaged> <https://launchpad.net/bugs/1541473>02:08
mupBug #1543216 opened: EOF uploading charm <charm> <ci> <deploy> <intermittent-failure> <tech-debt> <juju-core:Fix Released by cherylj> <https://launchpad.net/bugs/1543216>02:08
cheryljtych0: we've been busy testing master02:09
axwwallyworld_: sorry, I forgot to look at your branch earlier. it's ready to ship now02:10
cheryljtych0: now that we're releasing alpha2, CI should pick up other branches02:10
wallyworld_axw: np, ty. i've been caught up in other things02:10
tych0cherylj: ah, ok, cool.02:10
cheryljdavecheney: we will probably be doing 1.25.4 after 2.0-beta1.  So, sometime the week of Feb-2202:11
mupBug #1538868 changed: kill-controller failed: sudo: switch: command not found <ci> <kill-controller> <local-provider> <regression> <juju-core:Invalid> <juju-core api-command-rename:Fix Released by cherylj> <https://launchpad.net/bugs/1538868>02:18
mupBug #1541473 changed: ensure-availability brings down the state-server <ci> <ensure-availability> <regression> <tech-debt> <juju-core:Fix Released by cherylj> <juju-core 1.25:Triaged> <https://launchpad.net/bugs/1541473>02:18
mupBug #1543216 changed: EOF uploading charm <charm> <ci> <deploy> <intermittent-failure> <tech-debt> <juju-core:Fix Released by cherylj> <https://launchpad.net/bugs/1543216>02:18
anastasiamacaxw: u r all good to go! HUGE thank you for this PR \o/02:19
axwanastasiamac: thanks, accounts.yaml on its way (basically just a sed command, for real)02:19
axwanastasiamac: also, your PRs are reviewed02:19
anastasiamacaxw: u r amazing \o/ tyvm!02:20
mupBug #1538868 opened: kill-controller failed: sudo: switch: command not found <ci> <kill-controller> <local-provider> <regression> <juju-core:Invalid> <juju-core api-command-rename:Fix Released by cherylj> <https://launchpad.net/bugs/1538868>02:21
mupBug #1541473 opened: ensure-availability brings down the state-server <ci> <ensure-availability> <regression> <tech-debt> <juju-core:Fix Released by cherylj> <juju-core 1.25:Triaged> <https://launchpad.net/bugs/1541473>02:21
mupBug #1543216 opened: EOF uploading charm <charm> <ci> <deploy> <intermittent-failure> <tech-debt> <juju-core:Fix Released by cherylj> <https://launchpad.net/bugs/1543216>02:21
anastasiamacaxw: about show-contoller not showing last model... would be nicer to show all models for this controller with indication of current?02:22
axwanastasiamac: I think both would be good actually02:22
anastasiamacaxw: excellent \o/ i agree :D02:23
mupBug #1538868 changed: kill-controller failed: sudo: switch: command not found <ci> <kill-controller> <local-provider> <regression> <juju-core:Invalid> <juju-core api-command-rename:Fix Released by cherylj> <https://launchpad.net/bugs/1538868>02:24
mupBug #1541473 changed: ensure-availability brings down the state-server <ci> <ensure-availability> <regression> <tech-debt> <juju-core:Fix Released by cherylj> <juju-core 1.25:Triaged> <https://launchpad.net/bugs/1541473>02:24
mupBug #1543216 changed: EOF uploading charm <charm> <ci> <deploy> <intermittent-failure> <tech-debt> <juju-core:Fix Released by cherylj> <https://launchpad.net/bugs/1543216>02:24
axwwallyworld_: I'll fix your comments in my next branch for accounts02:26
wallyworld_np ata ll02:26
axwwallyworld_: when you remove the api-info command, can you please remove ModelCommandBase.ConnectionEndpoint (if you haven't already)02:41
wallyworld_axw: i'm waiting to merge your work and then will deal with the inevitable conflicts etc02:42
axwwallyworld_: ok02:42
davecheneycherylj: ok, i'll work hard to get all my fixes backported before then02:43
anastasiamacwallyworld_: axw: list-controllers output in spec shows "-" for not-currently-logged-in... AFAIK there is no precedence, can I just leave the column cell empty or is "-" highly desired?02:46
* wallyworld_ checks the spec02:47
anastasiamacp 9 (of 13)02:48
axwanastasiamac: I kinda like "-", but prefer consistency02:48
anastasiamacaxw: we can start consistency from here, if it's preferred :D02:48
wallyworld_"-" is for currently logged in02:49
wallyworld_not not currently logged in02:49
wallyworld_and * is for default02:49
anastasiamacwallyworld_: USER and MODEL columns02:49
anastasiamacwallyworld_: the prefix - is not consistent02:50
anastasiamacwallyworld_: and list-model uses * suffix to indicate current02:50
wallyworld_"-" is for the active controller02:50
wallyworld_you can be logged into more than one02:51
wallyworld_but only one is active02:51
anastasiamacwallyworld_: axw and I *thought* *suffix would be nice for active controller too02:51
wallyworld_how to we denote the default one in that case?02:51
wallyworld_i agree that  we should be consistent02:52
axwwallyworld_: we're talking about the "-" in columns that have no value02:52
axwwallyworld_: separately, there is some inconsistency in the spec w.r.t. using - vs * for active02:52
axwI think we settled on using * as a suffix02:52
wallyworld_yes :-(02:52
anastasiamacwallyworld_: axw: http://pastebin.ubuntu.com/15013438/02:53
wallyworld_i must admit, i would like to see "-" for empty region too :-)02:53
wallyworld_when in doubt, i think we adhere to spec (use "-" for model/user) and ask for feedback02:54
anastasiamacwallyworld_: yes, I think we should b consistent from now on to show "-" for empty in cmd output :D02:54
anastasiamacwallyworld_: axw: from the pastebin above, *suffix may be obscured by long names...02:55
axwwallyworld_ anastasiamac: added AccountStore02:55
anastasiamacaxw: u r powering thru this \o/02:55
axwanastasiamac: a simple sed command02:55
anastasiamacaxw: all simple things r ingenious :P02:56
mupBug # changed: 1382274, 1468972, 1497802, 1511659, 1515289, 1533275, 1533896, 1535165, 1536378, 1537717, 1540701, 154251002:57
wallyworld_in your current work, we are moving away from separate current-controller and current-model files right?02:58
anastasiamacwallyworld_: this is what it looks like form axw's recent PRs \o/02:58
axwwallyworld_: for now we still have current-controller, but current-model will be no more02:59
wallyworld_great, i only skimmed, just wanted to check :-)02:59
axwanastasiamac: ^02:59
axwwe can add current-controller to controllers.yaml too though?02:59
wallyworld_i think so02:59
mupBug # opened: 1382274, 1468972, 1497802, 1511659, 1515289, 1533275, 1533896, 1535165, 1536378, 1537717, 1540701, 154251003:00
wallyworld_axw: although, that embeds user state in that file03:00
wallyworld_so not sure. makes sense for models file as that's a cache03:01
axwwallyworld_: true. and you'd (probably) never share accounts.yaml.03:01
wallyworld_yes, i was thinking about accounts.yaml too and think i came to the same conclusion that it's ok to put current account in there, but still am not 100%03:02
wallyworld_i can see a case for a current-ccount file03:02
axwwallyworld_: it'll need to be a per-controller file03:02
wallyworld_true, so in that case, ok as is probably03:03
mupBug # changed: 1382274, 1468972, 1497802, 1511659, 1515289, 1533275, 1533896, 1535165, 1536378, 1537717, 1540701, 154251003:03
anastasiamacaxw: wallyworld_: could we plz update blueprint with all files we are unravelling? :P03:04
wallyworld_one day03:04
wallyworld_the files are in the spec03:04
wallyworld_i don't want to duplicate info03:04
wallyworld_the blueprint should just refer to the spec03:04
wallyworld_we should have separate as-built doco03:05
wallyworld_but that's not for the blueprint, we should have that stuff elsewhere03:05
wallyworld_we're not good at doing that though :-(03:05
wallyworld_the blue print can refer to the as-built doc03:05
wallyworld_the blueprint is more to track progess the way we're using it03:06
anastasiamacwallyworld_: also to communicate with QA03:06
anastasiamacwallyworld_: this info can b the backbone of Release Notes03:07
wallyworld_there needs to be separate doc for that IMO - a proper test plan03:07
anastasiamacwallyworld_: that we tend to write *after* the effect03:07
wallyworld_file implementation detail is not for release notes03:07
anastasiamacwallyworld_: \o/ proper test plan would b amazing!03:08
anastasiamacwhen and who would do it?03:08
wallyworld_all the people we don't have :-(03:08
wallyworld_should be done as the feature is planned, post one page spec03:08
anastasiamacfile implementation detail is not but what files to expect may b of interest to QA03:08
wallyworld_sure, but that's not what we are using blueprints for03:08
wallyworld_the spec in this case has the file detail03:09
wallyworld_but it's not detailed enough for QA - we need a test plan03:09
wallyworld_wish there were more hours in a day03:10
anastasiamacwallyworld_: i agree, jeez, a lot of planning and definition needs to happen after one pager :D Test Plans, Acceptance Criteria, Task break down03:10
wallyworld_we do the task breakdown03:10
anastasiamacwish we had a dedicated resource... :)))03:10
wallyworld_yes :-(03:10
anastasiamacwallyworld_: https://www.youtube.com/watch?v=Dj3GH5myc3M03:11
wallyworld_you look nothing like a donkey03:12
wallyworld_are you saying i am an ogre03:12
anastasiamacwallyworld_: not everything needs to be read literally (just btw the lines) :P03:13
wallyworld_i know, just going along with the theme03:13
anastasiamac:)) i know u know03:13
axwwallyworld_: a smaller, easier one: http://reviews.vapour.ws/r/3823/03:17
axwso we don't need to change crap that can just be deleted03:17
wallyworld_+1 to that03:18
wallyworld_axw: there may be some nuggets in api_test that need to be retain i suspect?03:20
wallyworld_or at least re-homed03:20
axwwallyworld_: the file's not gone, the branch is conflicted so it doesn't show up on RB03:21
axwwallyworld_: will fix in a sec, look on GitHub in the mean time please03:21
axwwallyworld_: so once controllers, models, accounts are all in, what else are you waiting on?03:35
axwwallyworld_: are you doing the changes to pass controller/model name through, or am I?03:35
axwpass through to API conn opening code that is03:35
wallyworld_i was working to eliminate cache.yaml03:36
wallyworld_remove bootstrap config etc03:36
wallyworld_and not write controller addresses, accounts etc03:36
wallyworld_you could do the work to pass in controller/model first03:36
wallyworld_i am currently adding private-key to joyent03:37
* thumper headdesks03:37
wallyworld_but the credentials stuff needs to pass in the optional attr03:37
axwwallyworld_: seems like something you'd do at the same time03:37
wallyworld_ok, np, i can do do, need to look at the code again03:37
axwwallyworld_: that's why I don't want it to be a separate attr, but a flag on the attr that allows you to specify a file03:37
axwwallyworld_: I'm happy to do it, just not sure how far into it you are03:38
wallyworld_the joyent stuff?03:38
wallyworld_just started looking03:38
wallyworld_so you can do that if you want03:38
axwwallyworld_: no, the API stuff03:38
wallyworld_i have a branch that i started that has a shit tonne of changes, but many will be conflicting with your recent work; i think it just needs to be discarded03:39
wallyworld_so you could look to do that next bit while you're on a roll in the area03:39
axwwallyworld_: ok. if I can, I'll leave stuff to be done in parallel03:41
wallyworld_axw: what i can do first if you want is remove the bootstrap config stuff03:41
wallyworld_that won't conflict too much03:41
axwwallyworld_: that would be good03:41
wallyworld_axw: in the interim, did you want to add that fromfile attr to credentials schema? CI needs it03:42
wallyworld_as they like to use the private-key directly03:42
axwwallyworld_: ok, I'll take a lok03:42
wallyworld_ok, ty03:42
wallyworld_axw: damn, such a long queue waiting to land, hopefully your account store is not at the end03:45
axwwallyworld_: maybe just pull my branch down and work off that?03:46
axwwallyworld_: actually you shouldn't need it03:46
axwif you're just removing bootstrap bits03:46
wallyworld_might go a bit further depending. wil grab a bite to eat03:47
anastasiamac /me is surprised thumper's desk is still standing03:50
davecheneythumper: PTAL04:02
davecheneytotally redone and made much simpler04:02
davecheneyno reviewboard link im afraid04:03
thumperhmm... I wonder why no reviewboard04:08
davecheney'cos I closed the old review04:14
davecheneyremoved the link from the description04:14
davecheneybut rb didn't add a new one04:14
* davecheney drums fingers while some huge file is uuencoded, sent to the remote bash at a few kb / sec04:33
davecheneythumper: if this bootstrap works04:37
davecheneyare you find with me to JFDI this fix onto master04:37
davecheneythen I'll roll up a big backport patch for 1.2504:37
thumperwhich fix is this?04:37
thumperthe error stuff?04:37
davecheneythis is all the groudnwork I think we need to implement retrying on all API requests04:38
thumperdavecheney: yes04:42
wallyworld_axw: the preferIPv6 flag is used to sort the host api addresses so that ipv6 appear first because we start at the top of the list and try each in turn04:45
* thumper is off now04:45
axwwallyworld_: I understand how, but not why04:45
axwwallyworld_: if we're trying them all, why does it matter...?04:46
wallyworld_because we will take the first one we connect to i think04:46
wallyworld_connect to successfully04:47
axwwallyworld_: yep... why does that matter? :)04:48
wallyworld_not sure, but i assume this has come from stakeholders who want juju to connect to ipv6 if at all possible04:49
axwwallyworld_: I *think* it might be because api-endpoints prints out the last-connected-to address,  and someone was relying on that being v4/v604:49
wallyworld_yeah, not sure either04:49
axwwallyworld_: no rush, but joyent change is here: http://reviews.vapour.ws/r/3827/05:37
axwwallyworld_: not tested because I have no account, but there's a unit test un the cloud package05:37
wallyworld_axw: awesome, ty. have almost finished unpicking the change from my conflicted branch that i want to keep, have tests to fix. btw you can get joyent creds from cloud-city repo05:38
axwwallyworld_: cool05:38
axwwallyworld_: yeah, being lazy. I'm pretty sure it's ok.05:39
wallyworld_axw: lgtm05:48
axwwallyworld_: ta05:49
axwwallyworld_: it's a bit more than immediately necessary, but my intention is to add things like this into environschema05:49
wallyworld_yep, i like it05:49
wallyworld_fark so many failing tests to fix :-( i hate tests that compare log lines05:51
=== \b is now known as benonsoftware
wallyworld_axw: remind me - restore will be broken with the current cloud-credentials branch i think you said?05:52
axwwallyworld_: correct05:52
wallyworld_axw: we'll need to fix that if we want to get this branch into master for beta1. is that something you could look at next rather than the next round of feature work05:53
axwwallyworld_: ok05:54
wallyworld_ta, i can't recall the scope of the problem offhand05:54
wallyworld_hopefully not too major05:54
axwwallyworld_: https://github.com/juju/juju/blob/cloud-credentials/cmd/juju/backups/restore.go#L13305:55
wallyworld_ah yes that's right05:55
axwwallyworld_: I'm assigning 5 points to that, sound reasonable?05:57
wallyworld_think so, or 8 if it turn ulgy05:58
wallyworld_axw: as a drive by, could you also change the version to "delta1" so that the QA guys can target the new scripts05:59
axwwallyworld_: delta? not beta?06:00
wallyworld_axw: master is beta1, we need something different for this feature branch just as the new scripts are being tested06:00
axwwallyworld_: ok06:00
wallyworld_we will change back to beta1 prior to merge06:00
davecheney        "github.com/juju/juju/apiserver/testing"06:51
davecheney        apiservertesting "github.com/juju/juju/apiserver/testing"06:51
dimitern:) my fav so far is "gitjujutesting"06:54
davecheneywhy do we have juju/api, the client07:54
davecheneyjuju/apiserver/client ... ?07:54
axwwallyworld_: the restore-rebootstrap code is broken in general. I propose we delete it.08:07
axwwallyworld_: ... and just require people to bootstrap-then-restore08:08
axwwallyworld_: (BTW, CI is not using this code path AFAICS)08:09
frobwaredimitern, voidspace, dooferlad: I'm back in the land of the living \o/09:10
frobwaredimitern, voidspace, dooferlad: and this (http://reports.vapour.ws/releases/3596) tells me maas-spaces is blessed. \o/ \o/ \o/09:11
frobwaredimitern, voidspace, dooferlad: nice work!09:11
voidspacefrobware: double yay!09:14
dimiternfrobware, hey! welcome back! :)09:19
dimiternfrobware, oh yeah! we're merging today09:19
jamespagedimitern, frobware: some comments on the extra bindings stuff but I think it will fulfill our requirements09:43
dimiternjamespage, awesome! tyvm09:43
frobwarejamespage, thanks. I think your "is this a list" can be answered by the "Future extensions" section09:44
jamespagefrobware, ack - i see09:44
dimiternfrobware, you can't hear me apparently09:44
frobwarejamespage, I was just typing that when your ^ message arrived. :)09:44
jamespageone comment on the network-get use for extra-bindings - feels odd with just the name but that might be just me09:44
frobwaredimitern, nope. problem is definitely at my end.09:44
dimiternfrobware, I can hear you fine though09:45
TheMuemorning folks o/09:47
voidspaceTheMue: o/09:55
voidspacedimitern: so, I'm building a function that builds a map of spaces from maas name -> SpaceInfo09:55
voidspacedimitern: we *could* use that in maasEnviron.Spaces to set ProviderId as well as Name09:56
voidspacedimitern: shall I do that in my branch or leave it for a follow up?09:56
voidspacedimitern: (we still need the space name from MAAS here so we can generate a juju name)09:57
dimiternvoidspace, let's chat during standup09:59
voidspacedimitern: kk09:59
dimiternfrobware, fyi, here's how I tested: http://paste.ubuntu.com/15015446/11:07
frobwaredimitern, do you currently see any failures with what's on maas-spaces now?11:10
dimiternfrobware, yes, that's the base line - about 10 out of 50, the one with the patch is running now11:11
frobwaredimitern, can you pastebin the patch11:11
dimiternfrobware, http://paste.ubuntu.com/15015467/11:12
dimiternvoidspace, ^^ that's what you meant, right?11:12
frobwaredimitern, ty11:12
dimiternfrobware, with the patch I can see the forked processes running almost in sync11:14
frobwaredimitern, and which version of Go are you using?11:15
dimiternfrobware, on first glance - fewer failures: 7 out of 50, but in reality less than 5 due to TestLoginsDuringUpgrade - the rest are mongo panics and other similar11:16
dimiternfrobware, go version go1.6rc2 linux/amd6411:17
frobwaredimitern, so I initially tried with loop(1..5). I got one failure. What's the failure message to look for without the patch?11:19
frobwaredimitern, upgradeSuite.TestLoginsDuringUpgrade?11:19
dimiternfrobware, usually upgrade_test.go:316:11:21
dimiternfrobware, actually with the patch it's always that same line - without it occasionally lines 138 or 129 fail as well11:23
frobwaredimitern, so it fails with the patch too?11:24
dimiternfrobware, it does, but about less than half of the unpatched runs11:25
dimiterns/of/compared to11:26
frobwaredimitern, so not a slam dunk in terms of landing this before merging maas-spaces back into master...11:26
frobwaredimitern, just trying to assess the risk of landing this change vis-a-vis merging what we have11:27
dimiternfrobware, I still think landing as is ok atm11:27
wallyworld_axw: sounds ok to me, sorry i was at soccer. if it didn't work anyway, then no point keeping it12:08
voidspacedimitern: not really12:21
voidspacedimitern: just the changes lines 28-29 in that diff are sufficient12:21
voidspacedimitern: the other changes are fine too I guess12:21
dimiternvoidspace, cheers12:24
* dimitern just realized we haven't tried maas-spaces on maas 1.10 - yet another maas to setup (mostly done anyway)12:25
frobwaredimitern, voidspace, dooferlad: I plan to postpone the release note sync. any objections?12:33
dimiternfrobware, +112:33
=== lazypower|travel is now known as lazyPower
voidspacedimitern: heh, so the spaces endpoint isn't implemented in the gomaasapi test server13:09
voidspacedimitern: so looks like I'm diverting into that13:09
dimiternvoidspace, yeah, unfortunately, but it can be worked around13:12
dimiternvoidspace, like what I did for the interfaces13:12
voidspacedimitern: well, I can patch out the function - but then that code isn't tested13:12
dimiternvoidspace, the parsing can be tested given a maasObject13:13
voidspacedimitern: well yes, but it still leaves untested code13:13
dimiternvoidspace, but at this point it's better to do it properly and start tackling the piling gomaasapi changes we need13:13
voidspacedimitern: so you're saying I should implement spaces for gomaasapi?13:14
voidspaceit's not hard13:14
dimiternvoidspace, yeah, exactly13:14
voidspacedimitern: cool13:14
frobwaredimitern, so I tested a bit, with/without the patch and looping(1..5) - failure rate seems identical.13:21
frobwarevoidspace, ^^13:21
voidspacefrobware: dimitern: The change is still good, but may not be the root issue for this test.13:22
frobwarevoidspace, yeah, we should separate the two. patch good, but there's something else too.13:22
voidspacefrobware: dimitern: I haven't had a chance to look at the test, but I suspect if it's just trying to login after upgrade and it hits a discovery in progress error it need to try again (in a short attempt loop)13:22
voidspacefrobware: hmm, it's in a LongAttemptLoop13:25
voidspacefrobware: what's the full failure message please13:25
frobwarevoidspace, bear with - I'll pastebin a failure.13:26
voidspacefrobware: ah, during discovery api.Open will fail - so line 316 should be "return err" not an Assert13:26
voidspacedimitern: ^^13:26
voidspacedimitern: that should fix it13:27
dimiternvoidspace, hmm possibly, yeah13:28
voidspacedimitern: I'm pretty sure  - I hit that in writing the code/tests for limitLogins13:29
voidspacebecause we're preventing login, not returning a restricted api13:29
voidspaceyou should see very similar code in the machine_test.go for limited logins during space discovery13:30
voidspaceapi.Open returns an error until discovery is completed13:30
voidspaceand it's intermittent because it's timing related - there's only an error if the login attempt is in the short window whilst discovery is in progress13:31
dimiternvoidspace, sounds good13:33
dimiternvoidspace, and we do have another test to check that client logins are blocked while agent logins are not?13:33
voidspacedimitern: yes13:35
voidspacedimitern: the test tries to login as a user - that fails with the right error message13:35
voidspacedimitern: then it checks it *can* login as the local machine and that succeeds13:35
voidspacedimitern: then after discovery completes the client login succeeds13:35
dimiternvoidspace, sweet!13:36
frobwarevoidspace, http://pastebin.ubuntu.com/15016031/13:38
voidspacefrobware: in featuretest/upgrade_test.go change line 316 from c.Assert to "if err != nil { return err}"13:40
voidspacefrobware: it *should* pass more reliably then13:40
voidspacethe real commit should have a comment too I guess that api.Open can fail if discovery is still in progress13:41
voidspaceand it should be *with* dimitern's fix to prevent deadlock13:41
frobwarevoidspace, ok - funnily enough I just got 3 successful runs ... hence why it took a while to pastebin. :)13:41
frobwarevoidspace, oh-la-la. I'm now out of disk space, so the tests are failing... Grrr.13:54
frobwarevoidspace, looks like mongo is leaving stuff in /tmp. Approximately 5GBs worth. :(13:55
* voidspace lunch13:58
frobwarevoidspace, I need to go eat too. bbiab.13:58
dimiternfrobware, dooferlad, voidspace, guys, I've prepared the merge PR: https://github.com/juju/juju/pull/4392 and I'll leave you to your judgment, but now's the best time to get it in14:59
* dimitern goes to get some rest (off tomorrow btw)15:00
katcoericsnow: standup time15:02
=== alexisb-afk is now known as alexisb
voidspacefrobware: I say ShipIt15:32
frobwarevoidspace, sure? ;)15:32
frobwarevoidspace, the err check instead of assert makes a diff.15:32
voidspacefrobware: does?15:34
voidspacecherylj: ping15:35
frobwarevoidspace, yes. just double-checking because the tests leave /tmp/test-mgo* files around so can't run it for a long time. Just working around that atm.15:35
voidspacefrobware: cool15:35
voidspacefrobware: I can do that on master with the potential deadlock fix15:35
voidspaceafter maas-spaces lands15:35
frobwarevoidspace, agreed. but give me 20 mins, as I'm part way through this already.15:36
voidspacefrobware: it will take that long to land maas-spaces anyway I think, waiting on cherylj to give the ndo15:36
frobwarevoidspace, btw, we just got a maas-spaces cursed build.15:36
voidspace*nod even15:36
frobwarecherylj, we have a both a blessed and cursed maas-spaces build. Choose your poison. :)15:38
voidspacecherylj:  frobware: leadership (uniter) tests failed in attempt 291 but succeeded in 29215:38
voidspace(trusty unit tests)15:38
voidspacethat was the only failure15:39
alexisbfrobware, cherylj are we need a CI on MAAS Spaces?15:49
voidspacealexisb: we had a bless15:49
frobwarealexisb, we've had two results today. 1 blessed, and just recently 1 cursed.15:49
voidspacealexisb: then a curse due to a flakey test15:49
frobwarealexisb, personally, I'm big fan of the first result.15:50
alexisbcherylj, sinzui are there any high priority branches we are needed to test asap?15:50
alexisbfrobware, heh15:50
sinzuialexisb: upgrade-mongodb315:50
alexisbsinzui, past that run can we queue up lxd-container-type for tych0 ?15:50
sinzuialexisb: we are retesting maas-spaces which had i386 and ppc64el failures.15:50
sinzuialexisb: oh, well, it is actually in the priority list after mongo315:51
alexisbsweet, thanks15:51
frobwarealexisb, sinzui: am I hearing we're waiting for another blessed build?15:52
sinzuifrobware: yes.15:52
frobwareboo said the crowd15:53
frobwaresinzui, is this some form of majority vote?15:53
sinzuifrobware: no. there are voters and non-voters. ppc64el unit tests vote. Ci will willing to test 3 times...juju is that bad on ppc64el. But maas-spaces might need 5 or 6 tries to pass15:55
frobwaresinzui, I meant in terms of status overall. this morning maas-spaces was blessed.15:56
dooferladvoidspace: can I pick your brains for a moment?15:56
voidspacedooferlad: sure15:57
sinzuifrobware: yes, but dimitern merged another change and now we don't have a bless. also. I forced a retest of ppc as you were waking up to get that bless15:58
dooferladvoidspace: hangout?15:58
voidspacedooferlad: sure15:58
voidspacedooferlad: give me ten minutes!15:58
voidspacedooferlad: send me a hangout link - just about to phone the bank15:58
dooferladvoidspace: no problem. Ping me when you are ready. Was going to use https://plus.google.com/hangouts/_/canonical.com/juju-sapphire?authuser=015:59
frobwaresinzui, which change? this one? http://reviews.vapour.ws/r/3829/15:59
sinzuimaybe http://reports.vapour.ws/releases/3597 points to https://github.com/juju/juju/commit/f4de48d033ce30be495756492ba3ab726a3ec68f16:01
sinzuifrobware: I am going to restart the host and retest16:01
frobwaresinzui, ack16:01
frobwaresinzui, is ppc64 using gc or gccgo?16:08
sinzuifrobware: gccgo-go16:09
frobwaresinzui, ok I temporarily have access to h/w - will try running the unit tests.16:09
sinzuifrobware: you have permanent access to the stilsons CI uses.16:09
voidspacedooferlad: frobware: alexisb: my daughter just had a fit (febrile convulsion - she's had them before)16:10
frobwaresinzui, heh. yes, I keep forgetting about that...16:10
voidspacedooferlad: frobware: alexisb: need to sign out for a bit16:10
sinzuifrobware: you have access to every CI host you need to do you job.16:10
voidspacealexisb: I'll probably miss our meeting tonight :-(16:10
alexisbvoidspace, take the time you need16:10
voidspacejust called for ambulance16:10
alexisbnot a problem16:10
dooferladvoidspace: sure.16:10
alexisbmy thoughts are with you guys16:10
frobwaresinzui, so... what can I do on those machines? clone my juju repo, build, et al?16:13
sinzuifrobware: those machines are on canonical's network. they are very restricted. I scp files up and run them on the host.16:16
natefinchgood luck voidspace16:16
frobwaresinzui, and cross-compile from x86?16:25
sinzuifrobware: no cross-compile. the compiler on ppc64el is not the compiler you have on amd64. We do everythign native. debs are created on the ppc. in the case of the unit-tests. we make a release tarfile, untar it on the host and run the unit-tests. the host has go ready or it can be installed. I think we need to use stilson-07 now because 08 is dedicated to go1.5 testing16:29
frobwaresinzui, ok. on the h/w I'm using I was using gcc-go-4.916:29
sinzuigo version16:30
sinzuigo version xgcc (Ubuntu 4.9.3-0ubuntu4) 4.9.3 linux/ppc64le16:30
voidspacenatefinch: thanks16:32
frobwaresinzui, I have:16:32
frobwarego version go1.4.2 gccgo (Ubuntu 5.2.1-22ubuntu2) 5.2.1 20151010 linux/ppc64le16:32
voidspacefrobware: alexisb: dooferlad: she's ok, she has these fits when she gets a high temperature (inherited from her mother unfortunately)16:32
frobwarevoidspace, all resolved? hope so...16:33
voidspacefrobware: alexisb: dooferlad: but I have to go to hospital with her now (they always take her in after them)16:33
voidspacefrobware: she's fine, she won't stay overnight16:33
voidspacebut I have to go :-(16:33
natefinchvoidspace: go go!16:33
frobwarevoidspace, sure16:33
alexisbvoidspace, I am glad it sounds like she is ok, take care16:33
frobwaresinzui, are the CI machines running wily?16:41
frobwaresinzui, for ppc6416:41
sinzuifrobware: trusty except for one wily that is not running any unit tests16:41
frobwaresinzui, ok, I suspect that explains the gcc go version difference.16:42
perrito666Bbl lunch16:54
mupBug #1544662 opened: cannot bootstrap with mongodb3 and manual-provider <bootstrap> <manual-provider> <mongodb> <juju-core:Incomplete> <juju-core upgrade-mongodb3:Triaged> <https://launchpad.net/bugs/1544662>17:25
mupBug #1544662 changed: cannot bootstrap with mongodb3 and manual-provider <bootstrap> <manual-provider> <mongodb> <juju-core:Incomplete> <juju-core upgrade-mongodb3:Triaged> <https://launchpad.net/bugs/1544662>17:28
mupBug #1544662 opened: cannot bootstrap with mongodb3 and manual-provider <bootstrap> <manual-provider> <mongodb> <juju-core:Incomplete> <juju-core upgrade-mongodb3:Triaged> <https://launchpad.net/bugs/1544662>17:31
mupBug #1544662 changed: cannot bootstrap with mongodb3 and manual-provider <bootstrap> <manual-provider> <mongodb> <juju-core:Incomplete> <juju-core upgrade-mongodb3:Triaged> <https://launchpad.net/bugs/1544662>17:35
mupBug #1544662 opened: cannot bootstrap with mongodb3 and manual-provider <bootstrap> <manual-provider> <mongodb> <juju-core:Incomplete> <juju-core upgrade-mongodb3:Triaged> <https://launchpad.net/bugs/1544662>17:41
mupBug #1526063 changed: TestNoContextWithLock <ci> <intermittent-failure> <test-failure> <juju-core:Invalid> <juju-core maas-spaces:Fix Released> <https://launchpad.net/bugs/1526063>17:56
mupBug #1544694 opened: [maas-provider] Juju is incorrectly  configuring os bridge interfaces <juju-core:New> <MAAS:Invalid> <https://launchpad.net/bugs/1544694>18:32
natefinchman, our pseudo transactions make separating the business logic from the persistence layer nearly impossible.18:41
natefinchericsnow: did you have ideas on how to increment the CharmModifiedVersion in the same transaction as SetResource?  I can do it the horrible hacky way, where resource persistence now has intimate knowledge of the service collection's info... not sure what my other options are, besides not having them in the same transaction, which seems very likely to cause bugs if something only half-succeeds.18:50
ericsnownatefinch: gah, yet again mongo transactions are making us leak persistence info into state :(18:52
ericsnownatefinch: I do have a plan18:52
natefinchericsnow: yeah, I hate that about our transactions....18:53
ericsnownatefinch: consider adding a "NewServiceCharmModifiedOps()" method to resource/persistence.PersistenceBase18:57
ericsnownatefinch: then the adapter in component/all/resource.go must provide that, exposing a similarly named helper in corestate18:57
ericsnownatefinch: or just in state/resources.go, since that is the point of that file :)18:58
ericsnownatefinch: (the point: avoid exporting more persistence concerns out of the state package)18:59
natefinchericsnow: ahh, yeah, ok.  I was trying to figure out how to get the left hand to talk to the right hand.  I was looking at state/resources.go and trying to figure out how to do it there.  Iget it.18:59
ericsnownatefinch: cool18:59
ericsnownatefinch: let me know if you get stuck on that or have any other questions19:00
katcoericsnow: natefinch: merge of master into the feature-branch just succeeded. may want to rebase19:01
ericsnowkatco: nice!19:01
natefinchkatco: huzzah!19:01
natefinchkatco: thanks so much for taking that on.19:01
katconatefinch: happy to so you guys can keep kickin arse :)19:01
ericsnownatefinch, katco: +119:02
katcoericsnow: natefinch: cherylj: we should see if we can get a CI run through by tomorrow so we can merge INTO master19:02
cheryljkatco: well, we're trying to merge maas-spaces now, so there will be one less branch to compete with for CI runs :)19:02
katcocherylj: :) is there any way to guarantee a run?19:03
cheryljlet me send you my paypal link19:03
katcocherylj: oh man if i had known this is how this works...19:04
cheryljwe were mostly focusing on master until we got the bless yesterday.19:04
cheryljmongo3 was up next, then lxd-container-type, then I think your branch19:04
mupBug #1544694 changed: [maas-provider] Juju is incorrectly  configuring os bridge interfaces <juju-core:New> <MAAS:Invalid> <https://launchpad.net/bugs/1544694>19:05
natefinchkatco, cherylj, alexisb: team meeting?  Or should we skip it?19:06
katcocherylj: sweet! is there somewhere i can view the queue?19:07
katconatefinch: in another meeting atm, so i can't attend19:07
cheryljI repeat19:28
cheryljalexisb, frobware ^^19:29
* natefinch gets the feeling from the all caps, that this is a big deal.19:30
cheryljYES IT IS19:30
* lazyPower throws confetti19:30
lazyPowernatefinch i have to confetti if you have fanfare under control :D19:30
katcocherylj: woo! grats to everyone19:35
* katco has a sinking feeling that she had better merge master again19:35
cheryljkatco: yes :)19:36
* katco cries into her coffee19:36
katco9 conflicts. here we go again.19:37
lazyPowerheyyyy cherylj19:41
lazyPowerdoes juju 2.0 alpha bidniss work with maas 1.8? or do i need to upgrade to 1.9?19:42
alexisblazyPower, you will need to upgrade to 1.919:44
lazyPowerah, well that explains the weird issue i'm seeing about series mismatches19:45
* lazyPower grabs a pick and returns to the maas saltmines19:45
natefinchericsnow: am I correct in assuming that a staged resource that is being activated will never be pending?19:58
* ericsnow thinks about it19:59
ericsnownatefinch: pending resources should be staged as well, for exactly the same reason we stage non-pending resources20:05
ericsnownatefinch: however, I think the staging code isn't doing it right (there's a slight race)20:06
ericsnownatefinch: in resource/persistence/mongo.go, stagedID() should accommodate pending IDs20:09
ericsnownatefinch: that isn't a problem now, but could be in the future20:10
katcoericsnow: hey can you hop into moonstone?20:19
* frobware looks through the scrollback and HEADS TO THE BAR! :-D20:43
* frobware ponders... maas-spaces has landed. Gravitational waves breakthrough. Surely just a coincidence.20:47
mupBug #1544724 opened: repeatedly checks /dev/fd0 when it doesnt even exist <juju-core:New> <https://launchpad.net/bugs/1544724>20:59
perrito666well, that was something from the past21:02
mupBug #1544724 changed: repeatedly checks /dev/fd0 when it doesnt even exist <juju-core:New> <https://launchpad.net/bugs/1544724>21:02
katcofrobware: you need to keep these jokes coming.21:02
mupBug #1544724 opened: repeatedly checks /dev/fd0 when it doesnt even exist <juju-core:New> <https://launchpad.net/bugs/1544724>21:14
natefinchman, I am terrible at bash21:24
perrito666natefinch: good that ensures you are not going around embedding bash in go21:24
natefinchericsnow: btw, I think there's a bug in how we load resource information... when you deploy a local charm, the resource info defaults to says it's a charmstore resource, which is not correct21:28
ericsnownatefinch: yep21:29
natefinchahh man, the uniter tests are so slow.....21:39
natefinchsigh... that feeling when I forget to filter out all the logging, so I can't even tell what failed...21:42
natefinchOMG these tests are so bad21:43
thumperfwereade: ping?21:56
tych0sinzui: alexisb: ok. i think i've updated. i basically just merged master and cherry picked a bunch of old commits on top because git didn't figure it out. i'd prefer a rebase, but it's not obvious to me how to accomplish that with juju's feature branch setup. https://github.com/juju/juju/pull/4396 is the PR21:58
=== natefinch is now known as natefinch-afk
katcoericsnow: can you please review natefinch-afk 's branch so we can possibly land that tonight before the ci run?22:09
ericsnowkatco: sure22:09
katcoericsnow: ty... i'll try and weigh in if i can get this merge through, but feel free to give it a ship it w/o me22:10
alexisbtycho, you can just merge the rebase into your feature branch22:15
alexisbonce it is merged we can just re-queue for CI22:16
katcotych0: generally you don't need a review for merging master into your feature branch22:17
katcotych0: rebases are nice, but rebasing a feature branch on top of master is a headache, so most people just merge master in22:17
tych0ok. so nobody will be offended if i $$merge$$ myself?22:17
katcotych0: nope, not if it's a feature branch22:17
katcotych0: and it's just a merge from master22:17
tych0well, it's a merge from master + some fixups22:17
tych0which are mostly just cherry-picks of previously reviewed commits because git sucks :)22:18
katcotych0: were the fixups reviewed at one point?22:18
katcotych0: yeah, you're fine :)22:18
tych0katco: cool, thanks.22:18
katcotych0: maybe do those in a separate pr next time, but not a huge deal22:18
* katco is currently merging from master as well22:18
tych0well, i can't because the resulting merge won't build without them22:18
katcoah i see22:18
menn0thumper: "juju migrate" is here: http://reviews.vapour.ws/r/3839/22:44
katcoericsnow: natefinch-afk: merge from master just landed22:56
ericsnowkatco: k, thanks!22:56
katcoericsnow: natefinch-afk: we're also last in the queue to get a ci bless, so have a bit of time to land patches.22:59
tych0looks like my merge failed, but i'm not entirely convinced the errors are mine: http://juju-ci.vapour.ws:8080/job/github-merge-juju/6362/console23:35
tych0katco: cherylj: thoughts? ^23:35
cheryljah, the dreaded bad record MAC23:36
cheryljjust retry the $$merge$$23:36
davecheneycherylj: was there a successful CI run last night ?23:36
cheryljdavecheney: yes, we got a bless on master23:36
davecheneyi want to make sure the logging and rpc changes I have made so far are blessed before I backport them to 1.2523:36
davecheneyexcellent, excellent, chrry-pick coming right up23:36
cheryljtych0: yeah, looks like the bad record MAC was the only problem, just retry the $$merge$$23:37
davecheneyjust the uusual mongodb is web scale problem23:37
davecheneyignore it23:37
tych0ok :)23:38
davecheneyoh my god23:40
davecheneywhy is the apiserver code base like an episde of hoarders23:41
davecheneyit's got every line of code ever written in thre23:41
davecheneynothing is every thrown away23:41
davecheneyeven the apiserver has a client, even though there is another client in another package23:41
davecheneythis is like buying one of every color in the store so they don't feel lonely23:41

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