[00:12] yesterday cherylj help me fix a hung jujud process. I'm wonder how to go about debugging that if it happens again. Is there any info that can be extracted by attaching gdb to the pid? [00:16] Bug #1579059 changed: MainSuite.TestFirstRun2xFrom1x fails on windows

[00:16] Bug #1579173 changed: wily configSuite.TestNewModelConfig test failure lxd

[00:22] a123: https://github.com/juju/juju/wiki/pprof-facility [00:23] davecheney: great. thanks [00:51] thumper: https://bugs.launchpad.net/juju-core/+bug/1580391 [00:51] Bug #1580391: juju/errors tests do not pass with go 1.6 [00:51] looks like this never worked [00:51] maybe I should delete the feature === natefinch-afk is now known as natefinch [01:01] Bug #1580391 opened: juju/errors tests do not pass with go 1.6 [01:11] Bug #1580391 changed: juju/errors tests do not pass with go 1.6 [01:17] Bug #1580391 opened: juju/errors tests do not pass with go 1.6 [01:19] easy review http://reviews.vapour.ws/r/4800/ [01:21] https://bugs.launchpad.net/juju-core/+bug/1580400 [01:21] Bug #1580400: cmd/juju/charmcmd is untested [01:31] davecheney: I can look at errors if you like [01:32] Bug #1580400 opened: cmd/juju/charmcmd is untested [01:44] Bug #1580400 changed: cmd/juju/charmcmd is untested [01:47] Bug #1580400 opened: cmd/juju/charmcmd is untested [02:21] davecheney: it did work... [02:22] davecheney: however, I think I'd rather remove it... [02:22] davecheney: it is a bit bollocks IMO [02:26] thumper: it probably did work [02:26] but it stopped working a looooooooooooooooong time ago [02:26] and nobody noticed [02:27] well... 1.2 [02:29] yup, so when I said "loooong time ago" and "nobody noticed", those statements are still correct :) [02:32] davecheney: o/ [02:33] anastasiamac: hello! [02:35] davecheney: what, 1.2 has only been out of date for 23 months [02:35] davecheney: r u coming? [02:36] anastasiamac_: where ? [02:36] davecheney: it's not like we can just build a binary that'll work without the right version of go installed [02:36] anastasiamac_: coming [02:37] davecheney: the meeting u've acepted :D sent u a link into pm [02:37] anastasiamac_: i'm in the hangout [02:37] nobodyu is talking [02:50] Bug #1580417 opened: Allow a different VPC ID to be used for hosted models [03:00] i don't *think* i want to put go 1.2 in my ~gophers ppa :) [03:08] mwhudson: that'll be a paddling [03:13] wallyworld: we need yet another endpoint in clouds.yaml: https://bugs.launchpad.net/juju-core/+bug/1580423 [03:13] Bug #1580423: provider/azure: add identity endpoint to clouds.yaml [03:20] Bug #1580423 opened: provider/azure: add identity endpoint to clouds.yaml === rodlogic is now known as Guest70790 [05:29] axw: osrry, missed ping, yeah saw bug [05:30] Bug #1580418 opened: Cached local charms should be deleted when their service is removed

[05:49] thumper https://github.com/juju/errors/pull/25 [05:57] davecheney, wallyworld or axw: easy fix for one of the blockers: http://reviews.vapour.ws/r/4804/diff/ [05:57] took me ages to get my windows VM working again to be able to repro it though [05:57] and quite a while to confirm the problem [05:57] lgtm [05:57] thanks :-) [05:58] wallyworld, axw: that was fast. thanks :) [05:58] np === rodlogic is now known as Guest52700 === frankban|afk is now known as frankban [08:01] menn0 fwereade_: tech board [08:10] hmmmm... when I install XChat-Gnome it still has the config from the lost XChat [08:10] although under Xenial XChat-Gnome looks even worse than XChat did [08:11] jam: ping [08:12] frobware: pong [08:12] jam: want to sync after standup? [08:12] frobware: sure [08:14] * frobware notes that swap days do lead to a huge accumulation of unread email... [08:24] Bug #1580485 opened: juju/tools/lxdclient: TestUsingTCP and TestUsingRemoteTCP not isolated from the host and failing

[08:29] voidspace: ping [08:30] thumper: o/ [08:31] thumper: is this even implemented: `juju add-user --share model username` ? [08:31] error: flag provided but not defined: --share [08:32] hmm apparently not [08:37] thumper: pong [08:41] dimitern: something like that should work [08:54] Bug #1580497 opened: juju2 on maas permanent " connection is shut down" msg and loss of connection [08:54] Bug #1580501 opened: cloudimg-base-url parameters not in Juju2 anymore === rodlogic is now known as Guest73940 [08:56] dimitern: babbageclunk: frobware: last of the test changes for AllocateContainerAddresses [08:56] http://reviews.vapour.ws/r/4805/ [08:56] a nice easy one to review :-) [08:58] thumper: I figured it out - first `juju add-user --models admin --acl read bob Bob` then running the suggested `juju register ...` [08:58] voidspace: looking [09:01] :w [09:02] dimitern: standup? [09:02] the plot thickens: ERROR storing charm for URL "cs:ubuntu-0": delegatable macaroon is not obtainable using admin credentials [09:02] voidspace: omw, sorry [09:04] jam thumper fwereade_ menn0: just stumbled across http://opentracing.io/ [09:04] looks interesting, I'll dig a bit [09:05] axw: that does look interesting [09:12] voidspace: reviewed [09:12] dimitern: thanks [09:34] jam: standup running over [09:35] frobware: k [09:35] not unexpected :) [09:41] morning [09:53] axw: interesting. though looking at it only 3 actual implementations and 2 of those are private [09:53] axw: appdash has direct support, and zipking has private support, lightstep has a private beta === akhavr1 is now known as akhavr [10:31] jam: I can see you on the hangout, not sure if you can hear me though. === akhavr1 is now known as akhavr [10:42] * dimitern is loving the $JUJU_HOME/aliases support! [10:43] juju bootstrap dimitern aws/us-east-1 --upload-tools -> juju bud aws/us-east-1 [10:51] frobware: I meant to ask, now you're back - any idea why I get -ve times from strace -cf? [10:51] frobware: http://pastebin.ubuntu.com/16357965/ === akhavr1 is now known as akhavr === rodlogic is now known as Guest74633 [11:30] dooferlad: ha, running the tests with tmpfs is heaps faster, unsurprisingly - the expensive bits in mongo 3.2 (index creation and DB teardown) are only 10x as slow then. [11:31] babbageclunk: :-) [11:46] could anyone review my pr ? http://reviews.vapour.ws/r/4809/ === rodlogic is now known as Guest32358 === akhavr1 is now known as akhavr [12:31] jam, fwereade__, frobware, cherylj ping [12:32] alexisb: omw === akhavr1 is now known as akhavr [12:48] hoenir: hi - I'm looking at it === rodlogic is now known as Guest63806 [13:25] Bug #1276444 changed: juju switch -l should look at jenv files to add to available environments

[13:25] Bug #1282217 changed: Specifying bootstrap-host requires editing environments.yaml

[13:34] Bug #1276444 opened: juju switch -l should look at jenv files to add to available environments

[13:34] Bug #1282217 opened: Specifying bootstrap-host requires editing environments.yaml

[13:37] Bug # changed: 1276444, 1282217, 1564511, 1580314 [13:38] hoenir: reviewed [13:40] Bug #1564511 opened: cmd/jujud/reboot tests fail with lxd container running

[13:40] Bug #1580314 opened: MainSuite.TestFirstRun2xFrom1xOnUbuntu Exec function not called on windows

[13:45] babbageclunk, thanks, found a bug , need to work on it more [13:55] Bug #1564511 changed: cmd/jujud/reboot tests fail with lxd container running

[13:55] Bug #1580314 changed: MainSuite.TestFirstRun2xFrom1xOnUbuntu Exec function not called on windows

[13:57] katco, morning [13:57] alexisb: morning [13:58] katco, I need someone to pickup this bug today: https://bugs.launchpad.net/juju-core/+bug/1580184 [13:58] Bug #1580184: Timeout in github.com/juju/juju/apiserver/service on windows

[13:58] alexisb: ack. we'll see who's doing what and get it done [13:58] katco, thank you [13:58] alexisb: np === rodlogic is now known as Guest68017 [14:25] fwereade__: interesting thought.... is there a way to list all machines from all models on a controller? [14:26] fwereade__: just got an AWS alert for high usage, and wanted to see how many machines I was using from my aws controller === akhavr1 is now known as akhavr [14:43] Bug #1580626 opened: provider/openstack: Don't use a hard-coded clock.WallClock in deleteSecurityGroup

[15:14] perrito666, ping [15:14] alexisb: pong [15:15] heya perrito666 I am available early for our 1x1 if you are also available [15:15] I am [15:15] usual channel? [15:16] yep [15:16] already there [15:31] natefinch: any pointers on how to reproduce races? [15:32] redir_: go test -race ? :) [15:32] redir_: link me to the bug? [15:35] natefinch: https://bugs.launchpad.net/juju-core/+bug/1580186 [15:35] Bug #1580186: Race in github.com/juju/juju/worker/instancepoller/aggregate_test.go

[15:36] natefinch: they seem to hit it on every CI run, but I can't make it happen here. [15:36] bbiab [15:36] redir_: k === akhavr1 is now known as akhavr [15:50] fwereade__: hey sorry, i have to move on from that PR and pick up another blocker. i'll circle back when/if i get some free time [15:53] natefinch, just fyi, for the blocker bug you picked up... if you need a repo case sinzui would be happy to provide access to the machine where the test is failing consistently [15:53] alexisb: consistent repros are my favorite thing in the world [15:54] :) === akhavr1 is now known as akhavr [16:12] ericsnow: hey i'm ocr today. what was the impetus for http://reviews.vapour.ws/r/4812/ ? [16:13] ericsnow: looks like maybe so we can stop consuming the loggo package just for a testing utility? [16:14] katco: it's a precursor I had kicking around for a follow-up cleanup patch (itself to be followed by a patch that may be useful for the syslog work) [16:14] ericsnow: ahh [16:14] ericsnow: well looks like you have a shipit at any rate [16:14] bbl [16:14] katco: trying to avoid moving code around in the same patch where I change stuff :) [16:14] ericsnow: excellent idea :) [16:24] katco: would like a review on http://reviews.vapour.ws/r/4801/ if you get a chance [16:24] natefinch: tal === frankban is now known as frankban|afk [17:04] ericsnow: errr... not even sure my comment to you went through, but if so, it looks like it was a connection issue on my end [17:05] katco: apparently it didn't go through as I don't know what you're talking about :) [17:05] ericsnow: ha. i was complaining that review board was acting up again [17:05] katco: great :( [17:05] ericsnow: then my irc bouncer kicked me from all the channels at once and i wised up :) [17:06] ericsnow: no, like i said i'm pretty sure it was just a connection problem on my end [17:06] katco: oh, phew :) [17:11] natefinch: you have a review. mostly just some questions [17:11] katco: cool, thanks [17:53] Bug #1580714 opened: ubuntu user not in 'lxd' group on xenial instances [18:23] Bug #1580717 opened: Juju agent stuck in pending if charm storage's minimum-size exceeds available disk space on machine [19:27] oh boy, interesting.... package launchpad.net/tomb: cannot download, http://bazaar.launchpad.net/~niemeyer/tomb/trunk/ uses insecure protocol [19:35] natefinch: i think he moved tomb over to github didn't he? [19:36] natefinch: https://gopkg.in/tomb.v2 [19:36] katco: we still seem to use launchpad.net/tomb somewhere in juju though [19:36] natefinch: time to standardize :) [19:37] katco: actually, we don't use the gopkg.in version at all [19:39] we could standardize oooor, we could make a script that uses launchpad, github or gopkg randomly [19:39] super fun to debug [19:39] only 3 deps in juju on launchpad, and 1 already exists in github.... [19:40] niemeyer3: do you know if there's a launchpad-side fix possible for the "package foo: cannot download, uses insecure protocol" error? Is there something the project owner can do to enable https? [19:41] natefinch: i believe he is on a plane [19:41] katco: ahh boo [19:42] katco: well if he can tweet a picture in flight he might very well have irc :p [19:42] lol true [19:42] perrito666: that's an excellent point. where are his priorities! ;p [19:44] heh: C:\Users\Administrator\src\github.com\juju\juju\apiserver\service>go test [19:44] 2016-05-11 22:44:03 WARNING juju.testing mgo.go:186 failed to start mongo: exec: "/usr/local/bin/mongod": file does not exist [19:45] ahhh how fun [19:45] natefinch: beware, that might be a red herring [19:45] I recall an error somewhere having the path hardcoded even though it was looking in the right place [19:45] but I am pretty sure I fixed all of those [19:46] perrito666: the windows tests require mongo? [19:46] natefinch: some of them [19:46] perrito666: we don't support windows as a controller [19:46] and if you are using the instructions provided by cloudbase you should have it installed [19:46] perrito666: I thought I was, but their instructions either have changed or I was looking at the wrong page [19:47] natefinch: well no, but our unit tests are little unit-ish and need a mongo to try the most elemental things [19:47] perrito666: it really boggles my mind that we don't actually support running mongo on windows or controller code running on windows... but the tests assume it works ok [19:48] natefinch: the tests run a fake controller [19:48] that sort of works [19:48] http://wiki.cloudbase.it/juju-testing [19:48] http://wiki.cloudbase.it/juju-testing#getting_a_ssl_enabled_version_of_mongodexe [19:48] (╯°□°)╯︵ ┻━┻ [19:48] to be accurate [19:48] yeah, I was there... I guess I just didn't scroll down all the way before [19:49] man, I can't get any of their powershell to run on this machine... I don't know what I'm doing wrong, if it's a different version of powershell or what [19:50] oops, well, that one was a typo [20:03] Bug #1564577 changed: Add MAAS 2.0 support in Juju 2.0

[20:03] perrito666: ping? [20:05] babbageclunk: sorry was having afternoon tea [20:05] how can I help you? [20:06] alexisb said that you knew about Mongo - I'm looking at bug 1573294 [20:06] Bug #1573294: state tests run 100x slower with mongodb3.2

[20:06] perrito666: just wanted to ask if you had any advice or suggestions. [20:07] perrito666: Basically digging into it it seems like everything is slower in Mongo3.2. [20:08] babbageclunk: you have to be kidding me [20:08] what could have changed since early versions of 3 :( [20:09] babbageclunk: that is very sad news, 2.x is EOL [20:10] perrito666: Well, from rereading the bug it sounds like it's really the wiredtiger storage engine that's the problem [20:11] babbageclunk: it is, but mmapv1 is also EOL [20:11] perrito666: Ah, that was what I was about to ask. [20:11] babbageclunk: it sounds almost as if mongo sucked :p [20:14] perrito666: Maybe it's because the tests are essentially sequential, while WT is more tuned for concurrent writes? [20:15] I reached a similar conclusion that the one in the bug, michael was nice enough to put data behind that [20:15] is it possible to have multiple lxd hypervisor as resources under one juju lxd controller? [20:17] babbageclunk: that is an interesting theory, I also suggested that the fixed cost of bootstraping stuff in WT was high and since our tests seem to do that a lot that causes the pain [20:18] babbageclunk: I wonder if we can design a few cheap benchmarks to see how can we better use wt? [20:18] perrito666: I've hacked out the index creation in the tests, and while that cuts out some of the cost they're still much slower under 3.2. [20:19] mwhudson suggests that he did a deep trace of the db run, do we have numbers of there are the other 10X? [20:22] perrito666: I was running it under strace, but that was giving nonsense numbers (negative call times in the child processes). [20:22] perrito666: I've hacked up something that can get the same numbers from perf trace, and it lets me split out the syscals in mongo from the ones in our test code. [20:23] perrito666: Actually, would you be able to run the strace command for me to see if you get the same weird numbers? [20:23] certainly [20:23] lets go priv [20:28] natefinch: Hey [20:28] natefinch: Sorry, what's the question again? [20:29] niemeyer3: go get as of 1.5? definitely 1.6 will refuse to download from launchpad without the -insecure flag, because it goes over http, not https [20:30] niemeyer3: sorry, gotta run, but will read responses. No rush, just wondering if there was a way for a package author on launchpad to enable https to avoid that problem [20:33] Bug #1564577 opened: Add MAAS 2.0 support in Juju 2.0

[20:33] natefinch3: ECONTEXT, feel free to reach out when you have time.. o/ [21:03] katco: http://reviews.vapour.ws/r/4813/ [21:03] pls [21:06] redir_: [21:06] redir_: sec [21:06] katco: np [21:06] easy one too === rodlogic is now known as Guest93911 [21:24] redir_: running into meetings, but i want to give this one more time. something looks fishy here === rodlogic is now known as Guest14174 [22:01] wallyworld, ping [22:18] Bug #1580184 changed: Timeout in github.com/juju/juju/apiserver/service on windows

[22:48] Bug #1580802 opened: NoContextWithLock fails on windows because of another process

[23:02] jam: (bit delayed) yeah, and I tried appdash and its UI is a little underwhelming. I modified jujud to create spans for each API connection, and subspans for each API call that comes in. easy enough, tho it's just a hack. it would be nice to be able to drill into timings for provider API calls too, so we'd probably end up exposing the tracing API to provisioners, firewaller, etc. === rodlogic is now known as Guest67947 [23:14] axw: anastasiamac_: redir_: perrito666: dtuck in meeting, will be a few minutes late [23:15] wallyworld: tis ok, we love you anyway [23:15]