[00:02] axw: can I borrow you to clarify a couple things? [00:35] thumper: do you have a sec to peek at https://github.com/juju/utils/pull/226/files and ship if it's ok today please? [00:35] sorry, that's not the RB link but close enough [00:43] rick_h_: ack [00:49] bbiab === veebers2 is now known as veebers === natefinch-afk is now known as natefinch [02:16] menn0: I found out why that ARCH patch was there... [02:16] menn0: fails on s390 [02:16] without it [02:21] thumper: argh right [02:21] proposing fix now [02:21] * menn0 doesn't really understand why [02:22] but fair enough [02:23] menn0: http://reviews.vapour.ws/r/5314/ [02:26] thumper: should that setup live in SetUpTest or will that cause other problems? [02:27] thumper: never mind... looking at the code it's only done in one other place [02:27] my gut tells me it may cause other problems [02:27] better to just do it when necessary [02:28] thumper: ship it with one sugestion [02:28] * thumper looks [02:28] ack [02:33] redir: sorry was out before. if you're still around I'm here now [02:35] thumper: I don't really understand why your change was needed. we're already setting agent-version earlier in the function -- why doesn't that work? [02:35] thumper: ahh, because building the tools bumps the version? [02:35] yep [02:40] axw: np, I'll just dig in tomorrow. [04:01] Bug #1607161 opened: generate-image metadata command help needs updating

[04:10] Bug #1607161 changed: generate-image metadata command help needs updating

[04:13] Bug #1607161 opened: generate-image metadata command help needs updating

[04:24] thumper: apiserver changes for migration status in the FullStatus response [04:24] http://reviews.vapour.ws/r/5315/ [04:24] * thumper looks [04:37] Bug #1607170 opened: enable-ha requires '-m controller' (i.e. doesn't do the 'right thing' by default).

[04:38] thumper: by error state do you mean ABORT? [04:38] if so yes it is (well actually ABORTDONE is the final, terminal phase) [05:08] menn0: chat? [05:15] abort done is only set after aborting is finished? [05:15] I was thinking about an intermediate error state [05:33] thumper: sure [05:33] going to to 1:1 [08:58] Bug #1605986 opened: Creating container: can't get info for image 'ubuntu-trusty'

[09:09] 2016-07-28 08:04:34 INFO config-changed subprocess.CalledProcessError: Command '['/opt/opendaylight-karaf/bin/client', '-r', '20', '-h', 'localhost', '-a', '8101', '-u', 'karaf', 'feature:list']' returned non-zero exit status 1 [09:09] odl-controller charm [09:09] anyone can help me ? [09:19] ejat: If you run that command on the machine, do you get any useful output? [09:20] u mean this : /opt/opendaylight-karaf/bin/client', '-r', '20', '-h', 'localhost', '-a', '8101', '-u', 'karaf', 'feature:list [09:20] or '['/opt/opendaylight-karaf/bin/client', '-r', '20', '-h', 'localhost', '-a', '8101', '-u', 'karaf', 'feature:list']' [09:20] ejat: yes, but without the quotes and commas. [09:21] /opt/opendaylight-karaf/bin/client -r 20 -h localhost -a 8101 -u karaf feature:list <-- ? [09:21] ejat: yes [09:22] $ /opt/opendaylight-karaf/bin/client -r 20 -h localhost -a 8101 -u karaf feature:list [09:22] client: JAVA_HOME not set; results may vary [09:22] Logging in as karaf [09:22] 175 [pool-2-thread-2] WARN org.apache.sshd.client.keyverifier.AcceptAllServerKeyVerifier - Server at localhost/127.0.0.1:8101 presented unverified key: [09:22] Error executing command: java.lang.NullPointerException [09:22] juju not set the java_home ? [09:24] ejat: It sounds more like there's a problem with the key, but I don't know anything about opendaylight. [09:25] ejat: You might be better asking in #juju - there are more openstack people and charmers there. [09:25] ok thanks babbageclunk [09:26] ejat: good luck! [09:26] grattitude [10:22] voidspace: welcome back! Just around for 2 days? [10:25] frobware, voidspace: a couple of reviews if you have a moment? http://reviews.vapour.ws/r/5312/ and http://reviews.vapour.ws/r/5317/ [10:27] babbageclunk: will get to it later; syncing with macgreagoir [10:28] frobware: cool, thanks - no big rush [10:35] Bug #1607000 changed: Immortal zombie Juju controller on MAAS

[10:35] Bug #1607303 opened: Help text for juju kill-controller needs to point to unregister

[10:42] macgreagoir: combined patch updated - https://github.com/frobware/juju/tree/master-lp1566801-strike3%2Blp1602054 [11:12] frobware: FTR, woo hoo! === tinwood is now known as tinwood-lunch [11:29] babbageclunk: yeah, just here for a couple of days [11:29] babbageclunk: off to cornwall with my dad for a week after that [11:30] voidspace: nice [11:32] babbageclunk: looks like everything changed whilst I was away... [11:32] voidspace: of course! [11:32] babbageclunk: your doing no doubt... [11:33] babbageclunk: I got a microbit from EuroPython! [11:33] babbageclunk: I haven't played with it yet though [11:34] babbageclunk: I'm currently trying to fix MAAS (again) [11:34] babbageclunk: a recent update broke my install [11:36] voidspace: stink. I recently redid my MAASes using frobware's scripts - took a bit of time to get set up, but now I can add/remove nodes really easily. [11:36] babbageclunk: ah cool - got a link [11:36] babbageclunk: I'll try and get that working [11:37] voidspace: In KVMs, or hardware? [11:37] babbageclunk: KVM [11:37] I hacked this one up: https://github.com/frobware/ubuntu-fan-hacks [11:38] babbageclunk: have you pushed your changes anywhere, or are they babbageclunk specific? [11:38] babbageclunk, voidspace: I should really update the docs there .... [11:38] ...to get the initial VM, then installed maas manually. [11:38] then used https://github.com/frobware/kvm-maas with minimal hacks to add nodes. [11:39] babbageclunk: what changes did you need for kvm-maas? [11:40] frobware: checking - only cut down the disk sizes a bit - I couldn't get it to respect the env vars. Nothing that needs pushing. [11:41] babbageclunk: hmm - I wonder why [11:42] voidspace: The only tricky thing with the initial maas controller setup is that you need a hacked cloud image from frobware that enables cloud-init to do network setup in a network where there's no dhcp [11:42] babbageclunk: ah, right [11:43] babbageclunk: so I'm relibably informed that that's now the default [11:43] voidspace: ok - so that might be fixed! [11:43] voidspace: if that's not true there's always my enhanced version: http://178.62.20.154/~aim/xenial-server-cloudimg-amd64-disk1.img [11:43] frobware: babbageclunk: cool, thanks - will have a play [11:44] voidspace: hang on, I'll push my hacked (and now badly named) ubuntu-fan-hacks branches - I've got one for maas2 and one for maas1.9 [11:46] frobware, hey there, are you OCR today? [11:47] mattyw: yes [11:47] frobware, I have a tiny one for you. http://reviews.vapour.ws/r/5319/. I'm about to got afk but if you had any questions I'll be back in an hour or so [11:48] frobware, but if you're busy - it's not that urgent :) [11:48] mattyw: me too - lunch is required [11:48] frobware, I'm planning on combining it with a cycle as the sun is out :) [11:48] babbageclunk: running VIRT_DISK_SIZE=12 ./add-node disk-size-tes maas19-net maas19 [11:48] babbageclunk: gives me: http://178.62.20.154/~aim/disk-test.png [11:49] mattyw: heh, sounds fab! [11:49] frobware: yeah, I may have been getting confused by something VMM was telling me - I think that's working fine. [11:50] voidspace: https://github.com/babbageclunk/ubuntu-fan-hacks/tree/maas-2-vm [11:50] * frobware really really really goes for lunch. Really. Almost. Really. [11:50] babbageclunk: thanks [11:50] voidspace: https://github.com/babbageclunk/ubuntu-fan-hacks/tree/maas-2-vm [11:51] voidspace: https://github.com/babbageclunk/ubuntu-fan-hacks/commit/5f5b736308bfdfb15af84fe7986d252b7ef0990b [11:51] voidspace: That should show the kinds of things you're likely to need to change. [11:51] babbageclunk: cool [11:52] voidspace: Still need to install maas manually and click around in the admin a little bit, but it's a bit simpler. [11:52] babbageclunk: I'm just doing the mass install at the moment [12:16] frobware: what's the right ppa for the latest and greatest maas [12:16] frobware: ah, experimental3 === tinwood-lunch is now known as tinwood [12:23] Bug #1607347 opened: Password for juju-gui not showing up after a change

[13:05] Bug #1607362 opened: TestMachinerMachineNotFound agent should be terminated

[13:11] Bug #1607362 changed: TestMachinerMachineNotFound agent should be terminated

[13:18] frobware: I think I'm showing --bind working here (with your latest eth0 patch and an lp:1603473-like net config): https://pastebin.canonical.com/161866/ [13:19] macgreagoir: yes, looks good [13:23] Bug #1607362 opened: TestMachinerMachineNotFound agent should be terminated

[13:23] Bug #1607365 opened: PresenceSuite.TestScale reported false

[13:23] Bug #1607368 opened: bootstrapSuite.TestBootstrapUploadTools model "foo" of type dummy does not support instances running on "ppc64el"

[13:32] Bug #1607365 changed: PresenceSuite.TestScale reported false

[13:32] Bug #1607368 changed: bootstrapSuite.TestBootstrapUploadTools model "foo" of type dummy does not support instances running on "ppc64el"

[13:35] Bug #1607365 opened: PresenceSuite.TestScale reported false

[13:35] Bug #1607368 opened: bootstrapSuite.TestBootstrapUploadTools model "foo" of type dummy does not support instances running on "ppc64el"

[14:04] katco: standup? [14:05] rick_h_: you doing standup today? [14:09] sinzui, mgz: it looks like we're building the CI tests with an old version of go? [14:10] natefinch: ouch, I will look into it. Which test/host? [14:10] sinzui: http://juju-ci.vapour.ws:8080/job/github-merge-juju-utils/125/console [14:10] natefinch: not ci then [14:10] sinzui: oh, right, landing bot, sorry [14:11] natefinch: found the problem. Not every merge job has the path set right. I can fix this job in a minute. I need to review about 8 jobs though [14:12] sinzui: ok... glad to help find problems :) [14:13] natefinch: replaying the job with the right juju http://juju-ci.vapour.ws:8080/job/github-merge-juju-utils/126/console [14:13] sinzui: yay, thanks. And glad that we'll be using the right version of Go. I intentionally removed the guardrails for go 1.2 vs 1.6 in this change.... we really just won't build with go 1.2 anymore. [14:29] Is there anything that logs the actual userdata we send when starting an instance (particularly in maas)? [14:46] babbageclunk: natefinch are your bugs the same: bug 1604474 and bug 1605777 [14:46] Bug #1604474: Juju 2.0-beta12 userdata execution fails on Windows

[14:46] Bug #1605777: munna/juju2 is too slow to deploy windows images

[14:49] sinzui, natefinch - I think they are, yes. [14:49] Good spotting! [14:52] sinzui: Should I just mark 5777 as a duplicate of 4474? natefinch's quite far along with it, as far as I can see. [14:52] babbageclunk: yep. [15:08] Bug #1605777 changed: munna/juju2 is too slow to deploy windows images

[15:19] natefinch: ping - some advice [15:19] natefinch: if you have the mental bandwidth (juju advice - not life advice, don't worry) [15:19] for that I go to babbageclunk [15:24] voidspace: for *bad* life advice [15:25] uh uh uh, bad life advice, I can do that, who is in need? [15:28] babbageclunk: well, naturally... [15:28] perrito666: hah, thanks - if the advice from babbageclunk ever turns out to not be bad enough I'll turn to you [15:29] perrito666: babbageclunk: frobware: juju bug 1599972 [15:29] Bug #1599972: juju2 beta11 unable to parse PORT during a maas bootstrap <2.0>

[15:29] perrito666: babbageclunk: frobware: the bug is that specifying the MAAS server in the form cloud.lol:5240 doesn't work [15:29] with an error that it's a malformed url [15:30] perrito666: babbageclunk: frobware: specifying instead http://cloud.lol:5240 however works [15:30] do you think that is worth fixing or should be a "won't fix"... [15:30] voidspace: is 'cloud.lol:5240' actually a URL? [15:30] frobware: well it is is you have cloud.lol in /etc/hosts [15:30] which I do [15:31] frobware: because ip:port worked so I needed a domain [15:31] voidspace: I'm really asking do you need http:// or ftp:// or ... [15:31] frobware: if you don't specify the scheme but do specify a port then the url is "incorrectly" parsed [15:32] voidspace: so GIGO [15:32] frobware: yep [15:32] maas/172.0.0.1:5240 doesn't work [15:33] stokachu: ah, it worked for me - trying again [15:33] voidspace: cloud.lol is a hostname [15:33] http://cloud.lol is a url [15:33] voidspace: cloud.lol:port is incredibly ambiguous as it could be both [15:33] oh maybe I'm mistaken [15:33] maas/cloud.lol:5240 is what doesn't work [15:34] and you're saying that we should be using maas/http://cloud.lol:5240? [15:34] stokachu: yep [15:35] stokachu: what I'm *really* saying is that if you don't specify a malformed url you won't get the "malformed url" error... ;-) [15:35] stokachu: but I was trying not to be a dick about it [15:35] stokachu: voidspace I was just clarifying why the erorr makes sense, I lack context for more detail [15:35] not trying very hard though... :-p [15:35] stokachu: although cloud.lol:5240 is what I would have tried as well [15:36] voidspace: in the same vein that $(firefox cloud.lol:5240) would just work, should juju? [15:37] frobware: it would be nice if it did [15:37] frobware: firefox is lenient, if you put $(firefox agoogleableterm) will also work [15:37] frobware: I don't think it's a critical bug that it doesn't [15:37] voidspace: nope, but I do like DWIM [15:38] this said, we should be a bit nicer in how we treat urls, if we expect a url and its not a url we could try a hostname [15:38] perrito666: frobware: on error just adding http:// and retrying would work - if it still fails return the original error [16:51] voidspace: sorry, was afk. [16:52] sinzui: are we sure that #1605777 is the same as the bug I fixed? The bug I fixed was not that windows was slow - it was entirely broken [16:52] Bug #1605777: munna/juju2 is too slow to deploy windows images

[16:57] Bug #1607457 opened: juju migrate fails to find model after migration

[17:00] natefinch: The bug does not say windows is slow, It says juju 2. cannot deoploy a windows machine by 1.25 can. The suibstrate is known to be slow [17:02] sinzui: I don't know what munna is, I assumed it was a windows machine in CI [17:04] sinzui: regardless, it sounds like it's completely different from the bug that I fixed, which was that the userdata script we run on windows couldn't connect to the controller to download the jujud binary, because we didn't support a ciphersuite that windows supported. [17:06] natefinch: okay. The issue we see is that the machine agent never comes up. Looks like a cloud init issue [17:09] sinzui: that *could* be the same problem, then, since it's the userdata script failing. But it would have been a new failure around July 7th [17:10] That's when we broke windows with the thing I just fixed [17:13] natefinch: We first see the issue in July 15 http://reports.vapour.ws/releases/issue/5789759d749a5616aa83b491 We stopped testing on the host last week. The previous failures were timeouts, but machine want not checked [17:15] natefinch: The last pass of the test for juju 2 was jujy 7! the first failure is on the 8th http://reports.vapour.ws/releases/4121/job/maas-1-9-deploy-win2012hvr2-amd64/attempt/624 [17:16] sinzui: sound suspicious :) Let's try it again with this fix and see how it goes. [18:39] / Fields holds a map from attribute name to information about that attribute. [18:39] ....so why isn't it called Attributes? :/ [19:02] Bbl [20:16] brb reboot [20:23] natefinch: ping [20:23] natefinch: re: https://github.com/juju/utils/pull/226#event-737900037 I saw this fail to land yesterday, and don't see a change in code, did the landing bot change? [20:23] sinzui: ^^ [20:24] thumper: yep [20:24] was it using an older go? [20:24] thumper: yep [20:25] ok [20:25] thumper: so, win-win... it landed and now we have a canary for jobs that use a too-old version of go. [20:26] I had intentionally removed the guardrails for some go 1.6 code in there, since we officially support it everywhere now. [20:31] natefinch: I updated all the merge jobs to include go 1.6 in the path. We wont see any go 1.2 in the current jobs [20:33] sinzui: huzzah [20:55] man I hate middle-click paste [20:58] simple PR for anyone - https://github.com/juju/schema/pull/13 === natefinch is now known as natefinch-afk [21:35] anyone here know anything about the model-config-tree details? [21:38] natefinch-afk: LGTM, FWIW [23:01] Bug #1607557 opened: wrong JSON field for storage endpoint in API [23:04] thumper: addition MigrationMaster.SetStatusMessage API (small and boring): http://reviews.vapour.ws/r/5324/ [23:40] axw: same bat station?