/srv/irclogs.ubuntu.com/2017/01/03/#juju-dev.txt

=== frankban|afk is now known as frankban
frobwarejam: you around today?09:03
* frobware tries to find a 2FA device to check mail/cal09:03
rogpeppejam: ping?10:50
frobwarerogpeppe: curious as to why we are merging into juju:master for zaputil10:53
rogpeppefrobware: it's not juju:master - it's a new project10:54
rogpeppefrobware: ... unless i've borked things up horribly :)10:54
rogpeppefrobware: ah, it's juju:master because github only prints the username not the project in that place10:55
frobwarerogpeppe: well, possibly not. the GH page says juju:master, but I notice the popup says juju/zaputil:master... horrible UI.10:55
frobwarerogpeppe: apologies for the noise.10:55
* frobware sticks to the truth (aka the CLI)10:56
rogpeppefrobware: np10:56
frobwareaccess to reviews.vapour.ws is broken. anybody around to kick it back into life?10:58
rick_hfrobware: wfm ?11:51
frobwarerick_h: hmm. I always get a 50011:51
rick_hfrobware: oh what page? I could login at least11:52
rick_hfrobware: and click on a release test result summary11:52
frobwarerick_h: http://reviews.vapour.ws/11:52
frobwarerick_h: ah.....11:52
rick_hfrobware: oh, that's the old reviewboard reviews site. Is it even up any more?11:52
rick_hfrobware: I think they took it down, or maybe they left it up for old data but no one's noticed it went boom11:53
frobwarerick_h: hehe. too much (little) tab completion.11:53
frobwarerick_h: nevermind...11:53
* frobware blushes11:53
frobwarerick_h: indeed, http://reports.vapour.ws/ wfm too. :)11:53
rick_hfrobware: :)11:54
frobwarerick_h: what doesn't work atm is dynamic bridges. :(11:54
frobwarerick_h: just digging through some of the changes jam made since I left11:54
rick_hfrobware: k11:55
rick_hfrobware: we talked a bit last week so I know about some of it, but not sure where it's at today11:55
frobwarerick_h: would be great to catch up; also trying to figure out what needs to also be in develop vis-a-vis 2.1-dynamic-bridges11:56
frobwarerick_h: this looks like a long list - http://reports.vapour.ws/releases/468011:57
rick_hfrobware: I'm going to guess if it's a long list that there's some dirty substrate stuff with folks not around to help keep an eye on things11:58
rick_hfrobware: but not 100% sure11:58
frobwarerick_h: I'm pretty sure https://bugs.launchpad.net/juju/+bug/1652161 will break the functional-container-networking tests11:59
mupBug #1652161: juju-2.1-beta3 cannot add LXD container after host machine has rebooted <lxd> <network> <juju:New> <https://launchpad.net/bugs/1652161>11:59
frobwarerick_h: with or without dynamic bridging12:00
rick_hfrobware: k, will have to look at that. So we're not detecting an already configured lxd and trying to reconfigure it?12:02
perrito666wow the channel is back alive12:06
rick_hholiday is over, back to work12:08
voidspacerick_h: yep :-/12:13
voidspace:-)12:13
voidspacerick_h: ping12:14
voidspacerick_h: I'm working on bug #1631254 - lxd containers do not autostart12:14
mupBug #1631254: [2.0rc3] lxd containers do not autostart <rteam> <juju:In Progress by mfoord> <https://launchpad.net/bugs/1631254>12:14
voidspacerick_h: which is marked as critical now12:15
voidspacerick_h: the *specific bug*, as described, has been fixed for us by lxd12:15
voidspacerick_h: https://github.com/lxc/lxd/issues/246912:15
voidspacerick_h: there is still a bug in juju that needs fixing, and I have a fix - just need tests12:15
voidspacerick_h: however I can't manually test becasuse I can't repro (which is what I've spent this morning trying to do) because LXD have fixed the issue for us...12:16
voidspacerick_h: I can manually verify that my fix "does the right thing" with regard to the generated config however12:17
voidspacerick_h: but the bug can be downgraded from critical if it's holding anything up12:17
rick_hvoidspace: so can we not reproduce by setting the config value to auto start, manually shut down the lxd machine, then reboot and it does not come up before vs should with the config update?12:17
rick_hvoidspace: looking at the lxd issue, that was just that "if the container is running, restart it"12:18
rick_hvoidspace: so setting the config that it should autostart, manually turning it off, and rebooting the machine should still demonstrate the change you're making?12:18
voidspacerick_h: nope, it now restarts fine even if you hard shutdown the host12:18
voidspacerick_h: without my fix12:19
voidspacerick_h: wait12:19
voidspacerick_h: manually shutdown the lxd before hard reset12:19
rick_hvoidspace: I mean lxc stop the container first12:19
voidspacerick_h: and with the fix we would expect an auto-restart and without we wouldn't12:19
rick_hvoidspace: then restart the host12:19
voidspacerick_h: yep, good point12:19
voidspacerick_h: thanks12:19
rick_hvoidspace: all good12:19
voidspaceI have my develop env bootstrapped, so I can try that *now*12:20
voidspacerick_h: appreciated12:20
voidspacerick_h: yep, on develop stopping then restarting the host does not restart the lxd container12:22
voidspacerick_h: I actually wonder if my fix *does* change that....12:22
voidspaceI'll find out...12:22
rick_hvoidspace: so my understanding is that your branch fixes the configuration in lxd that controls that so I cross my fingers12:24
voidspacerick_h: yep12:24
=== petevg_holidays is now known as petevg
* voidspace lunch break13:08
voidspaceno school today, so no problem with second standup, beyond my normal philosophical objections to having two standups a day... :-)13:09
frankbanwallyworld: hey13:41
wallyworldhi there13:41
frankbanwallyworld: I was going to ask a question but, as it happens, I just found the answer. sorry, happy new year :-)13:43
wallyworldno worries. and same to you13:43
=== mup_ is now known as mup
natefinchsinzui: my latest merge build failed with /usr/lib/go-1.6/pkg/tool/linux_amd64/link: running gcc failed: fork/exec /usr/bin/gcc: cannot allocate memory14:47
natefinchhttp://juju-ci.vapour.ws:8080/job/github-merge-juju/9935/console14:48
natefinchabentley: ^14:49
perrito666gcc?14:49
natefinchhmm yeah weird14:49
sinzuinatefinch: perrito666 : I don;'t know why gcc is being used, but I believe it has something to do with a linker that will not actually have work to do.14:53
sinzuinatefinch: We can retry the merge, maybe we need more memory for the test machine14:53
natefinchyeah, building Juju takes a ton of memory... I'm sure gcc is just something used in the background... we can retry, but if there's nothing in general wrong with the machine (errant processes running etc) then definitely more memory will be in order14:54
perrito666sinzui: seems to be using gc-go?14:54
sinzuiperrito666: /usr/lib/go-1.6/pkg is no the location of gcc-go.14:56
natefinchsinzui: multiple merge jobs have failed with OOM... if there's nothing actually wrong with the machine, then we need a bigger machine15:04
perrito666natefinch: something might have leaked15:04
sinzuiperrito666: natefinch .yep. I just deleted 10 containers to reclaim memory15:04
perrito666natefinch: sounds like something fixeable with a restart15:04
frobwaremacgreagoir, voidspace: PTAL @ https://github.com/juju/juju/pull/6758 - explains some of why dynamic bridging isn't currently working.16:07
=== deanman is now known as deanman_
balloonssinzui, natefinch, yea that job is still running on the old slave. It still does the actual merging16:20
natefinchballoons: how much RAM does it have, out of curiosity?  obviously 10 errant containers are going to skew things pretty badly16:21
balloonsnatefinch, 14 gig.16:24
natefinchballoons: so is the merge job the only thing we run there? Because 14 gig should definitely be enough (again, w/o containers)16:25
balloonsnatefinch, it is 14 gig. But I lied, I was thinking the merges where done by the old slave. But they aren't. All PR jenkins jobs are running on it16:25
natefinchballoons: but serially, right?16:26
balloonsnot entirely. The merges are, but the pre-commit jobs are open season16:26
balloonsnatefinch, however your job was the only thing running on the box at the time. Looking back over the last month there's hardly ever more than 1 job running, so16:28
natefinchso probably just the leaking containers.  maybe we need a job that runs occasionally when the machine is idle and cleans them up?16:29
natefinch(obviously the best answer is not to leak containers, but that can be hard)16:30
alexisbhappy 2017 all!16:40
perrito666alexisb: <grumpycat>NO</grumpycat?16:43
perrito666:p16:43
redirGood time-zone appropriate period, and happy new year juju-dev16:50
=== kwmonroe_ is now known as kwmonroe
redirfrobware: yt?17:44
frobwareredir: yep17:44
frobwareredir: qemu and spaces?17:44
frobwareredir: HNY :17:45
frobware:)17:45
redirHNY!17:45
redircurious frobware if you thought this https://github.com/juju/juju/pull/675817:45
frobwareredir: that's broken for sure. Was just adding tests for that case as we speak17:45
redirmight be related to https://github.com/juju/juju/pull/6748#issuecomment-268935865 frobware17:47
frobwareredir: I confess to not trying this with KVM. I ran out of time towards the end of the year.17:49
redirwhen that change lands I'll cherry pick and cross my fingers it puts the rabbit back in the hat17:52
frobwareredir: want to cherry-pick with no-commit to try out now?17:52
frobwareredir: not sure I'll end up with my tests completed before my EOD (soon-ish)17:52
redirI s'pose that can't hurt17:53
frobwareredir: given this is so broken I would actually land the change as-is, with the promise to add unit tests.17:53
frobwareredir: this would also give the 2.1-dynamic-bridges branch a chance to get a CI run overnight.17:53
voidspacealexisb: happy new year17:55
alexisbsame to you voidspace!  hope you had a great holiday17:55
voidspacealexisb: yeah, really nice - lovely combination of busy-ness with people and relaxation with family17:56
voidspacealexisb: I return to work with a renewed and invigorated hatred for my job17:56
voidspaceoops, didn't mean to say that17:56
voidspace;-)17:56
voidspacealexisb: I hope you had a good break too, and managed to get away from work17:56
alexisbvoidspace, I did, disconnected for 2+ week - it was awesome17:57
=== frankban is now known as frankban|afk
natefinchrick_h: you around?18:37
rick_hnatefinch: otp what's up?18:38
natefinchrick_h: nothing super important, just wanted to talk about that deployer - to - manual email18:38
rick_hnatefinch: k, meet you in core18:39
TheMueHappy New Year, my Juju fellows19:19
natefinchTheMue: happy new year!19:22
TheMuenatefinch: started to work yesterday? I did, first a bit tired from vacation, but today better. got a new colleague in my team and teaching him go. :)19:27
natefinchAhh cool, must be fun teaching someone Go.19:28
TheMuenatefinch: yes, and he has a good feeling for it. so now we are 3 in my project.19:29
natefinchTheMue: That's great19:29
TheMuenatefinch: absolutely, I'm happy I'm able to place Go in anew project.19:30
natefinchsinzui, balloons, abentley: is the vsphere listed in cloud-city's clouds.yaml running?  Is there anything other than being on VPN that I need to do to access it?  It's listed at 10.245.0.13420:44
* redir lunches20:44
babbageclunkHappy new year everyone!20:51
redirHNY babbageclunk20:53
babbageclunkperrito666: I added Cloud.Name in https://github.com/juju/juju/pull/6735 - you were right, it looks better.20:54
perrito666I saw your change pass by but was in EOY mode :p20:55
babbageclunkperrito666: yeah, I figured! :)20:55
babbageclunkperrito666: I'll get jam to take another look today.20:56
perrito666babbageclunk: looks much better indeed20:56
abentleynatefinch: It is running.  You probably need to be on a specific machine to access it.  I am looking that up.20:57
perrito666natefinch: you need to ask IS to grant you permission to see it on the vpn20:57
perrito666our vpn is quite fine grained20:57
perrito666natefinch: actually someone from oil needs to open a rt for is to grant you permissions <-- abentley sinzui  babbageclunk20:58
perrito666sorry that was for balloons20:58
perrito666meh, I cant bootstrap with lxd because I die waiting for address (using develop) is this a problem for someone else still?20:59
sinzuiperrito666: I set natefinch the .ssh/config to get to the host we test with.21:00
perrito666nice "hacky vpn/vpn" :p21:00
natefinchsinzui: thanks21:02
perrito666ok EOD until standup, see you all later21:02
abentleysinzui: ping for standup21:04
redirping alexisb21:35
alexisbheya redir omw21:35
redirk21:35
alexisbneed about 5 more minutes21:35
natefinchso, I apt installed golang-1.6 and .... I still have no "go" command?  WTF?21:38
rediralexisb: np21:38
babbageclunkwallyworld: reviewed https://github.com/juju/juju/pull/675722:17
wallyworldbabbageclunk: great ty, will look22:18
babbageclunkwallyworld: around for a hangout?22:18
wallyworldsure22:18
redirwallyworld: you're back?22:18
wallyworldsupposedly22:18
wallyworldsince yesterday22:18
redirHNY wallyworld :)22:19
babbageclunkwallyworld: https://hangouts.google.com/hangouts/_/canonical.com/babbageclunk22:19
wallyworldsame to you22:19
perrito666Natefinch golang-go is the package iirc23:02
perrito666meh my dns are really acting weird today23:47
perrito666at least I believe its the dns23:47
redirmaybe it's just actin like dns23:48
perrito666has anyone experienced juju develop not being able to bootstrap lxd because it nevers gets an address?23:57
perrito666I ERROR failed to bootstrap model: waited for 20m0s without getting any addresses23:57
blahdeblahAnyone know when we will get 1.25.9 in the stable PPA for xenial?23:57

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