[00:56] <anastasiamac> thumper: wallyworld: m pretty sure that this si not intentional but with the change of default behavior of create backup in 2.4, it's possible to not download a copy and not store it remotely
[00:56] <anastasiamac> https://bugs.launchpad.net/juju/+bug/1786253/comments/1
[00:56] <anastasiamac> i'd like to put a stop to that... what's the point of running backup if there is no output
[00:56] <wallyworld> well that doesn't look very reasonable
[00:56] <wallyworld> +1 to fix
[00:57] <anastasiamac> also, is it really practical for us to allow --verbose here?
[00:57] <wallyworld> in 2.3.9 even
[00:57] <wallyworld> if the behaviour was first doner there
[00:57] <anastasiamac> who would want the contents of backup from command output as stdout
[00:57] <anastasiamac> i'll check 2.3
[00:57] <wallyworld> --verbose should be for extra status info etc
[00:58] <anastasiamac> k, yes verbose is just metdata
[00:59] <anastasiamac> and confirmed that the change in behavior only went in from 2.4-beta2
[01:28] <anastasiamac> oh no.... ran backup unit tests and got all this test backup archive files left behind...
[01:28]  * anastasiamac adding proper filesystem cleanup to test suite
[02:33] <veebers> Sorry for all the spam you're probably getting wallyworld, I'm close to a fix. I'll just push the oneliner the fix for your issue then propose a PR for the re-do of the merge jobs etc.
[02:33] <wallyworld> no worries
[02:41] <babbageclunk> kelvinliu_: dep seems to work pretty great for me thanks! I had one hiccup where I normally work in a symlinked dir ~/juju, rather than in $GOPATH/src/github.com/juju/juju, and dep complains about that.
[02:42] <babbageclunk> Do you know whether there's any way around it?
[02:43] <anastasiamac> veebers: u r only sorry that wallyworld is spammed? :(
[02:43] <veebers> anastasiamac: hah I hadn't realised others would to, ah because you commented on it right? Sorry to you too
[02:44] <babbageclunk> kelvinliu_: also, do you mind if I turn off the -v on dep ensure? It makes building pretty noisy - if I'm just changing a leaf package I'd normally only see a few lines of output.
[02:44] <anastasiamac> veebers: i was just picking on u :) in fact, it improves my blood flow as everytime email arrives, my machine pings and I startle/jump
[02:45] <veebers> ^_^
[02:45] <babbageclunk> (The symlink thing isn't much of a problem for me, I normally build through a wrapper script anyway so I just change to the directory that works.)
[02:46] <kelvinliu_> babbageclunk, sure, i think the -v  could be just for debuging.
[02:47] <babbageclunk> kelvinliu_: yeah, I figured that might be it.
[02:51] <babbageclunk> kelvinliu_: oh, the other noisy thing is that we now list all of the packages in the go install line rather than just "github.com/juju/juju/...". Is that needed to avoid the vendored dirs?
[02:51] <babbageclunk> kelvinliu_: I'm tempted to hide it, if that's ok.
[02:53] <anastasiamac> veebers: is it k to propose against develop or will it make ur current task worse?
[02:53] <kelvinliu_> babbageclunk, awesome, great to see u tempted to hide the annoying long list of packages,
[02:54] <kelvinliu_> babbageclunk, the purpose of listing packages was to exclude vendor
[02:54] <veebers> anastasiamac: go nuts, if the check/merge fails for you let me know. I have a quick fix for the immediate problem, I'm working on restructuring the whole job system as as separate but related task
[02:54] <babbageclunk> yeah, that makes sense
[02:55] <kelvinliu_> babbageclunk, thanks for doing the enhancement.
[03:10] <wallyworld> anastasiamac: here's that PR :-D https://github.com/juju/juju/pull/9048
[03:10] <anastasiamac> wallyworld: yep, Awesome \o/ looking
[03:10] <wallyworld> ty, i owe you a beer
[03:11] <wallyworld> it's not too big actually
[03:11] <anastasiamac> beer?
[03:12]  * anastasiamac stops reviewing
[03:15]  * babbageclunk is sorely tempted.
[03:15] <babbageclunk> hey axw!
[03:19] <babbageclunk> kelvinliu_: any idea why the digest is 0: for golang.org/x/crypto in the committed Gopkg.lock? I just updated my first dependency in dep-world, it was fine except it's also updated the digest for the crypto package despite my not having touched it.
[03:23] <kelvinliu_> babbageclunk, i didn't know the digest was 0 for crypto.  needs to investigate
[03:24] <babbageclunk> kelvinliu_: cool, thanks
[03:34] <vinodhini>  wallyworld: could u please take a look at the PR : https://github.com/juju/juju/pull/9049
[03:34] <wallyworld> sure
[03:34] <vinodhini> who is that ?
[03:36] <vinodhini> wallyworld: the ca-cert for now i havent taken into the 2.4.2 - to me it looks that it is out of scope of this bug fix.
[03:36] <vinodhini> Le t me know ur inputs
[03:36] <wallyworld> sure. i wasn't intending to take thw who ca cert thing, just the small refactoring of certain parts of the code that were common to this change
[03:37] <wallyworld> i'll look at the pr
[03:38] <anastasiamac> wallyworld: reviewd but m raising ur beer offer to sparkly, at least...
[03:38] <wallyworld> ok :-)
[03:38] <wallyworld> ty
[03:39] <wallyworld> next week
[03:39] <wallyworld> monday even
[03:39] <anastasiamac> mayb...
[03:39] <anastasiamac> it's a short week in bne and am saving myself for partying
[03:43] <wallyworld> i'm too old to parety
[03:44] <babbageclunk> kelvinliu_: oops, my computer died.
[03:45] <anastasiamac> wot? my 99 yo granny still parties... u r not that old, wallyworld :) altho they do say that u r as old as u feel
[03:45] <wallyworld> sometimes i feel 21 again, mostly not
[03:46] <anastasiamac> :) early 20s tend to be crazy - taste of freedom and all that....
[03:46] <kelvinliu_> babbageclunk, i just found that digest updated from "0:" to "1:xxxx" as well. don't know why it was 1 tho.
[03:48] <babbageclunk> kelvinliu_: <shrug> I don't think it matters too much. I'll just leave it.
[03:49] <kelvinliu_> babbageclunk, yeah, let's see if it happens again in the future
[03:51] <babbageclunk> kelvinliu_: can you review this please? https://github.com/juju/juju/pull/9050
[03:56] <wallyworld> vinodhini: looks pretty good but needs tests. also don't forget to remove the template from the PR description
[03:58] <babbageclunk> veebers: I'm getting a weird build failure - can't install go snap. Is this a known thing or something new? http://ci.jujucharms.com/job/github-check-merge-juju/2898/console
[03:58] <veebers> babbageclunk: let me have aa quick look
[03:58] <babbageclunk> thanks!
[03:59] <veebers> babbageclunk: no, that's some weird failure that I've seen happen twice now. a re-run should go through properly
[04:00] <veebers> babbageclunk: hah, infact my test job (which shouldn't have triggered) is running that pr branch fine; http://ci.jujucharms.com/job/veebers-github-check-merge-juju/27/console
[04:00] <veebers> unfortunately it won't count
[04:00] <babbageclunk> veebers: ok, kicking it off again, ta
[04:01] <veebers> babbageclunk: I'm really not sure what the error is due to, will have another look at some point, but it seems udev is unhappy within the container
[04:01] <vinodhini> wallyworld: thats why its WIP - i am working on unit test part
[04:02] <wallyworld> vinodhini: oh, doh! sorry, i didn't see that
[04:02] <wallyworld> maybe time for glasses
[04:02] <babbageclunk> too much partying
[04:03] <kelvinliu_> babbageclunk, LGTM, thanks!
[04:04] <anastasiamac> well, m reading wallyworld's "glasses" as "friday drinks" too
[04:04] <wallyworld> i wish
[04:05] <babbageclunk> kelvinliu_: cheers
[04:11] <babbageclunk> veebers: This one looks bad too... http://ci.jujucharms.com/job/github-check-merge-juju/2899/console
[04:11]  * veebers looks
[04:12] <babbageclunk> doh, not on the vpn after rebooting
[04:12] <veebers> babbageclunk: hmm, yeah that might be a cloud-init issue, I've seen that recently before too. I've spoke briefly to someone about it, let me poke around a bit
[04:12] <babbageclunk> Thanks
[04:20] <babbageclunk> veebers: hah, your test job against my PR did count!
[04:20] <veebers> babbageclunk: erally? oops, luckily it's just the check and not an erroneous merge
[04:21]  * veebers deletes that job
[04:21] <babbageclunk> true that
[04:21] <babbageclunk> aww, what happened to the veebers-rulez container?
[04:21] <veebers> babbageclunk: I've just aborted that job that was stuck
[04:21] <veebers> babbageclunk: hah you on that machine? :-)
[04:22] <babbageclunk> yeah, once I remembered to start the vpn
[04:22] <veebers> babbageclunk: We had a sick machine that was gumming up the cleanup jobs, so thinigs wheren't being cleaned up. I think too many old lxd machines on there does something to the cloud-init on those jobs
[04:23] <babbageclunk> ah right. hey, did you just kill my merge job?
[04:23] <veebers> babbageclunk: so in short, I killed that machine, re-ran the cleanup to get grumpig cleaned up, killed that stuck job, You need to re-build that pr
[04:23] <veebers> hopefully it'll get through this time
[04:24] <veebers> babbageclunk: um, sorry yes
[04:24] <babbageclunk> ha, no worries
[04:24] <veebers> babbageclunk: it wasn't going to go anywhere. I'm hoping to get more insight on it, but I think the cleanup will sort it for the immediate future
[04:27] <babbageclunk> oops, think I've managed to schedule multiple merges - cleaning them up now.
[04:48] <anastasiamac> an easy review anyone? https://github.com/juju/juju/pull/9051 adds uuid back for compat
[04:51] <vinodhini> wallyworld: have a min ? i want to discuss abt this goose lib
[04:52] <wallyworld> ok
[04:52] <vinodhini> i wud like to HO please.
[04:53] <wallyworld> sure, i'm there
[04:55] <anastasiamac> babbageclunk: any chance u could have a look since apparently u r my patner in the original crime...?^^^
[04:56] <babbageclunk> anastasiamac: well, ok, as long as it's easy!
[04:56] <anastasiamac> babbageclunk: when is it ever?
[04:56] <anastasiamac> but PR is small
[04:57] <babbageclunk> anastasiamac: why have both? Is that going to be more confusing than just having uuid?
[04:58] <anastasiamac> babbageclunk: to ensure backward compatibility... and a way forward... we've been discussing it instandup the last couple of days :)
[04:58] <anastasiamac> babbageclunk: also, see thumper comments in the linked bug ;)
[04:58] <babbageclunk> ok, reading more
[04:59] <anastasiamac> babbageclunk: thank you :)
[05:00] <babbageclunk> anastasiamac: approved!
[05:00] <anastasiamac> babbageclunk: \o/
[05:01] <babbageclunk> anastasiamac: hey, how was I your partner in the original crime?!
[05:01] <anastasiamac> u were reviewer when i've renamed 'uuid' to be 'controller-' or 'model-'
[05:02] <anastasiamac> lol... how else?
[05:02] <babbageclunk> whoa, totally blanked that from my mind.
[05:04] <veebers> babbageclunk: I'm not sure that argument would hold up in court . . .
[05:05] <anastasiamac> kelvinliu_: m doing something wrong.. teething problem mayb? m was off develop locally, switched to 2.4 but running 'make install' or 'go install ./...' does not work
[05:06] <kelvinliu_> anastasiamac, u will need to run `make godeps` after switched to 2.3 or 2.4
[05:06] <anastasiamac> kelvinliu_: i have
[05:06] <anastasiamac> beofre trying to install
[05:06] <anastasiamac> i'll try again
[05:08] <kelvinliu_> anastasiamac, because we need to rm -rf ./vendor then ensure dependencies via `godeps` for 2.3/2.4
[05:09] <anastasiamac> kelvinliu_: sorted... yes... i was not runing make target but godeps directly... old habits die hard... i'll b better by monday i promise
[05:10]  * anastasiamac winces
[05:14] <kelvinliu_> anastasiamac, cool! hope the change doesn't impact ur workflow too much.  : )
[05:15] <anastasiamac> kelvinliu_: so far, pretty seamless !!! so kind of awesome
[05:22] <anastasiamac> and another awesome and laconic PR for review, plz - https://github.com/juju/juju/pull/9052 - help for create-backup and removing test artifacts
[05:24] <anastasiamac> wallyworld: any chance u could PTAL ^^ 2.4 one after all..
[05:24] <wallyworld> sure
[05:24] <anastasiamac> \o/
[05:29] <wallyworld> anastasiamac: done with a request to reject incompatible cli args
[05:31] <anastasiamac> wallyworld: so, the problem is that they are not incompatible. it's totally okay to say --keep-copy --no-download...
[05:31] <wallyworld> sure, so check the value
[05:32] <wallyworld> if keep==false and no-download then complain
[05:32] <anastasiamac> wallyworld: eve if the user says '--keep-cope=false --no-download', we will ignore --keep-copy
[05:32] <anastasiamac> ic
[05:32] <wallyworld> that's when we should error
[05:32] <wallyworld> make sense?
[05:33] <anastasiamac> -keep-copy is false by default
[05:33] <anastasiamac> i.e. when we read a flag we supply 'false' as default...
[05:33] <anastasiamac> how do u know if that false is user-supplied vs the default value?
[05:34] <anastasiamac> wallyworld:
[05:34] <wallyworld> doesn't matter how the true/false value gets there. if c.Keep == false and c.NoDownload == false then it's an error
[05:34] <wallyworld> NoDownload==true
[05:35] <wallyworld> ie we don't want to allow the user to accidentially ask for a no nop
[05:35] <anastasiamac> waht u describe is undetectable
[05:58] <anastasiamac> wallyworld: really made me sweat for i :D but found it... i'll update the pr.. want to review it before i land?
[05:59] <wallyworld> nah, all good
[05:59] <anastasiamac> ack
[05:59] <wallyworld> as long as there's a test
[05:59] <anastasiamac> yes, of course :) this is how i know it works too :D
[06:00] <wallyworld> :-)
[08:20] <veebers> stickupkid: I realise now that I said I was going to counter prpose your ci test changes but never did, I might try hit that real quick for you now
[08:23] <stickupkid> ta
[08:37] <veebers> stickupkid: FYI current-parameters
[08:37] <veebers> heh
[08:37] <veebers> stickupkid: actually FYI https://github.com/CanonicalLtd/juju-qa-jenkins/pull/73
[08:37] <stickupkid> ah, that solves it better than mine
[08:38] <stickupkid> CR your PR :D
[12:09] <manadart> Anyone got insight into what can cause repeated dependency engine errors like this?
[12:10] <manadart> ERROR juju.worker.dependency engine.go:587 "api-address-updater" manifold worker returned unexpected error: connection is shut down
[13:38] <jamespage> odd question but if I need to use pip to install a workload from a reactive charm, how do you break out of the virtualenv being used for the charm hook itself?
[13:38] <jamespage> cory_fu: ^^ ?
[13:41] <icey> jamespage `deactivate; pip install $MY_PACKAGE` ?
[13:42] <icey> probably break all kinds of other stuff though
[13:48] <cory_fu> jamespage: What icey suggested might work in a subshell.  I think you could also filter the environ dict passed to subprocess (maybe just pass an empty dict)
[13:50] <jamespage> cory_fu: is it VIRTUAL_ENV that causes pip to install to the venv rather than globally?
[13:50] <cory_fu> jamespage: There are a few env vars that get modified.  PATH, possibly PYTHON_PATH, maybe some others, I'm not sure
[13:58] <jamespage> cory_fu: I think if they fully path /usr/bin/pip it will dtrt - the issue is that the pip in the venv is being used
[13:58] <jamespage> or maybe not...
[13:58]  * jamespage puzzled
[14:17] <hml>  stickupkid: pr reviewed, lgtm with a few questions
[14:17] <manadart> stickupkid: Quick HO?
[14:20] <stickupkid> sure
[15:52] <stickupkid> rick_h_: you got 5 minutes for a quick HO?
[15:52] <rick_h_> stickupkid: for you, sure thing
[20:15] <magicaltrout> rick_h_: to your knowledge if jujucharms.com having any issues today?
[20:15] <magicaltrout> oop well thats funky
[21:17] <pmatulis> works here
[21:29] <magicaltrout> yeah its been a bit weird today
[21:29] <magicaltrout> one of my guys earlier was complaining about auth issues
[21:29] <magicaltrout> then i was searching for stuff i know exists that didn't show up
[21:29] <magicaltrout> then 5 minutes later, did
[21:29] <magicaltrout> weird
[21:37] <pmatulis> IT maintenance maybe, db upgrade?