/srv/irclogs.ubuntu.com/2016/09/06/#juju-dev.txt

menn0thumper: Fix for the "migrations get stuck" bug: http://reviews.vapour.ws/r/5599/01:46
* thumper looks01:46
thumpermenn0: just one Q on the review01:51
menn0ok01:59
thumperoh ffs, will this branch die...01:59
anastasiamac_thumper: u reviewed Casey's PR yesterday (http://reviews.vapour.ws/r/5588/) r u k to stamp it?02:35
axwwallyworld: still going on this, but if you have any time I'd appreciate if any early comments before I delve into unit tests: https://github.com/juju/juju/compare/master...axw:login-local-macaroon?expand=102:53
wallyworldok, give me 1002:53
axwwallyworld: in particular, there's some UX changes02:53
wallyworldor 502:53
axwsure, take your time02:53
anastasiamac_axw: menn0:wallyworld:thumper: any idea who is looking after https://github.com/juju/blobstore? us or Uros's team?02:56
wallyworldeither02:56
wallyworldwhoever needs to make changes does so02:57
wallyworldwe created it initially02:57
wallyworldthey have landed patches since02:57
menn0anastasiamac_: I emailed rog about it yesterday. what's up?02:57
anastasiamac_wallyworld: menn0: m trying to determine if the issues agaisnt this libraries are ours (and hnce need to be in launchpad) or theirs (and then I do not need to worry)02:58
anastasiamac_library*02:58
wallyworldi don't think we track bugs for that library on launchpad02:59
wallyworldany issues for those upstream libraries in github.com/juju are tracked in github IIANM03:00
anastasiamac_wallyworld: not all, there are some bugs in launchpad. for example, most recently, licence bugs were floating around03:04
wallyworldoh? ok. what project were they filed against?03:05
anastasiamac_wallyworld: and as far as I can see we r not *really* tracking in github03:05
menn0anastasiamac_: in case it's relevant, I emailed rog b/c we don't use github.com/juju/blobstore directly. We just the newer version at gopkg.in/juju/blobstore.v2.03:05
wallyworldoh, that needs to be fixed :-(03:05
menn0anastasiamac_: the only reason we still have it in dependencies.tsv is because charmstore.v5 still uses it03:05
anastasiamac_menn0: tyvm - u r the best \o/ we'll wait for rog :D03:06
menn0rog said that it shouldn't cause a problem b/c of the way it's used03:06
wallyworldwe could do tha one ourselves if we really wanted to :-)03:06
menn0and that they're moving away from using mongodb for a blobstore "soon" anyway03:06
anastasiamac_oooh.. what r they moving toward to?03:06
menn0anastasiamac_: no idea, he didn't say03:07
anastasiamac_intersting ... :D03:07
menn0this is what rog said: We actually plan to move away from using mongo-based blobstore at some point but in the meantime we should probably move to v2. I don't see any potential harm in having both versions as deps in juju though in the meantime - neither have any global state AFAIR.03:08
anastasiamac_menn0: \o/ k03:08
* wallyworld hates carrying unnecessary deps03:08
wallyworldadds to build time and landing bot time03:08
anastasiamac_wallyworld: so here is one for ur and thumper's delight: 2 PR against juju/cmd (on github) LGTM'ed in 2014.. one can be merged, one has conflicts :D03:09
wallyworld"delight"03:09
anastasiamac_wallyworld: thumper: could we plx decide if we close the or land them?03:09
anastasiamac_\o/03:09
anastasiamac_always "delight"03:10
anastasiamac_hand-in-hnad with  "wonder" :)03:10
* thumper relocates03:10
anastasiamac_to OZ?03:10
thumperdropping kid at bjj and working from cafe there03:10
thumperback online from cafe03:10
anastasiamac_wallyworld: i thin u've just been delegated03:10
* wallyworld is busy, will look later03:11
anastasiamac_awesome \o/03:11
anastasiamac_axw: menn0: wallyworld: is https://github.com/juju/errgo officially deprecated in favour of juju/errors?03:16
wallyworldonly in core03:16
wallyworldit's still used upstream :-( :-(03:16
axwwallyworld: does it? macaroon stuff uses gopkg.in/errgo.v103:17
wallyworldaxw: that's my point :-)03:17
axwcan't see any imports - I think it's dead03:17
wallyworldi want to use juju/errors everywhere03:17
anastasiamac_wallyworld: axw: so if there is a PR against this library, who is to review/land?03:17
axwwallyworld: I just meant there's gopkg.in/errgo.v1, which != juju/errgo. but yes, we still have two03:18
anastasiamac_r we responsible for it?03:18
wallyworldwe = rog03:18
anastasiamac_we != rog :D03:18
wallyworldjuju core doesn't maintain errgo03:19
wallyworldjuju core would like not to have to pull in errgo at all03:19
anastasiamac_\o/03:19
anastasiamac_i'll happily ignore these PRs then :D03:19
anastasiamac_wallyworld: axw: menn0: what about https://github.com/juju/errors?03:20
wallyworldthat's out error package03:21
wallyworldour03:21
menn0kinda need that one :)03:21
axwwe could just do better and stop creating errors03:22
wallyworldlol03:25
anastasiamac_oooh, axw, wallyworld: what about juju/go4?.. talking about lugging things around :D03:26
axwanastasiamac_: I think rog is using that03:27
anastasiamac_k :) 4? isn;t he on the tip of go7?03:27
wallyworldthat's not go the sdk03:28
anastasiamac_wallyworld: axwwhat about juju/ratelimit? is it rog too?03:43
wallyworldnot sure, would need to search the code base03:44
thumper:-(  cmd/juju/controller/listblocks_test.go           |  124 -------------03:55
thumper 49 files changed, 1014 insertions(+), 1238 deletions(-)03:55
thumpertim@elwood:~/go/src/github.com/juju/juju (block-cmd-rework)$ git diff master | wc -l03:55
thumper322903:55
thumperwallyworld: sorry03:55
wallyworldyay!03:55
wallyworldand i know you'r enot sorry at all03:55
thumperhttp://reviews.vapour.ws/r/5600/03:56
thumperwallyworld: just adding qa steps03:56
thumperwallyworld: it has been epic03:58
thumperwallyworld: that review addresses all the block related commands, and a few drive by fixes03:59
thumpersome of which were necessary, others not entirely so03:59
wallyworldok, am finishing something, then looking at andrew's diff then yours04:00
* thumper nods04:00
thumperthe change that introduced the import cycle that I had to fix was moving the CmdBlockHelper out of cmd/juju/common04:00
thumperit was the only file that brought gocheck into the package, which made gocheck a non-testing dependency04:00
thumpermoved it into testing04:00
thumperthat introduced an import cycle in the api package04:01
thumperdue to the mongo.Version structure over the api04:01
anastasiamac_thumper: what an adventure! thank you :D04:06
veebersthumper: That PR removes the block, unblock commands right?04:08
thumperveebers: it renames04:08
thumperveebers: block -> disable-command04:08
thumperunblock -> enable-command04:09
thumperblock list -> disabled-commands04:09
veebersthumper: sorry yeah renames. That's going to need a change to the assess_block CI test when it lands04:09
* thumper nods04:09
veebersthumper: I should be able to rough it up now, but not sure if it'll land before/when your PR does04:10
thumperveebers: the branch is still being reviewed04:10
veebersthumper: ack, I'll get something ready in the wings04:12
* thumper relocates again04:38
axwwallyworld: I've just updated my branch, I think I've got all the loose ends now. now I need to write a bunch of tests05:01
wallyworldaxw: ok, almost ready to look05:01
wallyworldaxw: looking now. i've add to my PR to address the comments, but also added machine and core count to show-controller. a difference with current behavior is that it does an AllModels() api call to get the model info to stick in the output, rather than simply only looking at locally cached info05:11
wallyworldand that's more correct anyway05:11
wallyworldbut it doesn't currently update the local yaml though05:12
axwwallyworld: doesn't update the cache for list-controllers you mean?05:15
wallyworldaxw: doesn't update the local model yaml05:15
wallyworldso this is for show-controller05:16
wallyworldpreviously, show controller would only look at local models yaml05:16
wallyworldnow it makes an api call05:16
wallyworldas it needs to get the machine and core info05:17
wallyworldand so it also then gets the latest models as well05:17
axwwallyworld: looks fine, but can you please get OCR to check too. it's over 500 lines now05:18
wallyworldffs05:18
wallyworldi was going to propose separately05:18
wallyworldshould have done that05:19
wallyworldat the expense of velocity05:19
menn0anastasiamac_: I improved that test as a result of your comment05:19
axwwallyworld: I only say because I'm actually having trouble keeping track of all the changes05:19
anastasiamac_menn0: veni, vidi, vici05:19
axwtiny brain05:19
anastasiamac_menn0: i've commented and stamped \o/ looks awesome05:19
wallyworldaxw: you don't need to import errgo just to call Cause() - juju/errors.Cause() will work IIANM05:23
axwwallyworld: ah ok, thanks05:24
wallyworldaxw: errgo and errors both define their causer interface identically05:25
axwwallyworld: it might be more useful for you to pull the branch and test it, than to try and review it as it is. I'll break it up later on and propose a few bits separately. mostly it'll have to be done as one though05:27
axwif you don't have time, that's fine05:28
wallyworldaxw: ok, can do. the fine detail of the auth stuff is a little out of my comfort zone05:28
axwwallyworld: no worries. me too, hence why it's taken so long :)05:28
wallyworldhave to wait for a 2nd review anyway :-/05:28
wallyworlddef get rog to look :-)05:29
wallyworldaxw: axtually, looks like there will be a conflict, wanna rebase before i pull?05:29
axwwallyworld: sure, just a minute05:30
axwwallyworld: pushed. I just realised I missed something, which is that "juju logout" will need to clear any cookies for the controller. I might do that in a follow up though05:34
wallyworldok05:34
wallyworldaxw: appears to work ok. password change, deleting go-cookies etc. i get prompted as i would expect for the password and then from there it seems to use the macaroon ok. any thing in particular else to test?06:01
axwwallyworld: not really, mostly just wanted to you sanity check. does it feel natural to you? as opposed to how it was, where you would get told that your login expired, and you had to run "juju login"06:02
axwwallyworld: FYI, I also tested "juju register" to make sure you get logged in automatically still06:02
wallyworldaxw: yeah, much nicer to have you simply prompted06:02
axwcool06:02
wallyworldIMHO06:02
axwwallyworld: thanks. updating tests now, will propose it for real later on06:02
wallyworldaxw: eg with github, if my cached password expires, it simply just prompts again06:03
* axw nods06:03
wallyworldanastasiamac_: if you get a chance, i'd love a 2nd look at this, it's a fraction over 500 lines sadly http://reviews.vapour.ws/r/5594/06:12
babbageclunkwallyworld: Morning!06:14
wallyworldhey06:14
babbageclunkwallyworld: Any idea what this maas user data that thumper's talking about is?06:15
babbageclunkwallyworld: I can't find it.06:15
wallyworldno :-(06:15
wallyworldi'm sure they maas guys told me to use tags06:15
babbageclunkSeems kind of unlikely that there's yet another way of searching for machines.06:16
wallyworldyeah06:16
babbageclunkok cool - I was feeling a bit silly there06:16
wallyworldyou and me both06:17
wallyworldmight pay to ask a maas person directly06:17
babbageclunkare there any in a convenient timezone?06:26
babbageclunkI'll ask allenap once he's up.06:28
=== frankban|afk is now known as frankban
anastasiamac_wallyworld: was afk - kids/school/lif (apparently, there is that). I can have a look after dinner/kids bedtime :D is it k?07:10
wallyworldanastasiamac_: sure, am working on the next bit in preparation07:10
anastasiamac_wallyworld: \o/ tyvm :) hope u have dinner in btw too :D07:11
wallyworldyeah, at some point :-)07:11
axwwallyworld menn0: to what degree is migration supposed to be stable in 2.0? there's some macaroon stuff in there that will need to change08:04
wallyworldaxw: you mean model migration or schema changes for upgrades?08:04
axwwallyworld: model migration08:05
wallyworldaxw: we have some lattitude - an external tool is being used to handle parts of the model the 2.0 agent doesn't know about08:06
wallyworldassuming the stuff you need done doesn't make 2.008:06
axwwallyworld: well I'm going to have to disable the ability to use macaroons to do the migration, so it would only be usable if you have a password in accounts.yaml08:07
=== MmikeM is now known as Mmike
frobwaremgz: you about?08:48
frobwaremgz: ah, no... you're crewing...08:48
voidspacebabbageclunk: ping10:35
babbageclunkvoidspace: pong10:38
* rick_h_ pokes head up and looks around13:04
macgreagoirfrobware: You have 5 mins? Wanna stay in that HO?13:11
frobwaremacgreagoir: I'm still in there13:11
macgreagoirfrobware: Hmmm... I can't see you...13:12
=== frankban is now known as frankban|afk
perrito666bbl lunch15:55
=== frankban|afk is now known as frankban
=== frankban is now known as frankban|afk
alexisbperrito666, ping17:28
perrito666alexisb: pong17:29
alexisbperrito666, have you pick up a bug yet?17:30
perrito666yes I have, https://bugs.launchpad.net/juju/+bug/161619717:30
mupBug #1616197: juju restore-backup error <backup-restore> <juju:In Progress by alexis-bruemmer> <https://launchpad.net/bugs/1616197>17:30
perrito666but I am in time to switch if you have something not backup related17:30
perrito666anyting at all17:31
alexisbperrito666, ok proceed thank you17:31
alexisbnope that is a good one for you to take17:31
perrito666k, Ill proceed17:31
* perrito666 puts the hazmat suit again17:31
thumpermorning folks20:28
perrito666going to fetch some meds, bbl20:33
alexisbmorning thumper20:39
=== natefinch is now known as natefinch-afk
thumperalexisb: morning21:16
perrito666Thumper sorry for whatever I broke that you where saying last night21:18
perrito666I already forgot what it was21:18
thumperperrito666: already fixed and landed21:19
thumperreminder though not to use external package structures over the api21:19
thumperapiserver/params was using mgo.Version21:19
thumperno21:19
thumpermongo.Version21:19
thumperanyway21:19
thumpernow it isn't21:19
veebersthumper: what's the easiest way to find out which beta version the block command was added to?21:20
thumperveebers: um... block has been around for a long time21:21
thumperdo you mean my change to block?21:21
veebersthumper: oh no, I meant originally added, this is for updating the ci tests :-)21:21
veebersthumper: all good, that pretty much answers my question21:21
thumperif `juju block list` returns non-zero, it isn't found and has new commands21:21
thumperveebers: it is in 1.2521:22
veebersthumper: ah right, thanks :-)21:22
thumperBOOM!21:24
thumpersuccess on that test21:24
menn0wallyworld: ping21:28
wallyworldhey21:28
veebersthumper: Am I right in my reading that you changed all-changes -> all for enable-command? (the PR says you did but the help text still says 'all-changes')21:35
thumperveebers: yes, "all-changes" is now just "all"21:37
veebersthumper: Cool, I'll file a bug for the help text21:38
mupBug #1620830 opened: destroyEnvSuite.TestDestroyEnvironmentCommandEFlag interface is nil <ci> <intermittent-failure> <panic> <regression> <unit-tests> <juju-core:Triaged> <https://launchpad.net/bugs/1620830>21:40
mupBug #1620832 opened: github.com/juju/juju/featuretests MSpan_Sweep: bad span state <ci> <go1.6> <intermittent-failure> <regression> <unit-tests> <juju:Triaged> <juju-core:Triaged> <https://launchpad.net/bugs/1620832>21:40
* redir goes to make tea and change computers, brb21:42
thumpermenn0 or other reviewer? http://reviews.vapour.ws/r/5609/22:25
menn0thumper: looking22:26
menn0axw: ping me when you're around regarding migrations + macaroon issues22:27
thumperupdate juju/testing dep http://reviews.vapour.ws/r/5610/22:29
thumperone line review22:29
thumperwallyworld: ^^22:31
menn0thumper: more ship its22:35
thumpermenn0: cheers22:36
axwmenn0: I'm here now22:55
thumperreview board not picking this up for some reason.https://github.com/juju/juju/pull/617223:12
thumperhttps://github.com/juju/juju/pull/617223:12
thumperwallyworld:  access stays unknown... http://pastebin.ubuntu.com/23143762/23:40
thumperhmm... older controller23:41
thumperperhaps wasn't sending right data back23:41
wallyworldthumper: correct. did you even read the messsage23:41
wallyworld--refresh23:41
thumpersee the last line23:41
thumperwith --refresh23:41
wallyworldoh23:41
wallyworldin that case the controller is sending back the access, i'll have to understand the repor steps23:42
thumperbootstrapping newer23:42
wallyworldthumper: at bootstrap, access and version is filled in, it should never say unknow unless tyoy are running an older version23:43
thumperwallyworld: I think it was probably due to it being an older controller23:44
wallyworldmy guess is you had an older version bootstrapped23:44
wallyworldyep, that won't work23:44

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