anastasiamacthumper: wallyworld: PTAL https://github.com/juju/juju/pull/9125 - fix for critical00:11
anastasiamaci've proposed against 2.3 and will forward-port once landed.00:11
anastasiamacbabbageclunk: veebers: any chance u could review ^^00:42
babbageclunkanastasiamac: sure00:42
anastasiamacbabbageclunk: my hero \o/ tyvm00:43
wallyworldveebers: sorry, just got off the meeting. i don't know what the issue might be off hand00:45
babbageclunkanastasiamac: approved!00:45
veebersanastasiamac: sorry was at lunch :-\00:52
anastasiamacveebers: how could u?00:52
thumperanastasiamac: I'm looking to forward port 2.3 into 2.401:20
anastasiamacthumper: k but my stuff will be there shortly (and on develop)01:20
anastasiamacthumper: m happy to review, once u propose if u want01:21
thumperanastasiamac: it is in 2.3 now01:21
thumperI noticed it there01:21
anastasiamacyep. i've pr'ed against 2.3 originally, this morning01:22
anastasiamac2.4 and develop landing is happening now01:22
anastasiamacgofmt against develop seems to be stricter?01:22
anastasiamacthumper: and merged into 2.401:25
veeberswallyworld: I believe one of my issues re: cleanup is conditionally setting statusOps (i.e. if caas use cloudcontainer key, otherwise unit) so removeOps is dying (I haven't touched that yet).01:26
thumperanastasiamac: I've found a problem with your patch :(01:26
thumperanastasiamac: was QAing it01:26
anastasiamac:( what's it>01:26
veeberswallyworld: is that the way forward, or instead of having the absence of a status via key be the decider should the caas check happen at the GetStatus end?01:26
thumperanastasiamac: I deployed keystone like the bug did01:26
thumperthen went "juju config keystone vip"01:26
wallyworldveebers: quick HO?01:26
thumperthere used to be no value01:26
thumpernow there is a blank line01:26
veeberswallyworld: sure thing01:27
veeberswallyworld: standup?01:27
veeberswallyworld: one sec, I smell smoke . .01:27
anastasiamacthumper: ahhh, yes coz it's Println... should it not be? just Print..01:27
thumperanastasiamac: confirmed with 2.2.6 cli01:27
thumperanastasiamac: just nothing01:27
thumperanastasiamac: I'll pastebin01:28
anastasiamacthumper: k. no need... i understand01:28
veebers;all good, no smoke or stroke01:28
thumperanastasiamac: always good to be explicit... https://paste.ubuntu.com/p/sgszfhQTpQ/01:28
thumperwallyworld: when do you want to talk watchers01:29
wallyworldotp now, maybe soon01:29
anastasiamacthumper: k. so I have added a *Println as jam's recommendation was to append newline tot he value01:29
thumperanastasiamac: which we should do for non nil values01:29
anastasiamaci'll go back and remove the newline :D thumper01:30
anastasiamacthumper: ack01:30
anastasiamacnewline wihen there is value and none where nil value01:30
* thumper nods01:30
anastasiamack. so hold off porting 2.3 into 2.401:30
thumperyep will do01:30
anastasiamaci'll propose agaisnt 2.3 ifrst01:30
* thumper nods01:31
babbageclunkwallyworld: trivial review for bumping up the timeout? https://github.com/juju/juju/pull/912801:35
wallyworldbabbageclunk: i wonder why this change got include in the lock file gopkg.in/goose.v2/testservices/neutronmodel01:37
wallyworlddoesn't seem related01:37
babbageclunkwallyworld: yeah, I was about to put a comment on the PR about that.01:37
babbageclunkI'm not sure. I was tempted to remove it, but presumably that would just mean that the next person to do a dep change would include it.01:38
babbageclunkkelvinliu: ^?01:38
kelvinliubabbageclunk, how's going01:38
babbageclunkkelvinliu: I got a weird change in Gopkg.lock when I updated an unrelated dependency, just wondering if you had any ideas about it.01:40
babbageclunkkelvinliu: https://github.com/juju/juju/pull/9128/files#diff-bd247e83efc3c45ae9e8c47233249f18R182001:41
kelvinliubabbageclunk, i guess if it's a dep of github.com/juju/pubsub01:42
babbageclunkNo, I don't think it's related.01:43
babbageclunkMaybe it's version differences between different people's dep binaries?01:43
anastasiamacthumper: so ha.. if the value is nil then we would not print a newline... if a value is '' (empty string), we would?... previously, say in 2.2.6, we did not... in fact, in 2.3- we have never had a newline...01:44
anastasiamac(unless it was included as the actual value for the setting)01:44
thumperanastasiamac: I'd say if the value is the empty string then yes to newline01:44
thumpernil or missing is no value01:45
* thumper thinks01:45
thumperthis is a break in our behaviour01:45
anastasiamacadding newline? yes01:45
thumperwe broke STSs scripts01:45
thumperI'm wondering if adding a new line will continue to break their scripts01:45
anastasiamacit could...01:46
kelvinliubabbageclunk, https://github.com/juju/juju/blob/develop/provider/openstack/local_test.go#L3701:46
anastasiamacadding newline was not required for the original fix anyway...01:46
thumperanastasiamac: I think we shouldn't add the new line now01:46
thumperanastasiamac: perhaps leave a note for Juju 3.001:46
kelvinliubabbageclunk, i think it's introduced before ur PR, but someone forgot to run make dep01:46
thumperto add a new line01:46
kelvinliubabbageclunk, sorry, it's make rebuild-dependencies01:47
babbageclunkkelvinliu: ahh, yeah, I think that makes sense.01:47
anastasiamacthumper: ack, i'll add the note, card in 3.0 list and a bug :)01:47
babbageclunkwallyworld: I think that's the explanation, I'm ok if you want me to pull it out of the PR.01:47
wallyworldbabbageclunk: if it's to fix a previous error then that seems ok to me01:48
babbageclunkIt seems like that section in Gopkg.lock is going to change in surprising/unrelated ways when someone changes a dependency, since anytime someone adds an import of a previously-unimported package from an existing dependency it's going to be added here.01:51
babbageclunk(and I don't think it would be reasonable to insist that people run rebuild-dependencies when they haven't really changed them)01:51
anastasiamacthumper: PTAL https://github.com/juju/juju/pull/912901:54
thumperanastasiamac: a suggestion and a question02:01
thumperbabbageclunk: I hit the session closed bug again in the lease manager02:02
thumperbabbageclunk: https://paste.ubuntu.com/p/84rSqXGsjg/02:02
* thumper goes to make coffee02:04
babbageclunkthumper: ah, thanks - looking now02:07
anastasiamacthumper: and updated PR02:08
* anastasiamac crosses fingers and makes tea02:08
babbageclunkthumper: ok - I guess the fix is to add a waitgroup to the manager and make sure it waits for all of the goroutines to be finished before stopping. Sound right to you?02:09
babbageclunkthumper: how do I reproduce it - running state tests under stress?02:09
* babbageclunk will assume that until further notice.02:10
thumperbabbageclunk: it is fucking hard to reproduce02:12
thumperbabbageclunk: re waitgroup, yeah I think something like that may be necesssary, I'm going to take a peek02:13
thumperbabbageclunk: got time to talk this through?02:17
babbageclunkthumper: sure02:20
wallyworldthumper: here's a PR to address the charm version issue https://github.com/juju/charm/pull/25802:29
veeberswallyworld: I'm either confused or my plan won't work, the unit workload status is initiall set at unit start (allocating), the caas provisioner updates the details (i.e. started container), if the unit status goes to 'unknown' or maintenance  state (i.e. agent error) then the cloud container status will never hit the criteria to update the status (unless it errors).02:29
wallyworldor anastasiamac ^^^^^ :-)02:30
veebersresulting in a status of 'maintanence, message: Waiting for mysql container' while the cloud container status is 'Started container'02:30
* anastasiamac looking02:30
anastasiamacwallyworld: could u ptal 9129, just in case?02:30
wallyworldveebers: right, remember this is a stop gap until we thread stuff through. if the charm sets status to maintenance, we won't show container status unless it's an error and we choose to use it02:31
thumperwallyworld: I thought the entire purpose of this was to not log a warning message02:31
veeberswallyworld: ack, but it also means that it'll never go to an 'active' status as that info is now piped though cloud container status (which won't overwrite any unit status)02:32
wallyworldthumper: 2 things 1. it was logging a warning unnecessarily due to a git issue, 2. IMO we do want to inform the user if the version generation fails, so i have improved the message02:32
wallyworldi think the bug was due to some bogus message wthout enough info02:32
wallyworldif there's a genuine error we should surface it02:32
thumperI would say it should be informational to the user, it isn't a warning02:32
wallyworldi can chabge it to INFO but that will be hiddn by default02:33
wallyworldit soimething fails surely that's a warning02:33
thumperso... if I understand this, if it is in revision control and it hits an error we get a warning02:33
thumperwhat was the problem they were seeing?02:33
thumperwhy was there an exit code of 128?02:33
wallyworldbecause "git describe" failed02:34
wallyworldit couldn't generate a version sha02:34
wallyworldand we were not surfacing the true error string02:34
wallyworldhence the changed cmd.exec code also02:34
wallyworldit was failing because there were no tags available for it to use in the output02:35
wallyworldso we needed to be passing in --always02:35
wallyworldso that it would at least print the sha (without tag)02:35
wallyworldand without complaining02:36
wallyworldmake sense?02:36
wallyworldveebers: for now she should also special case "active"02:37
veeberswallyworld: ack, have done so now :-)02:37
veebersif cloud container active use it, currently have && unit status == maintanence but I think that's a bit much02:37
wallyworldyeah it is02:38
wallyworldcould have been error->active02:38
wallyworldeg if we fix the stroage claim it will come good02:38
veebersack, thanks wallyworld!02:38
thumperwallyworld: can you jump into our 1:1?02:39
veeberswhat did we ever do without colour log output :-)03:09
babbageclunkthumper: I ended up reimplementing the waitgroups fix so I could check my test passed with it. I might just push it as a PR if that's ok? :)03:17
anastasiamackelvinliu: veebers just got a 'recipe for target 'dep' failed'... m assuming it's not related to my change, so m !!build!!Ing but would love to know what's going on.. http://ci.jujucharms.com/job/github-check-merge-juju/3211/console03:23
veebersanastasiamac, kelvinliu I *think* that's a network connectivity thing, it couldn't hit golang.org for some reason, I suspect it's transient; lets see how the rerun goes03:25
anastasiamacthumper: newline saga landed in 2.3.. feel free to move 2.3 to 2.4 :D03:25
thumperanastasiamac: ack03:26
anastasiamacveebers: \o/ thank you :) let's hope...03:26
kelvinliuveebers, anastasiamac yeah, let's see the re-run.03:26
wallyworldkelvinliu: i had to make a change to charm.v6 to fix some version string issues. can you add to your todo list making a charm build PR with the same fixes https://github.com/juju/charm/pull/25803:35
kelvinliuwallyworld, sure03:35
wallyworldgr8 ty03:35
babbageclunkthumper: review pls? I just extended the test to check for both claims and ticks. https://github.com/juju/juju/pull/913103:52
* thumper looks03:52
babbageclunkthumper: oh hang on - still pushing that last bit.03:52
babbageclunkbally gometalinter03:52
babbageclunkthumper: ok done03:54
* thumper has a failing test but I think it is a lxd test isolation bug03:57
anastasiamacthumper: fwiw, i'd rather forward-port my hcange manually.. there is a new test that was added in 2.4 that will fail so test needs to b adjusted too.03:57
thumperanastasiamac: ok, do you want to take my change as well?03:57
thumperif you do the 2.3 merge, it is just yours and mine03:57
anastasiamacthumper: honestly? no... but i can if u command ;)03:58
thumpermine is only 4 lines and clean mege :)03:58
thumperpretty please03:58
anastasiamacthumper: k03:58
* anastasiamac saluts03:58
anastasiamacsalutes even?*03:58
thumperanastasiamac: thank you04:00
wallyworldbabbageclunk: were you going to join us in this meeting?04:07
babbageclunkoh yes - send me an invite?04:07
anastasiamacthumper: merge PR - https://github.com/juju/juju/pull/913404:23
thumperanastasiamac: lgtm04:33
veeberswallyworld: you have a moment to talk migration_export? Do I need to add to juju/description a SetCloudContainerWorkloadStatus or so (so I can export the #container status data) or should I finagle it in with the existing SetWorkloadStatus()?04:45
wallyworldon a call give me 504:45
anastasiamac'finagle' is totally a word too04:45
veebershmm, I didn't realise it meant in a devious manner, I thought it meant in a tricky manner :-)04:46
anastasiamac'devious, dishonest' :( i did not know either ...04:47
veebersanastasiamac: I take it your 2nd build run went fine?04:49
anastasiamacveebers: yes, thnx :) must have ben connectivity like usaid...04:49
wallyworldbabbageclunk: here's a small dep change https://github.com/juju/juju/pull/913504:58
wallyworldveebers: free now04:58
babbageclunkwallyworld: looking04:59
veeberswallyworld: sweet, shall we HO ?04:59
veebershah omw04:59
babbageclunkwallyworld: approved. You and veebers should hang out.05:00
wallyworldty. we alrerady are :-)05:00
babbageclunkhey thumper are you happy with https://github.com/juju/juju/pull/9131?05:01
veeberslol babbageclunk the match maker ^_^05:01
veebershey, I know a cool guy. You should hangout and talk migration exports some time05:02
babbageclunkyou crazy kids05:02
anastasiamacnoticed, i've introduced a test where ordering matters on develop... working on a fix now :(05:27
veeberswallyworld: FYI https://github.com/juju/juju/pull/9081 I'll need to add another test for the conditional cloudcontainer/unit status usage too05:29
wallyworldok, ty05:29
wallyworldwill look soon05:29
veebersack, thanks05:30
anastasiamactrivial change, could someone review plz? https://github.com/juju/juju/pull/913605:41
wallyworldanastasiamac: looking05:48
wallyworldanastasiamac: also, i'm an idiot https://github.com/juju/charm/pull/25905:48
anastasiamacwallyworld: ta :)05:48
* anastasiamac looking too05:48
anastasiamacthumper: 2.4 branch is upto date as far as urs and my changes :) just wallyworld's one needs to come in and we r golden i *think*05:51
wallyworldi'll have to wait for snap to build i think soi can copy across05:52
babbageclunkwallyworld: can you take a look at this https://github.com/juju/juju/pull/9131, thumper has forsaken me.05:55
wallyworldbabbageclunk: i have some questions aboput the test06:01
wallyworldi think it may be racy and we need some extra selec/retry loops06:02
wallyworldwe use retry loops in other tests with an attempt strategy06:02
wallyworldvinodhini: how goes the export bundle fix?06:43
vinodhiniwallyworld: i pinged u this morning that i pushed the commit. in canonical #juju06:44
wallyworldvinodhini: ah, you said "wallyworld." not "wallyworld:"06:45
wallyworldso i didn;t get pinged06:45
wallyworldi'll take a look06:45
vinodhinii am sorry wallyworld06:46
wallyworldtis ok :-)06:46
vinodhiniwallyworld : do u need that to be changed ?06:54
wallyworldi'm leaving a few small comments06:54
vinodhinithat check* - now it checks "instance"06:54
vinodhinioh ok.06:54
wallyworldvinodhini: so yeah, just a few small things, but the relation scope test appears inadequate as written I think, but i could be wrong about that06:59
vinodhiniwallyworld u mean the unit test for relationscopeskipping ?07:00
wallyworldwe need to ensure that the would be an error if skiprelationscope were nnot true07:00
wallyworldi'm not sure that is the case but am not 100% sure07:00
vinodhiniwallyworld : I agree it is in adequate. But can we look at it in a way that if the relation scope is missing it just skips that.07:05
wallyworldbut we need to prove it in a test07:05
wallyworldit failed before, so we do a fix. but unless we test the fix, we are not doing the job07:05
wallyworldall bug fixes should have tests to catch the rror07:06
vinodhiniwallyworld I get what u say. like how it is proven in other tests07:11
vinodhinithis just going to hit the error : missing relation scope for wordpress:db mysql:server and mysql/0' if we comment that SkipRelationScope07:11
wallyworldvinodhini: if we can comment out the skip=true and the test fails and we add skip=true and it passes that is sufficient07:13
vinodhinithere needs to be still be some entity to prove when we skip that check07:13
vinodhiniyes it happens now.07:13
wallyworldvinodhini: also, FYI, i fixed an issue with charm version string. part of the issue was that there were more cases of errors being ignored https://github.com/juju/juju/pull/913507:13
wallyworldok, if the test fails that way then i think it is ok07:14
wallyworldi just wasn't sure it would07:14
vinodhiniwallyworld i do that PR charm version fix u did with develop.07:15
vinodhinithis timedelay issue i am working on 2.4.207:15
wallyworldvinodhini: i already fixed it just now, just wanted to show you so you knew about the issue i had to fix for learning07:16
vinodhiniok wallyworld this is something i did in June - charm version07:19
wallyworldyeah. it was broken but no one uses it yet really so no one noticed07:20
vinodhiniwallyworld :  so the unit test is for now ok. i can address the other 2 comments07:20
wallyworldi think so07:20
vinodhiniwallyworld : can u chk again now the PR ?07:53
vinodhiniso we can land it.07:53
vinodhinii have done state unit tests07:53
=== alephnull_ is now known as alephnull
manadartNeed a review of https://github.com/juju/juju/pull/913909:02
manadartjam, stickupkid ^ I am actually still doing the QA system testing steps, but some feedback in the interim...09:03
stickupkid"cmd/jujud/updateseries/updateseries_test.go:1::warning: file is not goimported (goimports)09:08
stickupkidservice/systemd/service.go:1::warning: file is not goimported (goimports)"09:08
stickupkidmanadart: you need to fix those files09:08
manadartstickupkid: Ack.09:09
stickupkidmanadart: nice one09:11
stickupkidmanadart: so 2.0.x LXD didn't send the server name, so when we're doing zone distribution the machine zone would be empty for all servers, even though there was one, it caused it to show the error message09:22
manadartYes, I surmised this when I saw your patch yesterday.09:23
stickupkidmanadart: the fix was easy enough, the issue is, now we've got the same issue as before, where by the it can't find the lxd-config file for the network setup09:23
stickupkidso i suspect something is amiss09:23
* manadart sighs.09:23
stickupkidmanadart: got a sec?10:36
manadartstickupkid: Yep. HO.10:37
* rick_h_ grumbles about relogging into irc and such11:38
rick_h_manadart: do you have time to review https://github.com/juju/juju/pull/9122 today please?11:40
manadartrick_h_: Yes.11:41
rick_h_manadart: ty!11:41
stickupkidnice, that's an annoying message in the status output11:43
manadarthml: Any chance you can look at https://github.com/juju/juju/pull/9139?12:37
hmlmanadart: sure.12:37
cory_fuHas anyone seen "could not get environ: Get Unable to connect to:" from Juju when provisioning machines on lxd before?  http://p.ip.fi/FWSs13:36
stickupkidcory_fu: edge or 2.4.x?13:46
cory_fustickupkid: 2.4.1, it would seem.  Note: I'm asking this on behalf of someone in #conjure-up13:47
stickupkidcory_fu: i've never seen it, was just wondering what code base to at least look at13:47
stickupkidmandart: you see that?13:48
rick_h_cory_fu: that seems like a network issue to the lxd api server. Is this already bootstrapped and failing on provisioning new containers? Or is this during bootstrap?13:57
cory_furick_h_: Link is a full juju status.  It's late in a deployment of openstack (machines 15 and 16), and, while retry-provisioning did nothing, remove-unit + add-unit worked and got them moving.  Maybe all that's needed is a retry, or even just getting retry-provisioning to do the right thing13:59
cory_fuObviously, there's not much juju can really do if it can't talk to lxd, but Juju does know that clouds are unreliable so should be able to recover more gracefully14:00
rick_h_cory_fu: yea, I wonder if retry-provisioning would work in this case to try again and pick up the machines?14:09
cory_furick_h_: It didn't14:09
cory_fuSeems like it should14:09
rick_h_cory_fu: :( then yea seems like it should14:09
hmlmanadart: unrelated to your pr, i found a false positive with the jujud-updateseries cmd15:14
hmlmanadart: just a heads up, not sure if it will impact the new work15:14
hmlmanadart: https://pastebin.ubuntu.com/p/hmfPPdgPMm/15:15
hmlmanadart: services are reported restarted when they weren’t15:15
manadarthml: I can't see the issue in the paste. They are loaded and active...15:20
hmlmanadart: they’ve been running for 5 min, but i ran the jujud-updateseries command 30sec before15:21
hmlmanadart: i’m pretty sure they would have failed during the restart… if really a restart, because the series of the link to tools wouldn’t have matched the current versions15:22
hmlmanadart: since i ran the command without doing the upgrade to check the links15:22
manadarthml: Do you still have that environment up?15:25
hmlmanadart: yes15:25
manadarthml: Can we do a quick HO?15:25
hmlmanadart: sure - omw15:25
aisraelcory_fu: I just ran into a case where a charm's reactive flag was cleared before a second @when was invoked, leading me to discover register_trigger (which is very useful). I had a @when('config.changed') that wasn't running, presumably, because the flag was cleared. Was there a change in that behavior sometime recently, or did I just get lucky before?15:55
jamguild: if anyone is around, https://github.com/juju/juju/pull/9140 is a potential fix for the errors they are running into in bug #1789211 I still need to grab VMWare credentials and actually test that it fixes the problem16:01
jambut it does seem likely to fix it.16:01
cory_fuaisrael: You must have just gotten lucky.  It's been the case that clear_flag re-checks the queue for some time, though I think it's a mistake16:01
rick_h_kwmonroe: cory_fu magicaltrout zeestrat bdx 37min warning for juju show time. I've started up an agenda/notes doc https://discourse.jujucharms.com/t/juju-show-38-wed-aug-29th-17-00-utc/20216:24
rick_h_feel free to add anything16:24
cory_fukwmonroe: You wanted me to mention the Azure integrator charm?16:25
cory_fukwmonroe: It doesn't look like the support for that has been released in the k8s charms yet16:26
kwmonroecory_fu: yeah, mention it anyway.  thataway, i can be like "and now, here's the vsphere integrator, which is totally functional and released in the k8s charms.16:33
cory_fukwmonroe: I'm happy to join, but it might flow better for you to just talk about both16:34
kwmonroeroger that cory_fu16:34
cory_fukwmonroe: Added to the Discourse16:40
rick_h_cory_fu: kwmonroe put into the main notes linup16:45
rick_h_cory_fu: kwmonroe bdx magicaltrout zeestrat https://hangouts.google.com/hangouts/_/dp4hqk72ubd7bllbbvr3hj2xhae is the HO url16:49
rick_h_for folks that want to watch the stream at home without joining https://www.youtube.com/watch?v=i5TGTiwXrmc16:49
kwmonroerick_h_: i have a problem making my discourse post about k8s/vsphere... the top K8s category is about deploying workloads on k8s (juju add-k8s).  where should posts about traditional k8s charms (like cdk) go?  perhaps Charming -> Kubernetes to follow the big data subcategory model?  i think we'll confuse people if we (read: you) don't get this right.18:48
rick_h_kwmonroe: bah, so I'm not a fan of all the k8s stuff in root ATM and all pinned. I agree there's some room for "best place".18:50
rick_h_Basically not sure for now. I'll have to get wallyworld on board with a better setup18:50
kwmonroeroger dodger18:51
stickupkidhml: thanks for QA'ing my PR :D20:12
hmlstickupkid: np20:14
bdxexcellent juju show - sorry I missed -  druid looks sweet!!21:03
rick_h_bdx: :) we missed you dude21:05
bdxim hustling trying to get this controller out the door21:07
bdxgetting cut by the "I'm the only person to try and do it" knife I feel21:09
bdxtrying to use jujucharms.com as identity for my own controller is not as straight forward as I was hoping21:10
bdxis there something I'm missing here ?21:10
bdxconcerning using jujucharms.com for identity21:10
rick_h_bdx: honestly I'm not sure. I saw your reply about the 509 error and I'd not expect that but I know we put some magic in Juju for "juju login jaas" to work21:11
rick_h_bdx: the guys that work on that stuff are in the EU and EOD atma21:11
bdxgot it21:14
bdxrick_h_: I think the problem is "no way to register external users on controller using jujucharms.com identity"21:23
rick_h_bdx: right, that's the issue to figure out. Juju knows about jaas baked in so any external users just login and they're off21:24
bdxI see21:24
rick_h_bdx: so I'm not sure if you can/need to preseed the controller info? Or if there's another tool/bit to it21:24
rick_h_bdx: the good news is that mhilton that wrote the post for you wrote most of candid and works on that stuff21:25
rick_h_bdx: so you're on the right track there and just need to prod them for moving forward another couple of steps in the doc he started21:25
bdxok perfect21:25
babbageclunkwallyworld: morning! Sorry I missed your qs about the test for that PR - I've put some answers in now. Alright for me to hit merge on it?21:37
veeberswallyworld: I'm looking for a better way, currently want to add something like "func (u *Unit) SetCloudContainerStatus(...)" so I can write this test (i.e. to mirror Unit.SetStatus() so I can prime statuses to then test the results22:06
wallyworldveebers: what do we do currently in the tests? i think we use UpdateUnitOps or something22:09
veeberswallyworld: ah good point I'll have a deeper look now22:09
veeberswallyworld: yeah that should do the job, thanks122:10
babbageclunkwallyworld: mind if I land that pr given the answers?22:13
wallyworldbabbageclunk: which PR? ECONTEXT22:14
babbageclunkwallyworld: sorry https://github.com/juju/juju/pull/913122:14
* babbageclunk lolles at ECONTEXT22:14
wallyworldbabbageclunk: yes, go for it. thanks for clarifying. i should have looked at the content of those methods sorry22:15
babbageclunkwallyworld: thanks! No worries!22:16
babbageclunkWhat the... how does running a state test suite require building apiserver/facades/controller/migrationtarget?!22:17
wallyworldrick_h_: that export bundle branch compile error you had - it has to be something in your setup right?22:33
wallyworlddid you run the make target which removes the vendor folder?22:33
wallyworldmake godeps22:33
wallyworldveebers: veeeeeebeeeeers23:17
rick_h_wallyworld: I didn't get anchance to go back at it.23:26
wallyworldrick_h_: no worries23:30
veebersOnly an agent can be status.Allocating right? A unit cannot?23:34
thumperwallyworld: any update on the firewaller test problems I see on my branch?23:42
wallyworldthumper: i'm still finishing all my morning meetings. almost ready to start23:43
* thumper nods23:43

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