/srv/irclogs.ubuntu.com/2016/10/27/#juju-dev.txt

wallyworldaxw: review done01:43
axwwallyworld: thanks01:45
menn0thumper: could you take another look at this please? https://github.com/juju/juju/pull/649202:10
thumperyep02:11
thumpermenn0: just one thing02:17
menn0thumper: ok thanks02:17
menn0thumper: good suggestion. I even looked for an existing file to put that func in but missed this one.02:18
* thumper nods02:18
blahdeblahAnyone able to tell me what this means?  Google has no hits.05:08
blahdeblahUnable to get the model status from the API: no such request - method ModelManager(2).ModelStatus is not implemented (not implemented).05:08
blahdeblahSeeing this using 2.0.0 client against a 2.0 rc3 controller05:09
blahdeblahwondering if it's an API incompatibility between the two05:09
blahdeblahIf I downgrade to the 2.0-rc3 client, I get this instead: "ERROR cannot update account information: unqualified user name "admin" not valid"05:31
=== frankban|afk is now known as frankban
menn0blahdeblah: that does indeed appear to be an API incompatibility09:06
menn0blahdeblah: the error means that the client is trying to call ModelStatus on the ModelManager facade but the API server doesn't have it09:06
menn0blahdeblah: I've done some digging and it looks like this was changed just before 2.0 final09:07
menn0blahdeblah: it was done as part of the fix for bug 161111109:09
mupBug #1611111: destroy-model is not blocking and should be <ateam> <oil> <oil-2.0> <uosci> <juju:Fix Released by wallyworld> <https://launchpad.net/bugs/1611111>09:09
menn0blahdeblah: your best bet would be to use a 2.0-rc3 client to do what you're trying to do09:09
menn0blahdeblah: and get that controller upgraded to 2.0 final09:09
blahdeblahmenn0: the 2.0-rc3 client had that unqualified user name problem, so I upgraded the controller to 2.0.0 (thanks wallyworld)09:26
menn0blahdeblah: ok great09:26
menn0glad you got it sorted09:26
menn0sorry for the trouble09:26
mgzhmph hmph, bug battle09:38
=== Guest21645 is now known as ahasenack
=== ahasenack is now known as Guest91977
lazyPoweris there a way to tune the timeout of an action?13:14
lazyPoweri have a long running command that can reasonably take up to 6 hours to complete. I'm noticing the action seems to be timed out after just an hour and nukes the process thread13:14
natefinchlazyPower: probably not13:15
lazyPowernatefinch this is unfortunate news indeed :/13:15
natefinchlazyPower: I'll double check the code to see13:17
lazyPowernatefinch - for context, we built a layer to run the kubernetes e2e suite so we cna just relate itto the cluster and kick off an action to validate it conforms to what both we and google expect from a deployed k8s cluster13:18
lazyPowernatefinch - the full gambit of tests can take nearly 24h in some cases, in our use-case the test runs take a few hours.  So any way we configure this we're headed for a tire fire it sounds like13:19
natefinchlazyPower: try using a numeric param called timeout and set it to the correct timeout in nanoseconds  (why the hell it's nanoseconds, I don't know).  I think that might work.  It looks like the timeout for juju run is specified that way, and it might work for actions, which use a lot of the same code13:28
lazyPowerok13:28
lazyPoweri'll give that a go and let you know if it yields diferent results in ~ 40 minutes13:28
lazyPowerseems like the timeout is closer to 30 minutes than my original estimated 1 hr13:29
natefinchlazyPower: worth trying a very short timeout on a long action first, to see if it does anything13:29
lazyPowerhttps://bugs.launchpad.net/juju/+bug/1637191  -- i filed this to track13:29
mupBug #1637191: [Feature Req] Make action timeout values tuneable via model-config <juju:New> <https://launchpad.net/bugs/1637191>13:29
natefinchlazyPower: like, try to get it to timeout after 30 seconds.... easier to test that it works that way, than try for a 40 minute run ;)13:29
lazyPowerhokay ;)13:29
rick_h_lazyPower: try a --timeout flag to the command?13:31
rick_h_https://github.com/juju/juju/blob/6cf1bc9d917a4c56d0034fd6e0d6f394d6eddb6e/cmd/juju/commands/run_test.go#L128 has me wondering13:31
rick_h_lazyPower: but I don't see where there's a hard coded timeout in juju for this in looking13:31
lazyPoweri added the timeout param to the action, building and upgrading charm13:32
rick_h_oh heh, natefinch got you the same info it looks like13:32
lazyPower:D I'm in good hands13:33
rick_h_natefinch: chat?13:34
lazyPowerThat doesn't appear to have changed the actual timeout.   http://paste.ubuntu.com/23388262/  -- is what i have in my actions.yaml13:35
natefinchrick_h_: yep.13:35
lazyPoweri would have expected this to terminate context nearly immediately13:36
=== frankban is now known as frankban|afk
rick_h_lazyPower: can you file a bug for the timeout exposure on actions please?13:38
lazyPowerrick_h_ i filed this: https://bugs.launchpad.net/juju/+bug/163719113:38
mupBug #1637191: [Feature Req] Make action timeout values tuneable via model-config <juju:New> <https://launchpad.net/bugs/1637191>13:38
lazyPowerare we looking for a different bug context on having specific actions take timeout values instead of blanket config in model-config?13:39
katcorick_h_: hey i just read your email, i'm double-checking13:50
rick_h_katco: k, I was looking at the github list of commits for each branch and got freaked13:51
rick_h_katco: balloons says I'm ok and to calm down :)13:51
katcorick_h_: i pulled from commit 5890aa9, not HEAD... maybe that's why you're seeing a difference?13:51
balloonsohh katco, you are around early. I reverted wallyworld's changes in your PR that you landed.13:52
rick_h_katco: sorry, been otp and not looked at the 'why' yet.13:52
balloonskatco, is there a reason you landed them without the bot?13:52
katcoballoons: you said the bot didn't watch the 2.0 branch?13:52
balloonskatco, I hope I didn't lead you astray by saying to manually do the merge to create the branch13:53
balloonskatco, $$merge$$ still works13:53
katcoballoons: oh... ok. why didn't we use that for merging the first set of commits?13:54
katcoballoons: sorry about that. i thought the bot didn't work on that branch13:54
balloonskatco, I didn't want to hold out waiting for a run as that exact code had already been through a full CI run13:54
katcoballoons: oh i see. ok, sorry for the misunderstanding13:55
balloonsand my expectation moving forward is we would just create the branch; not through a PR as it's already "landed" code we would be branching13:55
katcoballoons: well the branch was already there, i just merged code into it. moving forward all of these changes would have been done directly off the 2.0 branch13:55
balloonskatco, no worries. I would rather have already sorted this out but we're doing a release now with this mantra before finalizing the changes Alexis brought up13:56
balloonskatco, true. I was thinking there wasn't a brancg13:56
katcoballoons: we'll get it right eventually. so are wallyworld and axw's cherried commits in there? do i need to re-propose?13:57
balloonskatco, I just reverted them. They broke the build and maybe could have been fixed and merged, but you didn't get to see that before merging last night :-(13:58
katcoballoons: doh... ok. so i'll repropose? or is it too late now?13:59
balloonsI don't think you need to re-propose now as they will come naturally in the next release.13:59
katcoballoons: ok. sorry wallyworld, axw13:59
balloonsI'm sorry if I led you astray. But please do use the bot to land code, and test it yourself as well before proposing :-)14:00
rick_h_natefinch: ping for standup14:01
=== petevg_afk is now known as petevg
katcoballoons: hey i think rick_h_ may have been right14:12
katcoballoons: the commit alexisb_ suggested was for a branch prior to the commit to merge into develop14:13
katcoballoons: so i think i only brought in the commits from that work, not everything up to when that work was merged14:13
katcoballoons: i.e. 5890aa9 not bebd62714:14
balloonskatco, ahh. I verfied that what you brought in and the commit mentioned where exactly the same14:14
balloonsso if the commit you wanted wasn't the right commit, well ;-)14:14
katcoballoons: yep, kind of how i was operating14:14
katcoballoons: yep lol14:14
katcoballoons: so what can we do here?14:14
balloonskatco, well it sounds like rick_h_ is going to ask for 2.0.1 to have everything in it14:15
balloons?14:15
katcoballoons: yeah already has14:15
balloonsyou may as well re-do the cherrypick then14:15
katcoballoons: ok, i'm going to --force the correct merge in if that's ok?14:15
balloonsso you are going to merge bebd627?14:16
katcoballoons: yes, and then cherry the two commits?14:16
katcoballoons: and overwrite history14:17
balloonsok14:17
katcoballoons: with --force14:17
perrito666fun fact, Air Conditioning installers do not do the electrical wiring to the device....14:17
balloonskatco, can you do it so we get a pre-check run14:17
balloonsso we know the basics are sound -- otherwise we'll end up in the same place, since this isn't a straight pull anymore14:18
katcoballoons: sorry, can you restate? what do you mean "do it so we get a pre-check run"? how do i do that?14:20
balloonskatco, sorry. I was asking if you were going to propose it as a PR or not14:23
katcoballoons: oh, yes i am14:23
balloonskatco, ok, fire away14:23
balloons:-)14:23
katcoballoons: do you know what sha i should begin from? d25720a?14:23
=== frankban|afk is now known as frankban
balloonskatco, I don't know offhand. Let me see14:24
balloonskatco, to be far you could just grab the additional commits since the branch is up to 5890aa914:27
balloons*fair14:27
katcoballoons: i'd rather not have the messy history, plus we now have changes on top of that. not sure what git will do14:28
balloonskatco, then nuke the branch and redo it I think14:28
balloonskatco, use the 2.0.0 tag and re-make from that, then propose what you wish14:29
balloonshow's that?14:29
balloonsit's at 6cf1bc914:29
katcoballoons: sounds good to me14:30
katcoballoons: do you prefer i merge the pre-checked commits first and then $$merge$$ the cherries, or do you want them together?14:35
balloonskatco, I don't think I care. I don't think merging them all at once will cause issues in the future with merging to the branch. That would probably be my only concern14:38
balloonsall at once will make it easier from a test perspective14:38
balloonsso I guess that's my vote ;-)14:39
katcoballoons: ok, so all at once and then use $$merge$$14:39
katcoballoons: sounds good to me14:39
katcoballoons: rick_h_: sanity check: https://github.com/juju/juju/pull/651014:49
katcoi am a little confused because i'm seeing commits from nov. 2015, but i'm assuming this is from some kind of long-lived branch that landed14:49
alexisb_katco, are they the CMR commits?14:50
katcoalexisb_: trying to figure out what they are14:50
katcoalexisb_: yeah, looks like that's it14:50
rick_h_katco: cool that looks more like two weeks work from the team ty14:52
katcorick_h_: yep, sorry for the trouble14:52
rick_h_katco: all good, just glad we caught it before release ty for working on it14:53
katcorick_h_: yeah thanks for checking for me14:53
rick_h_katco: I'm getting back to the changelog. Where did you end up with the commit line?15:29
katcorick_h_: what do you mean commit line?15:30
rick_h_katco: what commit was the end of the line?15:30
rick_h_katco: if I were to look at develop and judge where to stop for the changelog notes15:30
katcorick_h_: oh, one sec15:30
katcorick_h_: bebd62715:30
rick_h_katco: you've not pushed to 2.0 on upstream yet right?15:30
rick_h_katco: cool ty15:30
katcorick_h_: no, i need to $$merge$$ looks like prechecks were successful15:31
rick_h_katco: cool thanks15:31
katcorick_h_: https://github.com/juju/juju/pull/6510 if you're curious15:31
rick_h_cool, that might be easier to go through15:32
katcoballoons: rick_h_: looks like it's all merged15:55
rick_h_katco: <315:56
=== frankban is now known as frankban|afk
mupBug #1637267 opened: Juju fails to restore state-server on xenial <ci> <juju-core:New> <https://launchpad.net/bugs/1637267>17:15
balloonsrick_h_, alexisb I'm bumping develop to 2.1-beta1. We comfortable with that as the "next" version17:52
alexisbballoons, I am17:52
rick_h_balloons: yes, wfm17:53
balloonsnatefinch, can you review? https://github.com/juju/juju/pull/651117:55
balloonsnatefinch, this one too :-) https://github.com/juju/juju/pull/651218:01
natefinchyep18:03
natefinchLGTM18:04
natefinchrick_h_: too much?  Just right? http://i.imgur.com/FAB5rSS.png18:42
rick_h_natefinch: yea, would drop the X18:43
natefinchaww... I like the X.  damn ascii purists ;)18:43
rick_h_:)18:43
natefinchseriously, I think the X is nice because it gives a visual indication that doesn't rely on color support for your terminal or your eyes.  But I know that's kind of pushing the envelope in the linux world18:45
rick_h_natefinch: right, but it's not something in our "visual language" to date18:46
rick_h_natefinch: so I'm hesitate to start adding new ways to indicate "error"18:46
rick_h_natefinch: without thinking/doing it thruogh the whole codebase in a clear way18:46
natefinchwell, we only have a couple interactive commands... but that's ok, it's easy enough to add and remove as needed.18:58
rick_h_redir: ping, I hear you have ipv6 on your network?20:08
rick_h_redir: e.g. you could in theory ping a public ipv6 address?20:08
redirrick_h_: whois me20:09
rick_h_whois redir20:09
redirProlly says I'm connected from an ip6 addresws20:10
rick_h_redir: can you ping6 2001:4800:7818:104:be76:4eff:fe04:78a620:10
rediryes20:11
rick_h_redir: k, ty20:11
rick_h_redir: can y6ou also try 2001:4800:7818:104:be76:4eff:fe04:c448 please?20:12
rediraffirmative20:12
rick_h_interesting, every node in rackspace gets a public ipv4 and ipv6 address20:13
rick_h_ty redir20:13
redirnp20:14
redirrick_h_: I've had ip6 for ~3years20:14
redirin NY and here20:14
rick_h_redir: cool20:15
rick_h_redir: yea, I think I can get my comcast setup to do dual but not bothered yet. Might be time to investigate that I guess.20:15
redirfor a while it broke android phones. They stopped reaching the net if they received an ip6 DNS server20:16
rick_h_heh, oops20:16
redirI didn't do anything.20:16
redirjust got it one day in NY20:16
redirand it is on here'20:16
rick_h_oic20:16
redirobviously ip4 still works20:17
redirand I've been on the same ip4 address for a year, but my ip6 address changes regularly.20:18
redirgo figure20:18
rick_h_heh, that's crazy. You'd think you'd just get one20:18
mgzit's probably a feature?20:18
mgzlike he does actually have a big range, but day to day moves around so people can't find him :)20:19
mgzrick_h_: I have a proposed fix for ssh from windows, am going to mail juju-dev with summary20:26
rick_h_mgz: ty20:26
mgzfood first :P20:27
redirapparently there's some ipv6 privacy stuff that let's you have multiple addresses, and they change regularly.20:37
redirBut I've disabled that on my host, because it causes chrom[e|ium] to show network changed errors when it happens20:38
balloonswallyworld, if you are said, you should be happy to hear your cherrypicks came in the end20:41
rick_h_balloons: <320:42
katcoballoons: they were actually part of the correct merge commit all along20:51
katcoballoons: no picking of cherries required :D20:51
balloonslol. nice20:51
redirdo we already have some simplestreams parsing code somewhere?21:00
redirutils or anywhere?21:01
rick_h_natefinch: or katco can you help redir please? ^21:11
redirI see somethign in environs/21:12
katcoredir: i saw your question, but i don't know :(21:12
redirenvirons/imagedata21:12
katcoredir: i would be surprised if we didn't21:12
katcoredir: wallyworld would probably know21:12
redircomparing to uvtool now21:12
redirdo we actually sync/mirror images in juju anywhere?21:27
rick_h_redir: what are you up to?21:29
rick_h_redir: I wonder if the context of what you're up to will help with a better answer21:29
redirremoving dependency on uvtool21:29
redirwhich mirrors cloud server images for creating kvm instances from21:30
redirloosely is what I am looking at right now.21:30
redirif we already have code to mirror/sync images I'd use it21:30
redirlxd does something like that21:31
* redir goes to look for lxd codee21:31
redirlayers I imaging21:32
rick_h_redir: lxd doesn't do any caching21:32
redirwhoops21:32
rick_h_redir: at this time, it's a feature we want to add21:32
redirtx rick_h_ alexisb21:39
alexisbperrito666, ping21:45
perrito666alexisb: pong21:56
alexisbheya perrito666, seems the vsphere updates are causing ci to fail21:57
alexisbcan you please take a look21:57
perrito666sure I can21:57
alexisbhttp://reports.vapour.ws/releases/4535/job/vsphere-deploy-trusty-amd64/attempt/24521:57
perrito666alexisb: from the rather unclear output of that test it would seem that it is just taking too long21:59
balloonsrick_h_, alexisb, are we confident that we'll land something else for 2.0.1?22:00
perrito666sinzui: abentley mgz  care to give me a hint?22:00
balloonsI ask only because https://github.com/juju/juju/pull/6512 landed, and will need reverted. I tried to be as quick as possible22:00
balloonsbut it would mean a little rework in this case if so :-)22:01
alexisbballoons, I think perrito666 needs to investigate22:01
alexisbballoons, can you help with his inquiry22:01
balloonsok, just please note that the version has to be revved back to 2.0.1 if you land something. ok perrito666?22:02
sinzuiperrito666: that looks like the open connection bug. Juju has exhaused the vsphere resources are repeated tests. we kas larry to restart vshere to relaim the 1000s of connections22:02
perrito666balloons: ack22:02
perrito666sinzui: can you confirm that using the vsphere dashboard thing?22:03
sinzuiperrito666: no, because I cannot see it22:03
perrito666sinzui: I have seen this work without the failure22:03
balloonssinzui, we need larry?22:04
perrito666we need something with access to the pannel22:04
sinzuiperrito666: indeed it does work until resoruce exhaustion. it happens about every 10 days22:05
perrito666sinzui: let me rephrase, I have seen juju work without taking resources22:06
balloonssinzui, we're trying to decide if this is a blocker22:06
balloonssounds like cursed ;-(22:07
alexisbperrito666, could we run the test steps manually and see if they pass22:09
perrito666alexisb: I guess we could, that is a question for someone in CI though that has both the env setup and knowledge of the tests22:09
perrito666alexisb: sinzui has been gracious enough to do these things for me :)22:10
alexisbperrito666, sinzui ok, this is currently a blocker for a 2.0.1 release so we will want to make sure we are not blocking needlessly or if there is an issue we need to investigate as a top priority22:11
=== akhavr1 is now known as akhavr
abentleyperrito666: Sorry, I'm EOD, but it sounds like you and Curtis have already figured out more than I knew.22:16
perrito666abentley: sorry I included you in the initial talk to keep all of QA in the loop, apologies22:16
abentleyperrito666: np22:17
axwballoons: still about? what did you revert? I can't see any revert commits on 2.022:17
axwalexisb: do you know? ^^22:20
balloonsaxw, branch was remade22:34
balloonsaxw, your stuff is in there22:34
axwballoons: ok, ta22:34
balloonsgood morning :-)22:34
axwgood evening :)22:35
alexisbperrito666, menn0, ping23:17
alexisbredir, ping23:17
redirtx23:17

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