[00:19] hey huwshimi [00:20] Hey [00:21] hatch: How's things? [00:21] going good - just trying to get everything done before next week [00:22] seems these things always happen a week too early haha [00:22] :) === uru_ is now known as urulama === urulama is now known as urulama-afk === urulama-afk is now known as urulama [11:01] morning party people [11:05] frankban: how goes this morning? [11:06] rick_h_: I was going to send you an email [11:06] rick_h_: I was able to give internet access to nodes, by doing the following: [11:06] - add a default 10.0.0.1 gateway to the MAAS network; [11:06] - fix the POSTROUTING netfilter rule in the host, i.e. [11:06] iptables -t nat -A POSTROUTING -s 10.0.0.0/24 -j MASQUERADE [11:07] cool, yea that's what I was just looking at in some google searches [11:07] rick_h_: Next error was related to juju tools: "cannot execute binary file: Exec format error”. [11:07] See https://bugs.launchpad.net/juju-core/+bug/1325968 [11:07] so the fix seems t be --constraints arch=i386 [11:08] ugh [11:08] rick_h_: with that constraint, juju finally bootstrapped! [11:08] yay! [11:08] rick_h_: so now I am testing quickstart with and without the constraints flag [11:09] frankban: <3 ok very cool. Glad you're unblocked enough. Thanks for helping throgh all this fun stuff [11:09] frankban: let's make sure we add notes to that maas doc so that we can make sure this works out as the hope is we'll have a maas lab going forward [11:09] rick_h_: yeah it was ages since my last iptables command ;-) [11:10] no kidding, I got that far last night and took that as a sign I was done :) [11:10] iptables and multiple glasses of wine never mix lol [11:10] :-) [11:11] rick_h_: now running ".venv/bin/python juju-quickstart --constraints arch=i386" fingers crossed [11:12] frankban: cool, maybe did I install i386 ubuntu on the maas controller? [11:12] rick_h_: unrelated, but ssh always disconnects me after a few minutes with a broken pipe [11:12] rick_h_: the kernel is 64 bit [11:12] frankban: yea, I wonder if I've got a network thing going on. It kept doing that to me ssh'ing out on my laptop last night [11:12] frankban: but my desktop stayed connected the whole time, so it's strange [11:13] rick_h_: the nodes are i386, and that's where tools are used [11:13] frankban: ic, but the nodes can be anything right? We can set them as amd64 in maas and it should install the right images from it's list of images? [11:13] rick_h_: in theory yes, in theory... [11:13] lol [11:14] frankban: ok, so you ssh'd in and verified they were running i386 images? [11:14] * rick_h_ is trying to phrase a reply to the bug and if we can dupe it we should work with juju core on it [11:14] rick_h_: no I have not but in the edit node page i see "i386/generic" arch [11:15] frankban: oh? oooh, ok. [11:15] I was looking at nuc2 which is amd64 [11:16] so maybe it's my failing to setup maas correctly [11:16] frankban: when you're test is done let me know and we'll edit the node and try to recomission it [11:16] rick_h_: ok [11:16] and see if we can get updated software on it [11:16] and then make it work sans constraint :) [11:17] rick_h_: woot, all's good for now: http://pastebin.ubuntu.com/8472118/ [11:17] yay! [11:18] so now the other thing to document is sshuttle into the 10.0 network on there so you can get at the gui [11:18] rick_h_: quickstart has not finished yet [11:18] * rick_h_ really doesn't want to redo this setup but wonders if trying to put all the nodes on the public network would be best [11:19] rick_h_: so, GUI deployed, but then quickstart exited with an error: juju-quickstart: error: unable to connect to the Juju API server on wss://nuc1.maas:443/ws: [Errno -2] Name or service not known [11:20] so, dns is not working [11:20] hmm, yea [11:21] so maas is setup and supposed to manage dns, but I bet it's because I set the dns on the hosts to google or something vs itself [11:22] * rick_h_ looks up format of dig commands [11:23] dig nuc1.maas @10.0.0.1 [11:23] works so cool [11:24] frankban: ok, updated resolv.conf and the nameserver for the 10.x interface and just a normal cli "dig nuc1.maas" returns now [11:26] rick_h_: cool, so, I'll re-run quickstart on the bootstrapped environment and this time it should be able to connect to the gui server [11:27] frankban: rgr [11:27] rick_h_: done, quickstart completed successfully: http://pastebin.ubuntu.com/8472146/ [11:27] \o/ [11:27] woooooooo! [11:27] it's early...but a happy dance it is [11:28] rick_h_: destroying the environment [11:28] hold on if you can [11:28] ops [11:28] ok, np [11:28] rick_h_: another thing I noted, when destroying an environment the node is left to Releasing failed status, and you need a manual "release" fot it to be back in ready state [11:29] frankban: yea [11:29] rick_h_: but I guess it's not our problem [11:29] frankban: ok so sshuttle -v -D -r ubuntu@maas.jujugui.org 10.0.0.0/24 [11:29] let's me load up maas on http://10.0.0.1/MAAS [11:29] rick_h_: cool, should wer try to switch nuc1 to amd64? [11:30] frankban: so the hope is that line will let us load up the gui and environment stuff once it's brought up [11:30] frankban: yea, let's do that [11:30] frankban: ok switched and comissioning [11:30] rick_h_: cool thanks [11:31] * rick_h_ wishes he had all the kvm stuff setup to watch it and make sure it's going...doesn't trust maas enough yet [11:31] frankban: yea, on the release node stuff, it's new and we've got some issues from the experimental branch [11:31] frankban: you'll be happy to know our big problem before was lack of the amt control tools package installed [11:31] frankban: the new software shows the error while before I assumed maas handled that bit [11:32] rick_h_: yay for experimental maas [11:32] definitely, so maas was never powering on/off our machines as hoped [11:32] ic [11:33] rick_h_: one confusing bit is that nuc1 pretends to have 10.0.0.151 addr, but when it starts it's on 10.0.0.100 [11:34] frankban: yea, I'm not sure what's up with that to be honest. [11:35] rick_h_: ok the node is ready [11:35] rick_h_: running quickstart without constraints [11:35] frankban: rgr [11:36] rick_h_: at least we now my branch with maas support works and generates a reasonable configuration [11:38] frankban: :) [11:39] frankban: yea, I had hoped it would be very little different other than generating the proper environments.yaml [11:39] rick_h_: yeah, it's basically just it [11:39] and then 3 days of getting maas and juju happy lol [11:39] across timezones [11:39] lol [11:40] hmm, it's green and says deployed? [11:40] rick_h_: yeah, qucikstart still bootstrapping [11:40] cool [11:41] rick_h_: and now it's deploying [11:41] rick_h_: so bootstrap succeeded without constraints [11:41] woot [11:41] lol, every change one step closer...and closer...and closer [11:41] waiting for the new shiny cs:trusty/juju-gui-9 unit to be deployed [11:42] rick_h_: done! quickstart is ready [11:42] sweet, so with the shuttle command can you load the gui in your local browser now? [11:42] trying [11:45] it works here https://10.0.0.100/login/ [11:45] rick_h_: it works! [11:45] woot! [11:45] jujugui we have a maas ! rock on frankban [11:45] rick_h_: password is 109b105fe8b699f4b0a2555563502331 [11:46] hah, /me wants to deploy something! [11:46] rick_h_: machine view shows 0-state service 4xGHz, 8.0GB, [11:46] \o/ [11:46] nice! [11:46] we might want to strip the final comma there ;-) [11:46] and with machine view we can container on that thing and deploy crazy stuff [11:46] lol [11:47] rick_h_, frankban: did anyone tried juju-gui on manual? i can play with that tonight before going to bed as I wanted to test using mongo shards and multi CS units [11:47] rick_h_: ghost charm on the bootstrap node? [11:47] frankban: yea, should work [11:48] urulama: 'juju-gui on manual'? [11:48] rick_h_: in a container? [11:48] rick_h_: manual environment [11:48] urulama: oh hmm, not sure. [11:48] frankban: yea, lxc it up [11:48] rick_h_: ok, i'll let you know [11:48] frankban: my understanding is that the lxc container will get a root 10.0.0.xx ip from dhcp [11:48] frankban: and be able to be exposed at the root network level [11:49] urulama: I think with machine view now you could bring up the machines, add them, and they'd show up in machine view [11:49] urulama: and then you could deploy to them directly [11:49] urulama: but you'd have to add them and such via cli/manual stuff first? [11:50] ok, bringing up elasticsearch on nuc2 hopefully [11:52] rick_h_: nuc2 is pending, 0/lxc/0 is pending. man, this GUI thing seems to work just fine [11:52] frankban: <3 [11:52] frankban: sometimes, we do damn good work :) [11:52] rick_h_: yes, adding will have to be done by add-machine unless there is code in GUI that makes this "add-machine" call. [11:53] urulama: right, it makes an add-machine call but I think in manual you have to provide extra info to that call as to where the machine to add is located [11:53] urulama: and the gui doesn't do that [11:53] rick_h_: http://10.0.0.155:2368/ \o/ [11:53] rick_h_: ok. [11:53] frankban: woot, ghost looks green and exposed [11:53] haha! [11:54] * rick_h_ does another round of happy dance and notices son is up...time to prepare for day bare biab [11:54] frankban: you rock thanks again for the pita work on this! [11:54] * rick_h_ goes afk for a little bit [11:54] and it's not "Just a blogging platform.". it's more "just a blogging platform deployed via a GUI which manages juju on top of a maas controller" [11:54] rick_h_, frankban: did you try deploying openstack? :P [11:56] urulama: I don't even know where to start to do that. but for our scope, ghost is as good as openstack ;-) [11:59] * frankban lunches [12:00] frankban: https://jujucharms.com/bundle/~james-page/openstack-on-openstack/25/openstack/?text=openstack [12:06] frankban: urulama http://marcoceppi.com/2014/06/deploying-openstack-with-just-two-machines/ :P [12:06] rick_h_: even better :) [12:06] frankban: urulama so I think I'll end up getting a 3rd nuc so we can easily test out bundles on two empty nodes and maybe then we can get openstack on them :P [12:07] oh boo, elasticsearch failed [12:08] oh, it can't reach out to the internet [12:09] no, it did reach out ok. It just got a forbidden for some reason ok [13:49] guihelp: is anyone available for taking a look at https://codereview.appspot.com/151200043 (quickstart/python)? thanks! [13:50] rick_h_: thanks for the nice email [13:51] frankban: <3 [13:53] frankban: i can look [13:53] bac: thanks [13:54] ah, i've missed rietveld. [13:54] like honestly? [14:00] compared to reviews on github, yeah. [14:00] +1 [14:00] the only difference I can see is that you can queue up review comments [14:01] but you can also do that on github wtihout clicking 'submit' [14:01] hatch: so you keep each comment in limbo and then submit them all at once? [14:02] no because queuing up the comments doesn't really bother me :) [14:02] but it is possible [14:02] hatch: i also don't like how it collapses and hides your comments when a new revision is pushed. i find it very hard to follow the review and determine if everything has been addressed. [14:02] hatch: there are other little things that I miss. For instance, better handling of long lines (when reviewing go code it is useful), ability to comment unchanged code and above all, a way to review patches and follow the review discussion that doesn't drive me crazy [14:03] reitveld frequently 'forgets' comments when saving so it's really a catch 22 [14:03] no, it is a toss up. it is definitely not a catch 22. :) [14:03] lol right [14:04] maybe if reitveld wasn't so ugly I'd like it more [14:04] gota get some designers up in that house [14:05] * rick_h_ changes location back to the house. afk [14:08] see, when doing the review i made a comment about some user-facing text but when i did the QA i saw my comment was bogus. deleted it. approved the review. no noise from premature comment sent out. [14:08] frankban: +1 [14:08] lol [14:08] hey luca__ [14:09] hi hatch [14:09] bac: thanks [14:09] frankban: amazing that it was so simple! [14:12] rick_h_: you need to add a photo of your MAAS to that document [14:16] luca__: sent an email to ya regarding the commit summary [14:16] rick_h_: i tried to ssh to the maas but cannot connect. did you do an ssh-import-id for all of us? [14:18] hatch: I haven’t seen any email... [14:18] hatch: I just got an email [14:19] :) [14:24] hatch: I didn’t know this feature was coming! [14:24] yeah we like to keep you on your toes :P [14:25] hatch: haha [14:25] I'm going to land this as is but would be nice to tweak it a bit for the next release [14:25] hatch: maybe we should schedule a call so you can talk me through it, it seems pretty straight forward. [14:25] hatch: do we have to land it? [14:26] well if I don't the code could go stale [14:26] ok [14:26] I'm pretty much free whenever [14:26] ok [14:26] with the exception of the standup in 45 [14:26] er 30 [14:27] guihelp: am I correct in thinking that a revived model (from a LazyModelList) should fire change events automatically? [14:27] kadams54: yes [14:27] hatch: *sigh* [14:27] lol why? [14:27] that sounds like a good thing [14:27] and pretty much the only reason to revive :P [14:28] hatch: it would be a good thing… if that's what I actually saw happening. [14:28] hatch: http://pastie.org/private/kndmzbqg3syfm9zgeh21pg [14:28] looking [14:28] That's the change I made. But now my onChange handler function isn't being called at all. [14:29] and you're sure the revive is being called? [14:29] Yeah, got a breakpoint on it [14:29] and what's the event handler? [14:30] just the registration part [14:30] I'll double-check though. I'm also wondering if something is breaking in event propagation. [14:30] I'm guessing that there is something wrong with your event registration [14:30] bac: I can add your keys, what pait are you using? I see four in your lp page [14:31] pair even [14:31] this.addEvent(db.machines.after('*:change', this._onMachineChange, this)); [14:31] hatch: ^ [14:31] frankban: just run 'ssh-import-id bac', if it hasn't been done [14:31] bac: cool [14:31] kadams54: yeah that won't work [14:31] frankban: i should probably prune what i've got on LP [14:32] bac: done [14:32] frankban: and i'm in. thanks. [14:32] kadams54: lazy model list doesn't have a change event for attribute changes [14:33] kadams54: and unfortunately reviving doesn't add the lml as a bubble target [14:34] The previous approach, of manually firing the change event, was buggy because the event object wasn't populated with the same info that a change would have in a ModelList. [14:34] ahh [14:34] well you can fire a custom event facade [14:34] the second param of fire() [14:35] nbd just another YUI bug that'll never get fixed now ;) [14:36] bac: woot welcome to maas :) [14:36] hatch: yeah, I know I can fire a custom event facade, but that means trying to populate it with the same info that YUI does for changes in a ModelList… and I'd like to avoid that. [14:36] well you only have to populate it with the information that your handlers expect [14:36] which shouldn't be too much [14:37] rick_h_: so you have some poor man's maas? no orange box? [14:37] bac: yea, I've got a 'rick bought 3 nucs, a switch, and a internet plan with static ips' maas [14:37] rick_h_: nuc-nuc-nuc [14:37] bac: so it took months to come up vs an orange box taking 10min [14:37] hatch: the whole point of an event architecture is to decouple so that the code firing the event doesn't have to know about the code subscribed to the event. [14:38] hatch: I mean, if that's what I have to do, I'll do it. I'm just hoping there's a better way. [14:38] you should name them larry, moe, and curly [14:38] bac: hah! [14:38] not without patching lazymodellist [14:38] bac: .95 manual focus lens just arrived wheeee! [14:39] rick_h_: can't you just turn off auto focus to get a manual focus? ;) [14:39] rick_h_: which one? [14:40] bac: voightlander 17.5mm for m4/3 [14:40] that'll be a real challenge to focus i would think [14:40] bac: so m4/3 focus peaking ftw [14:40] it'll take some practice for sure [14:40] but focus peaking should hopefully help [14:41] the hard part is I'm used to setting aperture priority and hitting go [14:41] being that wide, you've got some leeway [14:41] so now it's more set aperture on the lens, focus with peaking, adjust shutter speed to get desired exposure manually [14:41] but .95 is wiiiide open [14:42] wait, can't you leave it in AP and let the camera set shutter speed? [14:42] it doesn't know the aperture since it's done on lens [14:42] it has no contacts to talk to the camera [14:42] the camera thinks there's no lense there [14:42] rick_h_: no, but it has a meter. [14:42] rick_h_: no different from my rangefinder [14:43] * rick_h_ will have to check it out and try [14:43] a step in the right luddite direction for sure [14:43] bac: ah cool. So the aperture shows 0.0, but it does adjust SS for me when I hit go [14:44] rick_h_: but your exif data will be hosed! [14:44] yes, no exif data :/ [14:45] mine guesses based on having two out of three data points [14:45] gotcha [14:45] it writes to a different exif field showing that it is not sure [14:48] I just use my cell phone [14:48] rick_h_: are you available for a quick second review of https://codereview.appspot.com/151200043 ? [14:48] < --- professional [14:49] frankban: already looking [14:49] thanks [14:52] frankban: done, thanks for the changes. The goal will be to work on the release and completing the FFE bugs in the package for utopic after this please. [14:53] bac: now that's shallow https://www.flickr.com/photos/7508761@N03/15411328085/ :) [14:54] rick_h_: 404 [14:54] doh, perms [14:54] reload [14:55] rick_h_: cool thanks, I'll work on a 1.4.4 release after landing this [14:55] frankban: awesome. [14:55] rick_h_: that's some seriously shallow dof [14:56] what;s that, 1"? [14:56] hatch: 5 or 6 [14:57] Oooh. Jealous. [14:58] jujugui call in 4 [14:58] kanban it up please [15:12] so is my mic not working? [15:18] bac: yea, mic fail [15:18] camera fail [15:18] rogpeppe: under "Mailing List" do you have an Unsubscribe button? [15:18] bac: yup [15:18] rogpeppe: filter or spam catch it? [15:18] rogpeppe: nm, https://launchpad.net/~juju-gui-peeps/+mailing-list-subscribers [15:18] rick_h_: i'm looking in Spam [15:18] you are on it. [15:19] "gui maas now available...pretty much" is the exact title [15:19] rick_h_: i don't think so. when did you send the mail? [15:19] rick_h_: does everyone really have access? frankban had to manually add me [15:19] bac: no, I need to go through and import ssh keys [15:19] bac: but you me and frankban can now add that in [15:19] rick_h_: i'm guessing it might just be delayed [15:19] rick_h_: okey doke [15:20] bac: I'll try to run through everyone's LP ids at lunch [15:21] rogpeppe: shared the doc with you and included the email contents in the share notice [15:21] rick_h_: thanks [15:21] rick_h_: when did you send the email? was it today? [15:23] rick_h_: i've got that. i wonder what happened to the mailing list message. [15:23] * rogpeppe is concerned that he may have missed lots more [15:26] Makyo: did you forget to mark this one as committed? https://bugs.launchpad.net/juju-gui/+bug/1365260 [15:29] rogpeppe: whoa, you're not on yellow in LP strange [15:30] rick_h_: hmm. i am in ~juju-gui-peeps though. [15:30] rogpeppe: added you, should hopefully be all set, not sure on the peeps list but at least on the team wise [15:30] rogpeppe: yea [15:30] rick_h_: no archives, i'm guessing [15:30] rogpeppe: no, nothing much that you'd be interested yet I don't think [15:31] rick_h_: cool [15:31] rogpeppe: it's been very gui/machine view focused so far though it's starting to warm up with storefront stuff [15:31] it's handy as the UX team is also on that list, and a few core folks like ramm, kapil, etc [15:32] I'll try to remember to check next time I see something go across [15:32] jujugui ok added ssh keys for the LP users fabricematrat evarlast jcsackett hatch kadams54 makyo uros-jovanovic martin-hilton rogpeppe to the maas server [15:33] jujugui so if you're not on the list or can't get in let me know [15:33] bac: thanks for the prod [15:33] rick_h_: sweet… will check it out in a bit. [15:37] rick_h_: so what "can't" we deploy here? Does MAAS create vm's ? So it'll look to us like a bunch of machines? Or does it still only look like 2 machines? [15:37] hatch: it's two machines and you can lxc at will [15:37] so can install most anything you want but bundles of 5 machines can't work [15:37] ahh so is that where openstack comes in? to create vm's? [15:39] hatch: yes, thought without juju can you bring up a node with bare ubuntu image [15:39] so MAAS is really the hardware layer then openstack is the vm layer then juju is the orchestration layer [15:39] we got this stack! === uru_ is now known as urulama [15:48] hatch, sorry, bio break. I'll mark it as committed. [15:48] Actually, should be released [15:49] ahh yes [16:15] jujugui: i need two reviews and a pretty serious (requires real env) QA of https://github.com/juju/juju-gui/pull/600 [16:16] jcsackett: I can get on it once I'm finished mine but I'm running into some testing oddities that are being less than awesome [16:16] want to get this finished first [16:18] jujugui ok break time for lunch afk [16:40] Also lunching. [16:44] rick_h_: luca brings up a good point in his email re the conflict blocking - should I bench this branch until we can come to an agreement on a real plan? [16:53] hatch: well, I think we should leave the notification for now. [16:53] hatch: at least it's visible and non-blocking [16:53] hatch: and it's more info that we have today (no hidden config issues) [16:53] hatch: and we can work out the blocking path [16:56] rick_h_: ok so remove the blocking of the commit? [16:56] hatch: that's my idea atm, how does that sound? [16:58] works for me [16:58] I'll respond to the email with luca [16:58] hatch: k [17:05] jcsackett: lol nice typod bug [17:05] containser all the things!!! [17:07] our minimum wage just went up to $10.20/hr [17:07] and there goes the cost of living by the same increase [17:07] lol [17:15] jujugui does anyone know why the inspector breaks when the gui loses connection to its server? Anyone looked into this in their spare time? [17:45] jujugui I need reviews and qa https://github.com/juju/juju-gui/pull/599 [17:45] jcsackett: still need reviews and all that? === uru_ is now known as urulama [17:51] hatch: i do. looking to trade? [17:52] shitty trade but ok [17:52] :P [17:52] bwuahahahha [17:53] jujugui I've put the two big bugs into the urgent lane for next steps. They're big enough I'd like to do another release once we get those in. [17:55] cool [17:57] rick_h_: well there goes Ember from the list https://twitter.com/jdalton/status/517370990532497409 [17:57] :) [17:59] hello [17:59] bah [17:59] * rick_h_ test [17:59] * rick_h_ never thought ember was on the list [17:59] hmm, why did that not go out 3 times in a row :/ [17:59] everything is on the list :) I'm being open minded!!! [18:10] hatch: is there any way to set e.target when firing an event? I tried: list.fire('change', {foo: 'bar', target: model}) and the target that came through on the other side was still list (and not model). [18:15] hatch: Right now I'm considering, as a hack, something like this: http://pastie.org/private/cp1wjtfuigw4bsdxfhhynw [18:15] (with comments, of course) [18:32] rick_h_: just letting you know i was able to successfully get onto the maas stuff. [18:33] kadams54: looking [18:34] kadams54: so it won't let you set the target? [18:34] hatch: doesn't seem to. [18:35] kadams54: http://yuilibrary.com/yui/docs/api/classes/EventTarget.html#method_fire [18:35] yeah looks like it will [18:35] see the arguments definition [18:36] Yeah, I know, which is why I though my first approach would work. But that's not the story told in the debugger. [18:37] well no, the change event has a facade set [18:37] so it'll overwrite what you're trying to pass [18:38] kadams54: I think your approach in that pastie is acceptable with comments as per why it's there and where it's coming from [18:44] kadams54: there are other psudo appropriate ways to do it - setting a custom facade for the event...but no, that's even harder to reason about :) [18:44] just fyi :) [18:47] jcsackett: your branch qa's well on sandbox - I'm going to take lunch then will qa the real env [19:10] which is the preferred place to file bugs/feature requests: launchpad or github? [19:14] bic2k: launchpad [19:14] I think issues is disabled in github [19:17] jujugui running away for the day. jcsackett if you don't get a second review I'll try to look at it tonight after [19:17] rick_h_: have fun [19:18] urulama: get out of here :P I can't EOD before you TZ law [19:18] :D [19:18] * urulama goes *puf* [20:37] jcsackett: +1 [20:37] hatch: whoo! thanks. [20:37] so glad to be closing in on landing that mofo. [20:38] that db.reset call and 'var db =' thing in the tests were total PITAs to track down. [20:38] yeah I bet [20:42] jcsackett: I can't reproduce #1376353 nor does the var in question exist [20:42]