[01:56] yes you CAN run haproxy/ghost/mysql on a t2.small :) [01:56] oh, and the bootstrap node [01:58] using 412M of ram :) [01:58] oop 452 [01:59] lol [01:59] woot [01:59] ! [01:59] I need to somehow make a ton of requests to see if my mem usage goes up [02:00] ab or apachebench or whatever [02:00] yeah ab it is [02:00] ab is apache bench sorry, thinking of seige [02:02] hmm says invalid url [02:02] quotes around it? [02:02] been a while since I ran ab [02:02] ab -n 1000 -c 10 http://ec2-107-20-7-222.compute-1.amazonaws.com [02:02] I'm pretty confident that's correct [02:03] add a / at the end? [02:03] that was the trick [02:03] :) [02:03] :) [02:03] 468 [02:04] 468M was the highest usage with a 266ms time per request avg [02:04] not bad - I wonder if I can do this on a micro [02:05] dropping to 2 concurrent it only goes down to 219ms per request [02:05] so it's not being pegged :) [02:05] try 20 [02:05] :) [02:05] oh there we go lol [02:06] 489ms [02:06] :) [02:06] cpu bound [02:07] unfortunately I can't use the GUI, I need port 80 :( [02:40] darn, can't get mysql to run on a micro [02:40] heh [02:41] with the bootstrap node, haproxy ghost :) [02:46] although I think amazon is lying, they say on their site that it's 1GB of ram, htop is only showing 590 === urulama_afk is now known as urulama [11:04] yay tgif [11:06] rick_h_: indeed :) [11:25] * rick_h_ makes some coffee now that email is processed [14:02] * rick_h_ is going into a call so completely afk for the next little while [14:04] hatch: I have more questions about state :-) [14:04] well ubuntu vm decided it doesn't want to display anything but black pixels today :/ [14:05] kadams54 shoot [14:05] hatch: What is metadata generally used for [14:06] I don't understand the question [14:07] if you look in state.js you can see everywhere it's referenced [14:08] it's used to send the data about the state that needs to change [14:09] Any arbitrary data, or is it more limited than that? [14:09] well it's more limited than that because the state system and dispatchers needs to know what to do with the data [14:09] what are you thinking of putting in there? [14:10] hatch: mysql can run on a micro. its just a bug in the charm. [14:10] "just" [14:10] hatch: A flag for showing/hiding the help text [14:11] jrwren_ yeah I was poking at it, looked like it intentionally needs too much space [14:11] *sadface* [14:11] kadams54 hmm [14:11] hatch: I ran into it often when running on local in a small kvm. I'll propose a fix :) [14:12] metadata is the correct place for that....not sure we want it in there though :) [14:12] jrwren_ awesome! [14:12] I really want to run my blog for $5/mo :) [14:12] hatch: are micros that cheap now? [14:13] DO has a $5 droplet [14:13] and can handle 10 concurrent rps at about 170ms response time [14:13] which is more than enough for my blog atm [14:17] kadams54 adding it to the model might be quite a bit simpler than adding it into state [14:17] plus if it's in state it'll create a routable url [14:18] which doesn't make much sense in this case :) [14:18] yay back from black desktop in vm......stupid parallels [14:18] hatch: The whole point of going down this rabbit hole was to connect the showing/hiding of the text to a user's actions, rather than the model… [14:19] ehh it's not that deep of a rabit hole [14:19] hatch: the problem is that I'm having a hard time locating a good hook to hang this work off of. [14:19] there ever only was one spot to do this :) Now you learnt how this works haha [14:20] rick_h_ thoughts on making the hiding of the help text a routable url? [14:20] hatch: I've looked all through topology/service.js but I don't see anything all that different about how the inspector is created when clicking on a service vs. adding a new one. [14:21] hatch: rick_h_ is afk right now for a phone call… which is why you get the question :-) [14:22] ohh [14:22] well they are different methods which call the changestate [14:23] so in those different methods you would do something different to signify that you shouldn't show the help text [14:23] there is a 'flash' storage in the state which allows you to send data not in the url [14:23] but I think it only allows one record of things at a time [14:24] https://github.com/juju/juju-gui/blob/develop/app/views/topology/service.js#L740 [14:26] https://github.com/juju/juju-gui/blob/develop/app/views/state.js#L461 [14:27] then you'll reference metadata.flash.whateveryoucallyourproperty here https://github.com/juju/juju-gui/blob/develop/app/subapps/browser/browser.js#L586 [14:29] The _deployLocalCharm… is _deployFromCharmbrowser the other side of that coin? [14:30] deploy local charm is what handles when you drop a zip charm on the canvas [14:31] deploy from charmbrowser fires an event initateDeploy [14:32] so you trace that handler down and you'll see where it also fires a changestate event for the ghost inspector [14:33] It sounds like I need to add the flag in two spots: that handler and the _deployLocalCharm, and then extract it again in the renderServiceInspector method. [14:34] why deploy local charm? [14:34] Don't we want to display the help text for local charms as well? [14:35] that shows a totally different inspector [14:35] it should be visible by default and only hidden by the flag - then you only have to hide it in one place [14:36] in showServiceDetails [14:36] brb [14:37] hatch: Hah, my plan was to hide it by default (because the problem right now is that it's shown when we don't want it to show) and then only show it when a service is first added to the canvas. [14:47] ahh yeah - well there are more places where it's created (where we want it shown) than it's not :) [14:47] surprisingly [14:47] kadams54 have you deployed a local charm in the gui yet? [14:48] hatch: yes [14:49] hatch: Your ghost charm, specifically :-) [14:50] well then! [14:50] jujugui call in 10 [14:56] jujugui: i suggest postponing for 10min, so that rick_h_ finishes the current call [14:56] urulama: +1 [14:56] ok [14:57] urulama ack [14:57] jujugui call in 13 [14:57] :P [14:57] off call [14:57] jujugui call in 4 [14:57] lol [14:57] lol [14:57] rick_h_: sheesh :D [15:00] jujugui call time wheeee [15:00] hatch: jcsackett kadams54 ^ [15:01] oops. coming. [15:51] jujugui still need those reviews https://github.com/juju/juju-gui/pull/532 plz [15:54] hatch: you and kadams54 all good? [15:54] We'll know for sure in about 30 minutes :-) [15:55] kadams54: ok awesome. Let me know how it goes. thanks guys [15:55] But yeah, I'm back to coding again, so that's a step in the right direction. [15:57] :) [15:59] hazmat hey thanks for the update to the DO plugin, I'll be giving it a try tonight. [16:21] oh man this critical card was literally a 6 character fix [16:21] it's going to take about 1Mx longer to test than to fix lol [16:22] hatch: well that's good to hear at least then. [16:22] that it's not some fundamental giant of a thing [16:22] well....it sort of is...the charm is broken [16:22] whoops [16:23] but since the cli handles it correctly the guy should too [16:23] gui [16:23] rick_h_ https://gist.github.com/hatched/998e31e217c2a577d112 [16:24] hatch: :) [16:25] jujugui have to run up to the doc office for a little bit. biab. [16:25] enjoy [16:25] flu shots are in and the wife mandates I get one asap yay me [16:27] rick_h_: to actually get a flu, you'll have to leave your desk and computer sometimes ;) [16:27] lol! [16:27] that's why I get sick every time I leave the house [16:29] hatch: programmers should walk around in one of these http://main.cdn.grabone.com/goimage/440x267/i4a3clx.jpg [16:29] hatch: imagine our next sprint! :D [16:29] lol [16:29] it could be ON a lake [16:39] jujugui I'm going to be taken offline (power) in a few for about 30mins [16:40] jujugui still need one more review on #532, qa is already done [16:42] hatch: on it [17:05] urulama: her thing was "You better get this shot before you get on a plane with 300 sick people [17:05] " [17:05] and power! [17:06] woot [17:07] I no longer have a 100A power line running across my backyard [17:07] what could possibly go wrong with that? [17:08] new motox take my money! [17:08] wish I could get a 360 but I think my wife would kill me [17:08] hatch: want an LG watch? [17:08] and I'll get a 360? [17:09] haha no sorry [17:09] no shippping fee straight to brussels :P [17:09] https://twitter.com/FromAnEgg/status/507935819521613825 [17:09] hatch: I'm talking watches not phone :P [17:09] I know....but saving moneys [17:11] boo ok good plan [17:47] hatch: sorry i forgot to look at your PRs. moving charmworld along is surprisingly distracting. [17:47] heh np frankban got it [17:47] cool stuff. [18:28] of course my simple fix turned out to not be good enough :/ bleh [18:54] todays pro tip: if you're going to spill a bowl of borscht don't do it over your desk, keyboard cellphone, wallet, dog [18:56] ouch [18:56] time for a new phone! new moto x here we come! :P [18:57] Makyo: hatch got a sec to chat? [18:57] lol [18:57] rick_h_ yeah 2 mins just need to reposition stuff now that I don't have an external keyboard lol [18:57] hatch: all good [18:57] I'm chilling in today's standup room [19:04] someone finally fixed python: https://github.com/paradoxxxzero/nocolon [19:06] lol no [19:06] that's just wrong [19:06] it even says so in the readme haha [19:07] w...t...f [19:08] less ruby in my python kthx [19:08] haha +1 [19:09] I often forget colons when I Haven't done python in a while. [19:09] I always wished they were optional :) [19:11] nobody is happy [19:11] sheesh [19:12] I guess tonight I'll be taking apart my keyboard and running it through the dishwasher.....it no longer works so....maybe now is a good time to invest in an expensive one [19:16] * rick_h_ holds back [19:20] haha, are you back to using your kinesis? [19:20] yea, been using it for a while [19:20] typing too many docs these days :P [19:20] but I've got a closet of about 7 mechanical keyboards you could try :) [19:22] lol I've been eyeing this one.... https://www.trulyergonomic.com/store/index.php [19:22] unfortunately they are all on backorder [19:22] I may end up switching to one of my many old keyboards if this one doesn't come back to life [19:22] no good windows/meta key so :( [19:23] the one in the middle top i think is the meta key [19:23] yea but typing that easily a lot of times a day is not going to work ou [19:23] out [19:23] yeah true [19:23] https://www.trulyergonomic.com/store/model-differences--truly-ergonomic-mechanical-keyboard--cherry-mx-brown-red-blue-black-clear-keyswitches#Model-209 [19:23] here is one with a split alt key [19:24] could turn one of those into one maybe [19:24] hmm, perhaps [19:24] I've never seen/used this one in person so I really have no idea [19:24] yea, I've looked a lot at it but keep holding back [19:25] https://www.kinesis-ergo.com/shop/advantage-lf-for-pc-mac/ [19:25] that's what I use [19:26] coding with browns would suck [19:26] fingers would get so sore haha [19:26] yea, I do wish for blues, but browns get by [19:27] hah, let me show you my unicomp [19:27] that one I get sore hands for the first week of use every time [19:27] oh you're using browns and not reds? [19:27] https://d2s7foagexgnc2.cloudfront.net/files/49df782d46968ee719e7/kinesis_and_trackpad.jpg interesting layout....I'd probably do something very similar to this [19:28] yea, the browns here [19:28] hah, that's funny [19:29] jujugui ok I'm out. Everyone have a good weekend. [19:29] enjoy your weekend [19:29] cya [19:30] 12 cards left, with 5 people. Should be able to knock em out next week np [19:30] :) [19:31] agreed! [19:40] hatch__: So… it looks like the flash object is being stripped out before it gets down to renderServiceInspector. [19:41] hmm that's odd === hatch__ is now known as hatch [19:41] lemme pull up the code [19:42] kadams54 not sure that's possible - if you look at the renderLocalInspector method it references the flash property [19:42] and it's the same method that either calls the renderServiceInspector or renderLocal Inspector [19:43] Hmm. I see the flash property going in, but it's not coming out the other end. Still tracing through all the state code. [19:44] kadams54 put a debugger in the _inspectorDispatcher and see if it's available there [19:44] because that's where it would be passed to the renderLocalInspector if it had it [19:45] > metadata [19:45] Object {id: "7818300$"} [19:45] So nope. [19:46] hmm can you put a gist of your diff up [19:46] something funky is going on [19:46] This is what I see when I inspect the state object in the *:changeState handler, browser.js line 207: [19:46] > state.sectionA.metadata [19:46] Object {id: "7818300$", flash: Object} [19:47] So the flash is there when the changeState event is caught, but gone by the time it hits _inspectorDispatcher [19:47] gist coming up [19:48] https://gist.github.com/kadams54/1b382fca1c2c1d84071c [19:50] ok put a debugger in state.js on like 461 [19:50] oh that's the problem [19:50] that can never be hit [19:51] move line 461 to line 468 [19:51] well that's a temp fix to see if it works [19:52] will need a bit of a better fix for the long haul [19:53] then I think 126 needs to change from being set to an object to being set to undefine [19:53] d [20:01] kadams54 is that making sense? [20:04] Sorry, afk, back now [20:05] Not sure why line 126 needs changing, but I follow the rest [20:07] hatch: bam, that works. [20:14] it needs to be changed because we don't want it setting the flash metadata on every request [20:14] so that should now be if (this.get('flash')) {...} [20:14] and then where it resets it change it to set it to undefined [20:38] s [20:38] darn keyboard busted [20:38] :/ [20:55] kadams54 I have to run and buy a new keyboard [20:55] I'll likely be back in about 45 if you need some more help [20:55] Friday fun [20:55] I'm all set [20:55] yeah the store is closed on the weekends for the kinesis [20:55] Just working on a test now to make this official [20:55] so gota get there before eod [20:55] ok bbiab [22:18] kinesis keyboard with browns equipped [22:18] this will take a while haha