[01:21] another review, though no rush: I plan to go now :-) : https://codereview.appspot.com/8647043 [12:52] hi frankban. i'm trying to reproduce bug 1167344 but can't do it right now [12:52] <_mup_> Bug #1167344: Deployment via the juju-core API fails with a "no options to set" error in some cases

< https://launchpad.net/bugs/1167344 > [12:52] when i try to add dokuwiki i get a ghost service but then the service never fully deploys and i don't see the error you reported [12:53] bac: is this with trunk? [12:54] rogpeppe: i'm using a local branch based on trunk from ~1700UTC yesterday [12:55] trunk + some debug messages [12:55] rogpeppe, bac: I bet this is fixed in trunk [12:56] bac: I just succeeded in deploying docuwiki, which was one of the problematic charms [12:56] bac: hmm, i think i fixed it in rev 1139 [12:56] frankban: ok, i'll update and try again [12:56] rogpeppe: thanks! [12:56] bac: but i'd be interested to know if you're running a version before that [12:57] bac: because perhaps i didn't really fix the problem... [12:58] rogpeppe: actually i was running older than that. looks like 1130 [12:58] bac: hmm, i'm a bit surprised it doesn't fail for you [12:59] bac: but that bug doesn't have much info on how to reproduce it... [13:03] rogpeppe: we'll see what happens now [13:03] rick_h_, fwiw the jenkins failure we just had (in IE) was browser search widget: should support search change events: timeout of 200ms exceeded. I am guessing that's an intermittent one, unfortunately. retrying, but if you all have ideas on making that more reliable it would be fab [13:05] gary_poster: ok, hatch and I were talking about it a little bit the other day. There's one idea to try out on it. I'll file a bug/card for it. [13:17] cool thanks hatch [13:18] bah [13:18] cool thanks rick_h_ [13:22] gary_poster: I commented on the ID-munging branch not realizing that it had already landed. [13:23] I think the things I noted are worth another branch. [13:24] benji, thanks for safe id review comments. This is landed already, so I passed them on to Francesco, who is working in a related part of the codebase in his current branch. FWIW, my choice of underlines only within the .replace was conscious--in fact, I initially had what you describe. I decided I wanted a dash to separate the replaced value from the hash value, and to have the division between the two visually c [13:24] lear by not usuing hyphens elsewhere. I don't think the decision is problematic (unlike, potentially, your comment about needing to not start with a number--actually in html 5 you can, and in html 4 you can't but you also can't start with a dash, and I forget about the CSS rule but it is in the bug, but anyway it sounds fine to me). Francesco and I agreed that he would add the character, and I said that he ought t [13:24] o do whatever he wanted with the dash stuff, which IMO is entirely stylistic. [13:24] benji, I started writing the above not realizing that you would write about it here before I finished. Or that I would write a novel. :-) [13:24] heh [13:25] sounds like it's being handled [13:25] cool [13:29] frankban, rogpeppe: ok, with juju-core trunk i can now deploy dokuwiki. frankban did the error in bug 1167344 happen frequently? was there any way to trigger the failure? [13:29] <_mup_> Bug #1167344: Deployment via the juju-core API fails with a "no options to set" error in some cases

< https://launchpad.net/bugs/1167344 > [13:30] bac: the only way to dupe I know is deploying dokuwiki from the GUI. [13:31] frankban: how often did you see the failure? [13:31] interesting migration comparison story of backbone style to angular .. http://www.localytics.com/blog/2013/angularjs-at-localytics/ [13:31] bac: in trunk you should not get that error because that error no longer exists... [13:31] bac: IIRC, it always failed [13:32] rogpeppe: ok. [13:35] I am getting a connection refused error when connecting to the juju websocket; is this a known issue? [13:39] gary_poster: do we want to keep bug 1166153 around since we aren't using YUI 3.10 and it may be a bug in their code which is a pre-release version? [13:39] <_mup_> Bug #1166153: The env view slider does not work in YUI >= 3.10 when the thumb is used to zoom in/out < https://launchpad.net/bugs/1166153 > [13:40] bac, how bout mark as Incomplete [13:40] gary_poster: but keep the card in the lane? [13:40] bac no [13:41] delete card [13:41] gary_poster: +1 [13:41] thank you for the gardening [13:44] ugh, if you destroy a service and deploy another juju-core doesn't re-use the existing machine but creates a new one. rogpeppe did pyjuju work that way? is there a plan to re-use machines? [13:45] pyjuju reuses [13:45] pretty sure [13:45] bac: reusing is bogus, i'm afraid [13:45] bac: because a service has a right to assume it's being started on a pristine machine [13:45] rogpeppe: yuck. [13:45] bac: well, in a pristine *environment* anyway [13:46] bac: when we implement containerisation, we'll be able to reuse machines [13:46] containerisation will be the default deployment behavior then, is the plan? [13:46] gary_poster: yes [13:46] cool [13:46] gary_poster: (probably) [13:46] yeah, assuming it is stable enough etc :-) [13:47] bac: in the meantime, there's this in the offing: https://codereview.appspot.com/8520043/ [13:47] frankban https://codereview.appspot.com/8657043/ LGTM [13:48] gary_poster: cool, thanks [13:55] Makyo, could you please send a quick email to hazmat documenting what you think he can show off on the tablet? To re-re-reiterate, "pretty picture only" is completely fine, per re-re-reconfirmation yesterday from Robbie via Deryck [13:55] gary_poster, Sure. [13:55] thank you [13:58] morning [13:59] hatch: once you grab a coffee I wanted to see if you could block out some time to chat url adjusting, dispatch, etc for subapp/app/etc. [14:00] sure thing....10mins/ [14:00] ? [14:00] sure [14:06] very small review request for small sandbox fix (card is in story A/review): https://codereview.appspot.com/8663043 [14:07] gary_poster: i'll do the review but i don't find the card [14:07] bac thank you. I sometimes have to reload? "Fix sandbox setconfig and speed up test" [14:07] i stitle on board [14:08] I have to resize my root partition, it has reached capacity. I'll (hopefully) be back soon. [14:20] guihelp: could anyone confirm bug 1167967? thanks, if it's confirmed I will create a card [14:20] <_mup_> Bug #1167967: The GUI hangs when a service is destroyed from the service detail view < https://launchpad.net/bugs/1167967 > [14:20] (not related to juju-core) [14:23] frankban: yes, i see that problem on uistage [14:23] bac: ok thanks [14:23] after destroy it tries to stay on the service view, an error pops up, and the main canvas still shows the service. after reload the service is gone [14:25] bac: I'll add this info to the bug. card created in High-Priority Maintenance [14:26] gary_poster, alright if I land the add relation touch branch? [14:26] frankban, y [14:26] sorry Makyo y [14:26] gary_poster, thanks [14:43] bac thanks. one more review of https://codereview.appspot.com/8663043/, anyone? [14:46] It's so small! so fast! [14:49] I'll take it. [14:49] ty :-) [14:55] I now have more (unwanted) experience recovering from grub boot failures. [15:01] anyone need any reviews? [15:04] hatch, a little birdie (well, not that little actually ;-) ) told me you're a node.js expert and I have a few questions, how about a quick hangout when you have time? [15:04] rogpeppe: I see juju failing to destroy some services, maybe those in an error state, but not sure. Is it a known issue? if not do you want to try to dupe? [15:05] frankban: what are the symptoms? [15:05] rogpeppe: the service is still there after a "juju destroy-service" call [15:05] frankban: for how long? [15:05] frankban: does it remain forever? [15:06] rogpeppe: it's about five minutes now. in the delta, I've seen a "change" entry but not a "remove" entry. everything works well (and fast) for other services. [15:08] rogpeppe: the only difference is that the failing service (buildbot-master) is in an error state. could you try to dupe? just 1) bootstrap an env, 2) deploy buildbot-master, 3)wait till status says 'hook failed: "install" and 4) destroy buildbot-master [15:08] frankban: interesting [15:08] frankban: will try to dupe [15:08] rogpeppe: btw, the error state could be a coincidence [15:08] frankban: maybe... [15:09] bbiab [15:09] teknico: lol...damn birds! I'm just eating some breakfast right now :) chat in 5? [15:09] hatch, yes, five or so, back soon [15:09] bcsaller, we need to deliver the charm for IS by tomorrow. If you need additonal hands on task, please let me know. [15:10] We can improve charm next week, but we need functionality tomorrow [15:10] gary_poster: good info, I'll let you know at the meeting one way or the other [15:10] cool thx bcsaller [15:11] * gary_poster takes ubuntu touch tablet to UPS. biab [15:23] back. hatch, when you're ready, no hurry [15:23] sure now works [15:23] guichat [15:50] jujugui, like yesterday.please update kanban board to reflect reality in next few minutes. I will review 3 or 4 minutes before call to try to expedite call [15:51] gary_poster: can you fill us in on the canonistack stuff? The email didn't include any details :) [15:52] hatch, oh, right! I meant to advertise widely. summary: canonistack is not meant to be reliable. if you want something reliable, you have their blessing to look elsewhere. deryck is verifying [15:52] to make sure they say the same thing twice :-) [15:52] but that's what they said [15:52] so we can switch to ec2 or hp cloud [15:52] well there is a difference between reliable and not even able to start....but ok :) lol [15:53] yea, thought reliable was 'your instance can go away' not 'no instances for you ever' [15:53] but if we can switch that's also good news [15:53] rick_h_, hatch, heh, yeah, I said we could work with 95% reliable but not with <10% reliable :-) [15:54] haha [15:54] jeesh we could even work with 75% reliable :D [15:54] :-) [15:57] jujugui call in 2; reviewing kanban [15:57] 3, that is [16:09] jovan2 luca__ you guys have a minute? [16:10] hi rick_h: sure [16:10] https://plus.google.com/hangouts/_/5fea451cd85e565fc97faa543c4f904f4c39dac0?authuser=0&hl=en jovan2 luca__ [16:15] hatch: tomorrow's meeting i go to the roof! [16:16] haha is the weather that nice there all the time? [16:20] does anyone know of any two charms with multiple possible relationships? [16:20] jujugui ^ [16:21] save me from having to write out a fake charm data [16:22] hatch, hadoop to itself. [16:22] 8 possible relations. [16:22] nice [16:22] hatch: mysql-mediawiki [16:22] Yeah, that too, should have a slave and a db. [16:23] ooo I missed tests for peer relations [16:23] thanks Makyo :) [16:23] hatch, well, this was one hadoop instance to another, but hey, bonus :) [16:26] mysql and mediawiki don't have multiple possible requires=provides interfaces :( [16:27] Makyo: can hadoop actually relate to itself? it doesn't share any of the same interface names in it's provides/requires [16:28] it has namenode, ganglia, jobtracker vs secondarynamenode, datanode, tasktracker, mapred-namenode [16:28] hatch, I see 8 possible relations (well, four, but going both ways) on uistage [16:29] *EXPLICIT* *EXPLICIT* *EXPLICIT* *EXPLICIT* [16:29] so it relates on the interface type not the interface name :/ [16:32] ugh [16:35] so now I'm totally confused - why are there multiple ways to specify the same interface? [16:36] hadoop for example has secondarynamenode, datanode and mapred-namenode all pointing to dfs [16:36] are each one handled by a different hook? [16:42] for example serviceA:foo could relate to serviceB:bar if their interfaces are the same [16:43] the name is irrelevant [16:43] hatch, bcsaller is the one to talk to. Everyone else will be guessing at least in part (at least I would be) [16:44] hatch: the name of what you call the relationship in your charm determines what hook get called. Each relationship has its own hooks. [16:45] ok so serviceA:bar could relate to serviceB:foo as long as their interfaces are the same [16:45] even though they are named something totally different [16:45] hatch: yes, its the interface that is important, not what the charm calls it locally. Sorry if that wasn't clear [16:46] benji, your current work is, partly or completely, in juju core/go? [16:46] gary_poster: I think everything I need to do is constrained to the gui [16:47] (but I have to get juju to run to actually see it work) [16:51] ctrl+a del [16:51] cool thanks benji. did you get help from rogpeppe? [16:52] gary_poster: we're in the middle of diagnosing now; he and mgz seem to have tracked it down, but it doesn't sound like a quick fix [16:52] uh oh [16:57] jujugui, heads up: AIUI, Tuesday is the last day to make any changes to Juju Core. Hopefully we are set *now*, but if we are not, the sooner we discover a bug in Juju Core the better. [17:25] jovan2: ping, is the font color for the "Recent activity" line in the charm token that orange color used in links? [17:26] jovan2: it's blocked out as "(Charm activity info?)" and I wondered if that orange was part of noting it, or actual for the text [17:26] rick_h_: no it should not be orange [17:26] jovan2: the reason I ask is that otherwise the recent activity blends into the description below [17:26] jovan2: so I was trying to see how to seperate it visually [17:27] rick_h_: greg has done some work on the visual design today and hope to issue it to you tomorrow after review with Alehandra [17:28] jovan2: ok, I'll leave it as is for now then. http://uploads.mitechie.com/lp/recent_activity_blends.png is what I mean for the record [17:28] rick_h_: yeah I see it's a problem but leave as is for now [17:29] jovan2: k, thanks [17:48] doh and he's gone [18:52] AssertionError: expected 'No shared interfaces.' to equal 'No shared Interfaces.' [18:52] umm [18:52] anyone else see a difference there? [18:53] oh [18:53] nm [18:53] :) [18:58] case matters ftw [19:01] I swore I read that line 20 times and didn't see that hah [19:01] it could also be that i'm "hate coding" right now because I'm sick of rewriting this relations code :D [19:02] good news is that it gets better every time....well avoiding the fact that it was done to the wrong spec the first two times [19:02] s/avoiding/ignoring [19:02] :D [19:08] hatch: got a second for a quick review? https://codereview.appspot.com/8561048/ [19:08] sure [19:08] thanks [19:11] donezo [19:11] awesome, appreciate it ( jcsackett ^ ) [19:13] rick_h_: oh, thanks. [19:31] is anyone else seeing "[14:30:34.429] The connection to wss://ec2-72-44-32-91.compute-1.amazonaws.com:17070/ was interrupted while the page was loading. @ http://localhost:8888/juju-ui/assets/javascripts/reconnecting-websocket.js:80" when connecting to the go environment? [19:35] benji, I got that when I rebootstrapped, and had to go delete the certificates from FF, then re-accept them manually. [19:36] unfortunately, I have already gone through that procedure... no wait! I have rebootstrapped, thanks Makyo [19:41] Hmm. It appears that destroying a service while it's pending removes the service and unit(s), but not the machine(s). [19:42] And destroying a service that is stuck in pending or in error fails silently. [19:43] Not having any GUI-core communication issues, though. [19:57] great Makyo. please file those juju core bugs and maybe put them in our tracking? Also, benji or Makyo, would be cool to see that rebootstrap thing in the blog notes, like Makyo's earlier helpful blog. [19:58] review request: small changes, and would particularly like bcsaller to look because I touched the router. https://codereview.appspot.com/8680043 [19:59] gary_poster: reading it now [19:59] thank you [19:59] I'll do the blog thing real quick, then bugs. [19:59] cool thanks [20:58] any one up for a review of destroy_service addition to fakeness? https://codereview.appspot.com/8684043 [20:59] * bac -> dw [21:32] * gary_poster will if he has time before dinner [22:24] Makyo, you still around to review https://codereview.appspot.com/8680043/ by chance? (hatch, I'd ask you, but I bet you are trying to get your branch wrapped up) [22:25] I can do it [22:25] I'm around, too. [22:25] ok you can do it [22:25] :) [22:25] (bac I decided not to review yours till after you replied to Ben) [22:25] ok thanks, hatch, Makyo :-) [22:32] gary_poster: ok. thanks for the review bcsaller