[00:00] <niemeyer> SpamapS: Can we handle that separately?
[00:00] <niemeyer> SpamapS: That's a different use case from the problem we were talking above
[00:00] <SpamapS> Sure, but its going to come up with any complex apps.
[00:01]  * SpamapS sticks it in the "forget this today, remember tomorrow" bin
[00:01] <niemeyer> SpamapS: Hopefully we'll have _many_ problems to solve! ;-)
[00:01] <SpamapS> I'm ok with adding a 'shared-db' relation on the mysql formula when the time comes.
[00:01] <SpamapS> interface: mysql-shared
[00:01] <niemeyer> SpamapS: It's just that it's not productive to try to solve orthogonal problems at the same time
[00:02] <SpamapS> I wasn't aware that we were trying to solve another problem. :)
[00:02] <niemeyer> SpamapS: You have mentioned two things you'd like to do:
[00:02] <niemeyer> SpamapS: 1) I can't tell what is the right database to clean up on relation-broken
[00:03] <niemeyer> SpamapS: 2) I want to share a database between multiple services
[00:03] <niemeyer> SpamapS: These are orthogonal problems, with potentially independent solutions
[00:04] <niemeyer> SpamapS: If you want to switch over to (2) and leave (1) unsolved for the moment, I don't think a shared-db relation is the right thing to do.
[00:05] <niemeyer> SpamapS: This doesn't scale.. now I have two pairs of services that should use the same database
[00:05] <niemeyer> SpamapS: But each pair should use a different one
[00:05] <niemeyer> SpamapS: They can't all be put in the same shared-db relation
[00:07] <SpamapS> niemeyer: I see, the 2nd I'm not all that interesting in solving today, no. the 1st, I think I agree 100% that a relation-id, exposed to all hooks, solves the issue.
[00:07] <niemeyer> SpamapS: Cool, that's awesome
[00:07] <niemeyer> SpamapS: It's already scheduled for this milestone, so let's push it forward!
[00:09] <_mup_> ensemble/expose-provision-service-hierarchy r282 committed by jim.baker@canonical.com
[00:09] <_mup_> Spike on removing the observation around watches being stopped
[00:09] <niemeyer> Alright, and that's dinner time ticking here
[00:09] <niemeyer> I'll step out.. have a good time all!
[00:13] <jimbaker> niemeyer, take care
[00:56] <_mup_> ensemble/expose-provision-service-hierarchy r283 committed by jim.baker@canonical.com
[00:56] <_mup_> Fixed remaining non corner case tests
[01:04] <_mup_> ensemble/expose-provision-service-hierarchy r284 committed by jim.baker@canonical.com
[01:04] <_mup_> Removed unnecessary splitting of callbacks from their watch functions
[04:34] <_mup_> ensemble/expose-provision-service-hierarchy r285 committed by jim.baker@canonical.com
[04:34] <_mup_> Cleaned up watch_exposed_flag callback
[05:02] <_mup_> ensemble/expose-provision-service-hierarchy r286 committed by jim.baker@canonical.com
[05:02] <_mup_> Naming cleanup and other style changes
[05:51] <_mup_> ensemble/expose-provision-service-hierarchy r287 committed by jim.baker@canonical.com
[05:51] <_mup_> Ensure testing does not get confused with background activity around watch removal
[05:54] <_mup_> ensemble/expose-provision-service-hierarchy r288 committed by jim.baker@canonical.com
[05:54] <_mup_> Removed unnecessary guard from watch_exposed_flag
[09:03] <kim0> morning everyone
[11:29] <daker> anyone knows how to execute a sql file with MySQLdb in python ?
[11:44] <kim0> daker: why not execute mysql < file.sql
[13:20] <daker> woo my formula is working now :D
[13:40] <niemeyer> Greetings
[13:46] <m_3> morning
[13:49] <kim0> morning
[13:51] <kim0> niemeyer: Is the ppa going to always be daily build from trunk ? 
[13:51] <niemeyer> kim0: It should
[13:51] <niemeyer> kim0: You mean it's not working, or are you wondering about the future?
[13:52] <kim0> future :)
[13:52] <kim0> niemeyer: and universe should be an older and better tested version ?
[13:52] <kim0> for 11.10 at least
[13:52] <kim0> I'm doing the zero to ensemble screencast .. and was wondering how to frame it
[13:59] <niemeyer> kim0: Yeah, in the future we'll have more stable point-in-time releases for sure
[14:00] <niemeyer> kim0: We're just moving fast for now, and deploying important features very often
[14:00] <kim0> ok
[14:00] <niemeyer> kim0: So it makes more sense to have these well tested and having people able to play with them than freezing
[14:13] <daker> anyone want to test my formula ?
[14:16] <kim0> daker: is it working well for ya :)
[14:16] <daker> yep
[14:16] <kim0> woohoo
[14:16]  * kim0 hugs daker 
[14:16] <kim0> I'll definitely play with it tonight
[14:17] <kim0> daker: I'm sure we're interested in your overall experience developing it
[14:17] <kim0> let us know any thoughts you have
[14:18] <niemeyer> daker: Yeah, where is it?
[14:18] <daker> one sec
[14:24] <daker> niemeyer, bzr branch lp:~daker/+junk/joomla-formula
[14:25] <niemeyer> daker: Cool
[14:29] <daker> niemeyer, are you going to test it ?
[14:29] <niemeyer> daker: Yeah, checking it out already
[14:30] <daker> to access the admin just type : public_dns/administrator
[14:31] <daker> login : admin|passe: ensemble
[14:33] <niemeyer> daker: Cool
[14:33] <daker> it works ?
[14:37] <daker> niemeyer, ^
[14:37] <niemeyer> daker: Sorry, multitasking, but still firing it up
[14:37] <daker> ok
[14:56] <niemeyer> daker: Works!
[14:57] <niemeyer> Sweet
[15:02] <daker> whooo :D
[15:06] <niemeyer> daker: We need a way to communicate values such as username/pass to the admin
[15:06] <niemeyer> daker: For the time being, it may be worth putting these instructions in the description
[15:07] <daker> yes
[15:58] <hazmat> niemeyer, alternatively those could be default values for settings, although we would need a corresponding admin cli way of inspecting those
[16:00] <niemeyer> hazmat: Oh, that sounds much better
[16:00] <niemeyer> hazmat: We certainly need a way of inspecting the configuration either way
[16:01] <niemeyer> hazmat: We can have a default type of value which is automatically generated if unset
[16:01] <niemeyer> This would avoid everyone having the same password by default
[16:02] <hazmat> niemeyer, that sounds nice as well
[16:02] <_mup_> Bug #797241 was filed: Ensemble cli subcommand for inspecting config-settings <Ensemble:New> < https://launchpad.net/bugs/797241 >
[16:03] <_mup_> Bug #797243 was filed: A service config field schema that allows for a  random default value <Ensemble:New> < https://launchpad.net/bugs/797243 >
[16:19] <niemeyer> hazmat: Thanks for filing these
[16:21] <hazmat> niemeyer, np. 
[16:22] <hazmat> niemeyer, out of curiosity have you had any stability issues on natty?
[16:22] <niemeyer> hazmat: Yes.. I have had to kill compiz every few days
[16:23] <hazmat> niemeyer, thanks good to know for comparison.. my uptime is averaging about 6-8hrs, even reverted to classic, but still getting random freezes.. just ordered a new laptop hopefully it can help out
[16:24] <niemeyer> hazmat: Holy crap
[16:24] <niemeyer> hazmat: 6/8h is on the low side indeed
[16:24] <hazmat> niemeyer, yeah.. its quite disruptive, it was better during the beta cycle
[16:24] <niemeyer> hazmat: What kind of issue you notice?
[16:26] <hazmat> niemeyer, mostly hard X locks, which aren't recoverable. also attaching/detaching external monitors also seems to trigger an unrecoverable state, the old virtual terminal unity --replace trick works sometimes.. but not always.. typically a slew of session errors, some referencing dbus, some referencing device errors.
[16:27] <niemeyer> hazmat: Hmmm, ok
[16:27] <_mup_> ensemble/standardize-log-testing r256 committed by jim.baker@canonical.com
[16:27] <_mup_> Merged trunk
[16:27] <niemeyer> hazmat: Sounds related to what I see
[16:27] <niemeyer> jimbaker: Morning
[16:27] <jimbaker> niemeyer, hi
[16:28] <niemeyer> hazmat: Generally CTRL-ALT-F1 + compiz kill + unity restart works
[16:28] <niemeyer> Not nice, though
[16:28] <hazmat> niemeyer, yeah.. its better than killing the xsession.. i'm curious if there's equivalent to that for classic mode
[16:30] <niemeyer> hazmat: This is just restarting the window manager, I think
[16:30] <niemeyer> hazmat: I recall doing that for metacity long ago
[16:40] <niemeyer> While working on that debug-hook stuff, I'm quite proud of this system we came up with
[16:45] <m_3> niemeyer: I've got lockup problems with natty/compiz/unity too... half-minute lockups every time I do something compiz-fancy like alt-tab or switch viewports
[16:46] <niemeyer> hazmat: jimbaker: Do you want to have a quick look at these doc changes before I commit as part of the debug-hooks branch: http://paste.ubuntu.com/626637/
[16:46]  * hazmat checks it out
[16:46] <jimbaker> niemeyer, sure
[16:46] <niemeyer> Ben correctly requested the changes during review
[16:48] <SpamapS> daker: reading your joomla formula now. :)
[16:48] <_mup_> Bug #797263 was filed: ~/.ensemble permissions <Ensemble:New> < https://launchpad.net/bugs/797263 >
[16:51] <jimbaker> line 47: "At this point," would be better; line 59: maybe better word choice "same way performed";  line 102, "work" instead of "worked", but maybe better word choice there too
[16:51] <jimbaker> niemeyer, looks good with these changes
[16:52] <niemeyer> jimbaker: Super, thanks
[16:53] <hazmat> niemeyer, looks good, there is one bug people will hit following those discussion though, namely that the relation api isn't available yet outside of relation hooks, the relation identifiers is the first step to resolving it. it might be nice to reference the bug as a caveat.. its pretty easy to spot and remove it after the bug is fixed.
[16:53] <hazmat> s/discussion/documentation
[16:54] <niemeyer> jimbaker: Just the "worked" I'm going to keep as it is.. it's describing the process of transition and choice we made at a particular point, rather than a statement of how it works _today_
[16:55] <niemeyer> hazmat: Sounds good.. do you have a link to the particular bug you'd like to see ther?
[16:55] <niemeyer> e
[16:56] <jimbaker> niemeyer, ok, i guess the context is clear enough that this is the case
[16:56] <hazmat> niemeyer, bug 767195
[16:56] <_mup_> Bug #767195: Ensemble should have hook cli apis to enumerate and interact with all the relations of a unit. <Ensemble:New> < https://launchpad.net/bugs/767195 >
[16:57] <niemeyer> hazmat: That doesn't seem related
[16:58] <hazmat> niemeyer, yeah.. its addressing the same issue from a different angle.. which is relation api usage in non relation hooks
[16:58] <m_3> Have a what-if scenario I'd like some advice on...
[16:58] <m_3> let's say a mysql formula exposes a config parameter for something like query_cache_size
[16:59] <m_3> (in config.yaml as I understand)
[16:59] <m_3> when another formula adds this as a relation
[16:59] <m_3> and needs to change that config parameter
[16:59] <niemeyer> hazmat: I see what you mean, but this isn't the actual problem
[16:59] <m_3> can it do something equivalent to the 'ensemble set' command
[17:00] <niemeyer> hazmat: Fix Releasing this bug doesn't necessarily enable people to use relation-get there
[17:00] <m_3> i.e., are there ways to parameterize a relation?
[17:00] <niemeyer> hazmat: Well, enumerate _and_ interact.. sorry
[17:00] <niemeyer> hazmat: My bad
[17:01] <niemeyer> We should probably split that off, though.. there are really two different problems
[17:02] <niemeyer> m_3: No, it'd probably be better to follow up a bit on the documentation to get a better understanding of how things get together
[17:02] <m_3> ok, gotcha
[17:02] <niemeyer> m_3: The goal of configuration is "human oriented" settings
[17:02] <m_3> niemeyer: ah... ok
[17:02] <niemeyer> m_3: That includes "ensemble set", config-get, etc
[17:03] <_mup_> ensemble/standardize-log-testing r257 committed by jim.baker@canonical.com
[17:03] <_mup_> Remove unnecessary new import
[17:03] <niemeyer> m_3: Inter-unit/service communication is done via relations
[17:04] <niemeyer> As an aside, it's really good the fact that the config system we're putting in place is read-only from the formula perspective
[17:04] <niemeyer> I've seen quite a few people saying "Hey, I want to write to the configuration to communicate between formulas"
[17:05] <niemeyer> The fact it's read-only guides people to do the right thing naturally
[17:06] <m_3> niemeyer: so relation-specific configuration has to be either in the relation hooks themselves or passed in from the config of the dependent formula?
[17:07] <_mup_> ensemble/standardize-log-testing r258 committed by jim.baker@canonical.com
[17:07] <_mup_> Remove similar unnecessary import
[17:07] <niemeyer> m_3: relation-specific configuration is always performed by the formulas in the formula hoks
[17:07] <niemeyer> hooks
[17:08] <m_3> niemeyer: ok, thanks
[17:11] <niemeyer> m_3: np
[17:11] <niemeyer> hazmat: Is this what you meant: http://paste.ubuntu.com/626659/ ?
[17:11] <niemeyer> hazmat: /Limitations
[17:12] <hazmat> niemeyer, sounds good... although relation-broken should still work afaicr
[17:12] <hazmat> its a relation hook
[17:13] <niemeyer> hazmat: It's not a running relation hook
[17:13] <niemeyer> hazmat: It's a relation hook, though
[17:13] <hazmat> the open bug on relation-broken is really about adding the relation name as an additional environment variable to the relation hook.
[17:13] <hazmat> or the id
[17:13] <hazmat> hmm.. actually it was getting the related service name into the environment
[17:14] <niemeyer> hazmat: relation-broken has the same issue as the other hooks.. there's no specific relation on the other side to talk to
[17:14] <niemeyer> s/specific relation/specific unit/
[17:14] <hazmat> niemeyer, sure.. but in terms of implementation ... relation-list, relation-get, relation-set should still work in relation-broken
[17:15] <hazmat> ie. its unrelated to the bug referenced
[17:16] <niemeyer> hazmat: Hmm,  I see
[17:16] <kim0> zero to ensemble screencast → http://www.youtube.com/watch?v=qxMhKbDSbOw
[17:17] <hazmat> kim0, awesome! we should embed onto the wiki frontpage
[17:17] <niemeyer> hazmat: replaced broken with upgrade there
[17:18] <hazmat> niemeyer, cool, +1
[17:18] <_mup_> ensemble/debug-hook-fixes r248 committed by gustavo@niemeyer.net
[17:18] <_mup_> Fixed documentation to refer to tmux, as requested by Ben. [r=hazmat,jimbaker]
[17:18] <niemeyer> kim0: Woah
[17:18] <kim0> \o/
[17:19] <kim0> pushing on planet-ubuntu 
[17:20] <niemeyer> kirkland: I suppose you're not that guy: http://twitter.com/#!/bigtexan13
[17:21] <kirkland> niemeyer: sweet!
[17:21] <kirkland> niemeyer: you've found my alter ego
[17:21] <niemeyer> kirkland: A bit unfortunate, to say the least
[17:22] <niemeyer> kirkland: What's your actual twitter account?  I'd like to retweet your formulas post with actual credits
[17:22] <kirkland> niemeyer: dustinkirkland
[17:22] <niemeyer> kirkland: Hah!
[17:22] <niemeyer> kirkland: Thanks :)
[17:23] <kirkland> niemeyer: highly unoriginal
[17:23] <kirkland> niemeyer: i'm working on two more, for AjaxTerm, and screenbin
[17:23] <niemeyer> kirkland: Sweeeet
[17:24] <kim0> kirkland: oh that is awesome
[17:24] <m_3> ok, so what's the party line here... formulas or formulae?
[17:24] <kim0> we can use those in next irc tuition weeks :)
[17:24] <kirkland> m_3: i say formulae, as I love my latin ;-)
[17:24] <kirkland> m_3: oh, and principia is a latin word ;-)
[17:24] <niemeyer> m_3: We're a heavily distributed company, so we can't really rule that out! ;-)
[17:25] <m_3> kirkland: ha!
[17:25] <hazmat> npr kojo n. show is doing right now on cloud computing
[17:25] <niemeyer> I personally Go with formulas, because that matches the Portuguese word as well
[17:25] <kirkland> m_3: oh, you need to meet negronjl 
[17:25] <niemeyer> and the "ae" termination is alien to me
[17:25] <niemeyer> and I suck at languages and am lazy.. all of these
[17:25] <kirkland> m_3: he's written some puppet modules for deploying hadoop;  would be a great place to start your ensemble/hadoop work
[17:26] <m_3> kirkland: great thanks
[17:26] <m_3> kirkland: yeah, just finished with travel stuff and basic canonical setup stuff (although I question if I'll ever get mumble set up properly)
[17:26] <hazmat> yeah.. i'd vote for formulas as well.. no need for additional latin obscurity ;-)
[17:27] <kirkland> m_3: heh
[17:27] <m_3> kirkland: just branched ensemble and am digging through code
[17:27] <m_3> niemeyer: and docs!
[17:27] <negronjl> hi m_3
[17:28] <niemeyer> m_3: Yeah, I guess we'll have to do some voting there.. :-)
[17:28] <m_3> hi negronjl, great to meet you
[17:29] <negronjl> m_3:  nice to meet you.
[17:29] <m_3> negronjl: Dustin sent me some repo links a couple of days ago... lemme dig
[17:30] <negronjl> m_3:  ping me when you get settled and if you have questions.  I am sure that we should be able to port the orchestra-modules to ensemble formulas relatively quickly ( hadoop one comes to mind )
[17:30] <m_3> negronjl: awesome man... thanks!
[17:31] <negronjl> m_3:  np
[17:31] <_mup_> ensemble/trunk r253 committed by gustavo@niemeyer.net
[17:31] <_mup_> Merged debug-hook-fixes branch [r=hazmat,bcsaller,jimbaker]
[17:31] <_mup_> This branch fixes a number of problems in the debug-hooks functionality,
[17:31] <_mup_> and switches to using tmux for solving some of them. For instance:
[17:31] <_mup_> - joined and departed hooks are now valid
[17:31] <_mup_> - Sometimes screen was being fired with the ubuntu user
[17:31] <_mup_> - Sometimes screen was firing two independent sessions with the same name
[17:31] <_mup_> - The exit handler was only called on HUP
[17:31] <_mup_> - If the hook.sh shell died for whatever reason, it would hang forever
[17:32] <m_3> negronjl: I was planning on starting with something simple like hdfs or zookeeper from scratch, but now that we have the orchestra modules, I'll start with those.
[17:33] <niemeyer> Well written spam.. got through the spam filter, and I almost thought it was a real one
[17:34] <niemeyer> This was the issue, though: PS should I speak with someone else at niemeyer?
[17:35] <negronjl> m_3:  sure.  let me know if you have questions about it.
[17:35] <negronjl> m_3:  the plan is, at some point, to port the modules into formulas so both projects benefit from them.
[17:36] <m_3> negronjl: thanks! I'll hit you up for a higher-bandwidth conversation about it after I dig through them
[18:10] <kim0> Oh we did hit linuxtoday yeeha .. I did submit it, but somehow missed it when they published it
[18:11] <kim0> http://www.linuxtoday.com/it_management/2011060900341NWSVUB
[18:13] <_mup_> ensemble/expose-provision-service-hierarchy r289 committed by jim.baker@canonical.com
[18:13] <_mup_> Better logging
[18:23] <niemeyer> kim0: Do you have a blog post for the video yet?
[18:23] <kim0> niemeyer: yeah pushed
[18:23] <kim0> http://cloud.ubuntu.com/2011/06/zero-to-ensemble-in-5-mins/
[18:23] <kim0> niemeyer: ^
[18:24] <niemeyer> kim0: Awesome, will tweet that
[18:24] <kim0> great thanks
[18:25] <niemeyer> kim0: Thank you!
[18:27] <niemeyer> kim0: Just watching it now
[18:28] <kim0> oh if it's horrible we can do others to replace it :)
[18:28] <niemeyer> kim0: So far it's awesome :-)
[18:28] <niemeyer> kim0: Very smooth and easy going
[18:29] <kim0> glad you like it
[18:31] <niemeyer> kim0: Yeah, brilliant stuff
[18:31] <kim0> woohoo great
[18:33] <_mup_> ensemble/expose-provision-service-hierarchy r290 committed by jim.baker@canonical.com
[18:33] <_mup_> PEP8, comments, demoted log levels
[18:42] <_mup_> ensemble/expose-provision-service-hierarchy r291 committed by jim.baker@canonical.com
[18:42] <_mup_> Cleanup wrt review points
[18:44] <kim0> btw the docs at the installation from ppa step .. is missing a "sudo apt-get update" .. should I file a bug for that ;)
[18:56] <niemeyer> kim0: In a next screencast, it might be worth pointing out that both formulas use the same base image, and that it's the formula itself that defines how it works
[18:56] <niemeyer> kim0: Also that we'll enable multiple formulas in an image, etc
[18:56] <niemeyer> kim0: Erm, multiple formulas in a machine
[18:57] <niemeyer> kim0: The perfect is the enemy of the good enough, though.. it's fantastic to have _something_ out there
[18:57] <kim0> Yeah, this one only scratches the surface ..
[18:57] <kim0>  there will be deeper dives
[18:59] <niemeyer> kim0: We have to attempt to emphasize the points that really make it stand out
[18:59] <kim0> yeah agree .. I'll make sure to mention that next time indeed
[19:04] <jimbaker> niemeyer, do you want to have standup today?
[19:04] <jimbaker> regardless, i'm pushing on getting my outstanding branches fixed with respect to their reviews
[19:13] <niemeyer> jimbaker: Have you read the email related to standups from yesterday?
[19:15] <jimbaker> niemeyer, reading it now
[19:16] <niemeyer> jimbaker: Hold on, I'll add an extra entry there:  - Read your email. ;-)
[19:16] <jimbaker> niemeyer, ok :)
[19:19] <koolhead17> hi all
[19:19] <koolhead17> kim0: thanks!! :)
[19:19] <koolhead17> SpamapS: hey
[19:20]  * kim0 nods
[19:20] <kim0> koolhead17: have fun :)
[19:20] <koolhead17> :P
[19:20] <niemeyer> jimbaker: On expose-watch-exposed-flag, you don't have to move it to WIP
[19:20] <niemeyer> jimbaker: It's approved, and still pending another review
[19:21] <niemeyer> koolhead17: Hey there
[19:21] <koolhead17> niemeyer: hello :)
[19:22] <jimbaker> niemeyer, ok - i just wanted to indicate i'm making some small changes in response to the review points, which also address downstream needs, and they are about to land...
[19:23] <niemeyer> jimbaker: That's fine.. but by moving it to WIP you remove any chance of someone else looking at your branch at that time
[19:23] <jimbaker> niemeyer, understood - i will be more careful for sure
[19:23] <niemeyer> jimbaker: Sometimes it's worth moving it to WIP, when the changes are pervasive and you'd rather have the next reviewer looking at the new version instead
[19:24] <niemeyer> jimbaker: For this branch, that's not the case since it was basically a +1 with mionrs
[19:24] <niemeyer> minors
[19:24] <jimbaker> this is one aspect of the new review process i was not aware of, so basically leave it in the state left by the reviewers, unless otherwise negotiated
[19:25] <jimbaker> stating it this way describes the continuity of our review process, of course :)
[19:27] <niemeyer> jimbaker: It really depends on the intention
[19:27] <niemeyer> jimbaker: If you want to get a second review, don't move it to WIP
[19:27] <niemeyer> jimbaker: If you want the second review to look at the new version because there's something important/significant there, move it to WIP
[19:28] <jimbaker> niemeyer, got it
[19:43] <kirkland> hey guys ... when writing a formula that requires user input at deploy
[19:43] <kirkland> ie, i need to prompt the user to choose a password
[19:43] <kirkland> where does this go?
[19:43] <kirkland> i need to obtain it from the user at deploy time, and then some how get it over to the install hook
[19:45] <hazmat> kirkland, there's some work on service config that's almost done to help with service configurable settings
[19:46] <kirkland> hazmat: hmm, okay
[19:47] <hazmat> kirkland, there's a new hook cli api to retrieve the settings, and a new hook that gets invoked when the config changes as part of that work
[19:47] <kirkland> hazmat: neat
[19:58] <niemeyer> kirkland: We were also just talking about this today
[19:58] <niemeyer> kirkland: We'll have a kind of configuration that will create passwords automatically
[19:59] <niemeyer> kirkland: So that the user doesn't need to intervene during deployment
[19:59] <kirkland> niemeyer: how is the user informed of the password?
[19:59] <kirkland> niemeyer: in my case, they need it
[19:59] <niemeyer> kirkland: Meanwhile, our recommendation is to have a pre-defined password and putting it in the description
[19:59] <niemeyer> kirkland: They can inspect the settings of any service
[20:00] <kirkland> niemeyer: hmm, okay
[20:01] <niemeyer> kirkland: Or, that's the plan, anyway.. config settings is one of the things being worked on right now (and for the past couple of months)
[20:01] <kim0> aren't formulas just using `pwgen 10 1` to create passwords
[20:01] <niemeyer> kim0: Slightly different case
[20:01] <kim0> ah the user needs to be informed of it
[20:01] <niemeyer> kim0: They could _also_ use that system, actually
[20:02] <niemeyer> kim0: But the opposite isn't true: the mechanism we're using for inter-formula communication won't help kirkland
[20:02] <kirkland> niemeyer: https://ec2-184-73-37-114.compute-1.amazonaws.com/
[20:02] <kirkland> niemeyer: that's what I'm working on
[20:02] <kirkland> niemeyer: a formula for setting up ajaxterm
[20:03] <kirkland> niemeyer: for that login to be useful, though, i need the user to have a username/password
[20:03] <niemeyer> kirkland: hah! Sweet!
[20:03] <niemeyer> kirkland: Right, makes sense
[20:03] <kirkland> niemeyer: as unfortunately, ajaxterm does not support ssh keys :-(
[20:03] <kim0> yummy
[20:03] <kirkland> niemeyer: so there has to be a user/pass in there
[20:03] <niemeyer> kirkland: That config idea we mentioned above is perfect for thta
[20:03] <niemeyer> that
[20:03] <kirkland> niemeyer: perfect
[20:04] <kirkland> niemeyer: the config settings you mean?
[20:04] <niemeyer> kirkland: The user will be able to say something like ensemble get ajaxterm password
[20:04] <niemeyer> kirkland: Or similar
[20:04] <kirkland> niemeyer: okay
[20:05] <niemeyer> kirkland: Isn't ajaxterm just using the real machine users?
[20:06] <kirkland> niemeyer: yes
[20:06] <niemeyer> kirkland: If so, you can recommend people to ssh into the machine in the interim, for sorting the password
[20:06] <niemeyer> kirkland: This works: ensemble ssh $SERVICE/$N
[20:07] <niemeyer> kirkland: and also: ensemble ssh $N (where $N is a machine number)
[20:07] <kirkland> niemeyer: cool, will do
[20:24]  * koolhead17 just executed wordpress formula  :)
[20:24]  * koolhead17 is happy
[20:34]  * niemeyer high-fives koolhead17
[20:38] <koolhead17> niemeyer: :D hehe
[20:45] <niemeyer> jimbaker, bcsaller, hazmat: Do you have any summary for an Ensemble talk that I could reuse?
[20:45] <hazmat> niemeyer, i think i sent around my pycon lightning talk notes
[20:46] <bcsaller> gustavo: I have to prep slides for my upcoming talk, but I don't have anything yet
[20:46] <niemeyer> hazmat: I mean a summary for a talk to send the event organizers
[20:46] <bcsaller> gustavo: are you looking for prose or slides?
[20:46] <hazmat> niemeyer, ah.. hmm
[20:46] <niemeyer> bcsaller: Just the usual blurb for the schedule
[20:47] <niemeyer> I can easily come up with something, but laziness made me check if you already had one readily available.
[20:47] <jimbaker> niemeyer, not yet
[20:47] <bcsaller> sorry, I don't
[20:47] <niemeyer> Cool, no worries
[20:50] <hazmat> niemeyer, this is my unsuccessful submission for plumbers - https://pastebin.canonical.com/48481/
[20:52] <niemeyer> hazmat: Ah, sweet, thanks!
[21:20] <_mup_> ensemble/expose-provision-service-hierarchy r292 committed by jim.baker@canonical.com
[21:20] <_mup_> Removed watch_exposed_flags changes from this branch to move upstream
[23:28] <kirkland> niemeyer: relation-set ip="$ipaddr" port=443 hostname=`hostname -s`
[23:28] <kirkland> niemeyer: is that what I need for this byobu/ajaxterm formula?
[23:28] <kirkland> niemeyer: since the service runs on 443?
[23:28] <niemeyer> kirkland: Maybe.. a URL might do as well
[23:29] <niemeyer> kirkland: set url=...
[23:29] <SpamapS> kirkland: lol, hahaha I should have read here before answering you over there. :)
[23:29] <kirkland> SpamapS: ;-)
[23:29] <kirkland> SpamapS: i moved those conversations here, instead of bothering you there ;-)
[23:30] <SpamapS> kirkland: nice. I've been thinking about building an rsync formula to solve the "shared upload" problem until I can wrap my head around gluster/nfs
[23:30] <SpamapS> basically just a formula that provides a box which will rsync a dir from all related service units to all related service units
[23:30] <kirkland> SpamapS: i'm going to follow byobu-web with a byobu-classroom, that depends on byobu-web
[23:30] <kirkland> SpamapS: which sets up the one-writer, many readers classroom mode of byobu (previously called "screenbin")
[23:31] <SpamapS> kirkland: isn't that all still just one machine?
[23:50] <SpamapS> niemeyer: btw, how is config settings doing? Any branch we can help test out or anything? At this point it is being cited as the solution for a lot of stuff. ;)
[23:50] <niemeyer> SpamapS: bcsaller would be the right person to report on this
[23:52] <SpamapS> bcsaller: what it is brother? Got the 411? :)
[23:52] <kirkland> niemeyer: and is this correct?
[23:52] <kirkland> provides:
[23:52] <kirkland>   website:
[23:52] <kirkland>     interface: https
[23:53] <bcsaller> SpamapS: I have a branch thats working, I can effect changes on working ec2 deployments with it, however there are some issues with the testing of the branch since I tried merging it with trunk. Partly but not completely resolved. Tests somewhere are bleeding setup and changing the outcome depending on execution order
[23:53] <niemeyer> kirkland: The interface is actually an Ensemble level interface
[23:53] <niemeyer> kirkland: Rather than the protocol itself
[23:53] <bcsaller> SpamapS: Kapil did had some comments on the branch that needed some changes as well and I am working on those now, but its pretty close 
[23:54] <SpamapS> bcsaller: sounds fun. :-P alright, well color me interested in testing ASAP. Lots of settings to add. :)
[23:54] <niemeyer> kirkland: We still have to settle on a good set of those
[23:54] <bcsaller> SpamapS: great, thanks
[23:54] <kirkland> niemeyer: okay;  so what should it be in my case, right now?
[23:54] <niemeyer> kirkland: I think "url" would be a good name, for instance, for something that provides _only_ a URL
[23:55] <SpamapS> website has been the standard name for the relation which provides you with "how to access me" .. though right now it just spits back IP ... eventually I think it should provide status url(s) and possibly the desired canonical hostname.
[23:55] <niemeyer> kirkland: Even though, perhaps it's a good idea to make that more specific
[23:56] <niemeyer> kirkland: Not sure.. go with "http" for the moment I guess
[23:56] <kirkland> niemeyer: k
[23:56] <niemeyer> kirkland: (not https.. anyone handling http will most likely handle both)
[23:57] <kirkland> niemeyer: k
[23:57] <kirkland> SpamapS: for your IP=...
[23:57] <kirkland> SpamapS: what about: ip=$(wget -q -O- http://169.254.169.254/latest/meta-data/public-ipv4)
[23:57] <niemeyer> SpamapS: That's an interesting idea
[23:57] <niemeyer> kirkland, SpamapS: +1 on website
[23:57] <kirkland> niemeyer: okay, so change http to "website" then
[23:58] <SpamapS> niemeyer: err.. https and http are vastly different protocols. ;)
[23:58] <SpamapS> kirkland: won't work on lxc
[23:58] <kirkland> SpamapS: hrm
[23:59] <SpamapS> kirkland: and thats not necessarily the IP we want.. since a load balancer should use the internal IP
[23:59] <SpamapS> kirkland: been bouncing around the idea in my head of a provider-agnostic "machine info" script that would be part of ensemble.
[23:59] <kirkland> SpamapS: facter ;-)