[04:42] <TheChistoso|2> is there an environment variable i can use to point to environments.yaml?
[04:42] <TheChistoso|2> and for that matter, config.yaml?
[09:20] <_mup_> Bug #1175958 was filed: New peer relations not created when upgrading charms <amd64> <apport-bug> <raring> <juju:New> <juju (Ubuntu):New> <https://launchpad.net/bugs/1175958>
[11:27] <sidnei> jamespage: hi! replied to your review, the install hook actually adds the ppa and installs the dependencies.
[11:37] <jamespage> sidnei, I'd missed that
[11:38] <sidnei> yeah, might need to adjust the readme to mention so
[11:38] <jamespage> sidnei, how do you guarantee that the versions of tools you are pulling from the ppa are the same ones that the charm was tested with
[11:38] <jamespage> ?
[11:39] <sidnei> thats tricky. and also the fact that a ppa is not frozen, so an update to the ppa could break the charm.
[11:40] <sidnei> wedgwood_away created a stable and dev ppa, im using the stable one, which supposedly would only get updates if it doesn't break the charm.
[11:43] <jamespage> sidnei, yeah - its tricky - we have exactly the same issue with the openstack charms
[11:43] <jamespage> sidnei, which is why we actually embed the helper in the charm
[11:44] <jamespage> (s)
[11:44] <sidnei> 'static linking' ftw. ;)
[11:44] <jamespage> that way there is no requirement for a PPA; and DC's where there is limited outbound network access don't break
[11:44] <jamespage> sidnei, man - its the way to fly!
[11:44] <jamespage> *(apparently)
[11:47] <sidnei> jamespage: so what do you suggest? include a copy of the deps?
[11:48] <jamespage> sidnei, yep
[11:48]  * jamespage ducks behind his desk to avoid being shot
[11:48] <jamespage> sidnei, its a compromise but its the only one that meets the objective that a charm should be installable by default from the main ubuntu archive
[11:49] <sidnei> jamespage: the other one being getting charmsupport into the archive, which is a pain on itself.
[11:50] <jamespage> sidnei, well its not that bad
[11:50] <jamespage> we could package and drop it into saucy and then backport back through to precise
[11:50] <jamespage> backports is an official part of the archive so does get mirrored
[11:50] <sidnei> and then you get into SRUs :)
[11:50] <sidnei> ah, backports
[11:51] <jamespage> sidnei, yeah - which is why actually taking the static link method is good
[11:51] <jamespage> sidnei, I know it goes against all our ingrained distro knowledge
[11:55] <sidnei> ok, i'll go that route then. will make it so that 'make sourcedeps' can be used to automate the update of those copies, so that you can just run that and commit
[12:00] <jamespage> sidnei, makes sense to me
[12:50] <hazmat> jamespage, do you need a fix for that peer rel issue on upgrade?
[12:51] <jamespage> hazmat, well other than myself I'm guessing that no-one has hit that issue right now
[12:51] <jamespage> as the HA branches are still in final testing
[12:51] <jamespage> but I think that any juju based openstack deploys out there right now would be blocked on taking advantage of these features as a result
[12:51] <jamespage> hazmat, is it a huge amount of work?
[12:52] <hazmat> jamespage, not really.. http://paste.ubuntu.com/5629022/
[12:52] <hazmat> that's the diff for the fix
[12:52] <jamespage> I see - not huge then
[12:52] <jamespage> hazmat, in that case - yes please!
[13:12] <jcastro> marcoceppi: ping me when you're around
[13:16] <jamespage> jcastro, folks are already trying out the OpenStackHA stuff - had a direct email this morning
[13:22] <jcastro> hah yeah
[13:22] <jcastro> someone asked about it on AU and I put a link there too
[13:39] <marcoceppi> jcastro: ping
[13:42] <jcastro> marcoceppi: pong
[13:43] <jcastro> marcoceppi: let me fire up reminders to our social networks
[13:43] <jcastro> and then G+?
[13:43] <jcastro> shoot for the top of the hour?
[13:43] <marcoceppi> Sounds good, I'll be here
[13:45] <marcoceppi> jcastro: ack
[14:02] <jcastro> marcoceppi: G+ing you
[14:02] <jcastro> hey
[14:03] <jcastro> which account should I be inviting to G+ you?
[14:29] <mthaddon> does anyone know if there's any issue with running goJuju and pyJuju on the same machine (currently using pyJuju on Openstack, want to test using goJuju without disrupting existing envs)?
[14:33] <mgz> mthaddon: you may want to use seperate env files, otherwise no
[14:33] <mgz> you do want to be on raring for the shiny packages
[14:34] <mthaddon> mgz: this is a production env that's precise - are there backports of shiny, or if not, will there be at some stage?
[14:34] <mgz> you can deploy on precise... you just want to use the clients from the latest distro packages
[14:35] <mthaddon> mgz: do you mean use the raring packages?
[14:35] <mgz> yeah
[14:37] <mthaddon> mgz: any ideas how easy that'd be to backport to precise?
[14:37] <mgz> pretty simple, the point is you want both juju 0.7 and juju-core 1.10.0 with the packaging changes to make them co-installable
[14:38] <mgz> and the ppas don't currently have that
[14:40] <mthaddon> mgz: ah, I see, so we need juju 0.7 to allow it to be co-installable with juju-core 1.10.0 - gotcha
[14:41] <marcoceppi> Does juju-core work with local provider?
[14:42] <mgz> marcoceppi: no local provider is implemented yet
[14:42] <marcoceppi> mgz: just checking, thanks :)
[14:42] <mthaddon> mgz: how about openstack provider? (sorry, I should know this)
[14:43] <mgz> mthaddon: there is an openstack and a maas provider now, though some things are still not quite as polished as in 0.7 yet
[14:44] <mthaddon> mgz: understood - so if we have 0.6 installed, but are using a custom branch in juju-origin for a specific env, all local commands (juju status, etc) are using 0.6, but the env itself is deployed with that custom branch, correct?
[14:45] <mgz> mthaddon: that's right
[14:46] <mgz> for co-installbility, only the local package matters, as environments aren't compatible anyway
[14:46] <mthaddon> mgz: how do you mean, environments aren't compatible?
[14:47] <sidnei> the environments.yaml have different keys and both complain about unknown keys iirc
[14:48] <mthaddon> sidnei: they require different formats for environments.yaml?
[14:48] <sidnei> not 'formats' just some keys within environments.yaml are different
[14:48] <mthaddon> right
[14:49] <mgz> and you can't bootstrap with pyjuju, then do status with gojuju
[14:49] <mthaddon> was just about to ask that :)
[14:50] <mthaddon> how safe is an upgrade of pyJuju from 0.6 to 0.7?
[14:50] <mgz> safe.
[14:51] <mgz> an existing environment will be using a set version on the machines anyway, so all you're changing is your local client
[14:53] <mthaddon> cool, thx. And last question (I think). Can we upgrade juju in a running env, or is that a goJuju feature only?
[14:57] <mgz> I think that's gojuju only, hazmat might correct me.
[15:01] <hazmat> go juju only on that one
[15:02] <hazmat> mthaddon, probably next month, but there will be some tooling in place to enable impl migration
[15:02] <hazmat> mthaddon, its a topic for next week
[15:04] <jcastro> mthaddon: backports of everything to precise are on the team's todo
[15:04] <jcastro> Daviey was telling me a "few weeks"
[15:04] <jcastro> but I don't know how accurate that is lately
[15:05] <Daviey> hmm
[15:05] <Daviey> That is still the plan.
[15:06] <Daviey> it will go in via backports tho
[15:07] <mthaddon> thx
[15:08] <marcoceppi> hazmat: did the local provider fix for juju make it in to the backports?
[15:08] <marcoceppi> (on raring)
[15:14] <hazmat> marcoceppi, i believe it did.. mgz?
[15:15] <hazmat> marcoceppi, i added it to trunk since it wasn't there, but the 0.7 for raring was marked fixed
[15:15] <marcoceppi> hazmat: cool, I'll give it a shot. I saw it was fixed commited but the ppa appears to be behind what's in backports
[15:16] <hazmat> the trunk ppa is behind backports?
[15:16] <hazmat> odd
[15:16] <marcoceppi> ppa:juju/pkgs is still the pyjuju ppa, correct?
[15:35] <Youssefk> Hello, does anyone of you guys know if the reboot problem in juju has been fixed or not?
[15:36] <Youssefk> Hello, does anyone of you guys know if the reboot problem in juju has been fixed or not?
[15:38] <Youssefk> Hello, does anyone of you guys know if the reboot problem in juju has been fixed or not?
[15:46] <utlemming> macroceppi: is there an issue with the mediawiki? When deployed in the GUI, setting the user/password for Admin makes it go straight to config error
[15:47] <hazmat> marcoceppi, yes
[16:09] <AskUbuntu> Can I use Juju/MAAS on VPSes? | http://askubuntu.com/q/290122
[16:17] <jcastro> 45 minute warning on Juju Charm School!
[16:33] <jcastro> 30 minutes until Charm School!
[16:37] <mgz> jcastro: is there a link already, or should I look again on the hour?
[16:38] <jcastro> mgz: ubuntuonair.com for the stream
[16:38] <jcastro> I will post the hangout URL here if you want to join in the actual session
[16:38] <jcastro> but this one is more for new users
[16:38] <jcastro> getting started installing and configuring, etc.
[16:38] <jcastro> since we're starting anew
[16:39] <mgz> right, I think it's worth seeing what issues come up there, so was going to listen in
[16:56] <jcastro> he's having some problems with instances coming up on AWS
[16:57] <marcoceppi> mgz: http://paste.ubuntu.com/5629682/
[16:57] <marcoceppi> Deployed two services, but they're not launching instances in AWS
[17:04] <saras> so the ubuntu on air should be starting soon right
[17:04] <marcoceppi> saras: it's already started
[17:05] <saras> hum
[17:05] <saras> is that why their big white spot on ubuntu on air page
[17:05] <ahasenack> marcoceppi: where, ubuntuonair.com? I don't see the viedeo, just a white area where the video should be
[17:05] <marcoceppi> saras: not sure, bad embed code maybe; Here's the stream:" http://www.youtube.com/watch?v=9h5hgfnZcBQ
[17:05] <TheChistoso> if i lose my juju bootstrap node, can juju still function?
[17:05] <ahasenack> marcoceppi: thanks
[17:07] <TheChistoso> and if a node dies (power supply dies), can i still remove (terminate) it?
[17:08] <saras> can i use salt stack with juju
[17:09] <sarnold> saras: you may be able to create a salt subordinate charm, so that salt is available on all your nodes
[17:09] <TheChistoso> and simply rebooting a node shouldn't cause anything to necessarily change w/ juju, correct? that is, after services have been deployed to a node, rebooting the machine would have no ill effect on juju (the availability and services that node was running is a different matter)?
[17:12] <saras> what we can't talk about stack i will be using nginx | node.js | mondb
[17:12] <marcoceppi> On air page fixed
[17:13] <saras> at least their not using webex agrr 10gen
[17:15] <saras> metoer.js on node.js that know to talk to mongo
[17:15] <ahasenack> jcastro: you should mute, or else the focus switches to your screen everytime you make a noise :)
[17:16] <ahasenack> unless you want to speak, of course
[17:16] <jcastro> noted
[17:16] <jcastro> we're taking questions in here if you have questions
[17:16] <FunnyLookinHat> And remove the unity bar from the left...  works against the full-screen view because it's not snapping :)
[17:16] <brejoc> shell was better ;)
[17:16] <jcastro> he's going to switch to the browsers a bunch
[17:17] <Ben___> did I miss the juju charm school?
[17:17] <jcastro> http://youtu.be/9h5hgfnZcBQ
[17:17] <jcastro> nope
[17:17] <jcastro> in progress right now!
[17:17] <FunnyLookinHat> Awesome - thx :)
[17:17] <jcastro> feel free to ask qurestions
[17:17] <FunnyLookinHat> weird...
[17:17] <Ben___> Is there a hangout link?
[17:17] <jcastro> https://plus.google.com/hangouts/_/5ce263e3acf52d2a8604ad61d7fa1b7ef01377e7?authuser=0&hl=en
[17:18] <ahayzen> jcastro, hit F11 for fullscreen terminal?
[17:18] <jcastro> marcoceppi: ^^
[17:18] <Ben___> thanks!
[17:22] <mattrae> jcastro: are you using the ppa version of juju?
[17:23] <saras> should I write a charm for metoer.js
[17:23] <Guest76074> I wrote a juju charm for a service called boundary, wondering about the process to submit it for inclusion.  It is a paid SaaS.
[17:23] <mattrae> jcastro: cool thx
[17:24] <mattrae> neat
[17:25] <jcastro> aaronj____: https://juju.ubuntu.com/docs/charm-store.html
[17:25] <jcastro> if you prefer to have it on github that's fine too, just plop the link in the bug report
[17:25] <TheChistoso> i just realized you guys are doing a web cast :D -- did you happen to catch my questions from earlier?
[17:25] <aaronj____> jcastro: thank you
[17:25] <saras> should I write a charm for metoer.js? jcastro:
[17:25] <jcastro> TheChistoso: nope, feel free to AskUbuntu
[17:26] <jcastro> saras: oh yes, absolutely!
[17:26] <jcastro> that would be nice
[17:26] <jcastro> TheChistoso: I meant feel free to ask again
[17:27] <TheChistoso> if i lose my juju bootstrap node, can juju still function? how would i re-bootstrap?
[17:27] <aaronj____> is this console where juju bootstrap was run an EC2 instance or a local machine.
[17:27] <paraglade> jcastro: our company has made a big investment in puppet and I know it is possible to have juju and puppet work hand in hand.  I would love to see a charm demonstration of this.
[17:29] <aaronj____> paraglade: +1
[17:30] <paraglade> jcastro: can one bootstrap node handle many environments?
[17:33] <TheChistoso> i'm looking at using juju + maas to deploy openstack. how stable are the openstack charms for use with raring?
[17:34] <saras> salt
[17:35] <FunnyLookinHat> TheChistoso, I stepped out for a sec- did someone answer your question about losing the juju bootstrap node ?
[17:35] <TheChistoso> FunnyLookinHat: yes
[17:35] <FunnyLookinHat> TheChistoso, the [short] answer?
[17:35] <TheChistoso> right now, it's a known issue. they're working on HA for bootstrap nodes.
[17:35] <saras> lxc
[17:36] <FunnyLookinHat> jcastro, Timeline on colocation ?  I know it's dependent upon the Go rewrite...
[17:36] <FunnyLookinHat> thx :)
[17:36] <FunnyLookinHat> TheChistoso, Ah ok good to know
[17:37] <TheChistoso> jcastro: i had a question (up above)
[17:37] <jcastro> TheChistoso: yep, I'll hit it up next
[17:37] <TheChistoso> jcastro: (thanks! :D)
[17:38] <TheChistoso> jcastro: is there a provisional release date for 2.0?
[17:38] <jcastro> over the next month or so
[17:39] <TheChistoso> i'm working on a PoC, is there a way to easily use an RC/beta version before the official release w/o having to setup an entire development environment?
[17:41] <TheChistoso> i don't have 28+ nodes (the required # for use w/ juju) to deploy openstack
[17:42] <TheChistoso> (openstack HA that is)
[17:42] <TheChistoso> jcastro: tyvm
[17:42] <jcastro> oh, I can help you there too, I'll get to that one next
[17:42] <jcastro> https://wiki.ubuntu.com/ServerTeam/OpenStackHA
[17:42] <jcastro> for those interested in HA Openstack
[17:42] <paraglade> jcastro: in the pyjuju version what is the "open-tunnel" command used for?
[17:43] <saras> jcastro: how can I till what the verison of sevices is in a charm as their are 3 mongodb charms
[17:43] <TheChistoso> i've been following that -- i do intend to use ceph -- that page mentions using precise w/ the quantal kernel. how do i set that up? or is that a maas question?
[17:43] <jcastro> http://jujucharms.com/charms/precise/mongodb
[17:44] <jcastro> saras: I'll explain why there are three in a minute
[17:44] <saras> thanks
[17:45] <ahasenack> jcastro: highlight that it's just not a service up and running, but up and running in a scalable fashion
[17:45] <ahasenack> jcastro: otherwise people could reach the same state with just a bunch of customized amis
[17:46] <jcastro> sure
[17:48] <arosales> ahasenack, good point. The scaling embedded with relations is a very nice feature.
[17:49] <ahasenack> or even just using fabric
[17:49] <TheChistoso> jcastro: what if i want to colocate multiple web apps. is that possible?
[17:50] <FunnyLookinHat> Now every DDoS his haproxy to test!  :D
[17:51] <FunnyLookinHat> TheChistoso, He talked about that - not currently possible... planned as a post 2.0 feature
[17:52] <TheChistoso> okay, thanks funny!
[17:53] <FunnyLookinHat> fo sho :)
[17:54] <TheChistoso> jcastro: the openstack HA page mentions using precise w/ the quantal kernel in order to get kernel updates for ceph. how do i set that up? or is that a maas question?
[17:54] <jcastro> I'm still looking up how to answer that.:)
[17:54] <jcastro> I think that's more of a cloud image question than either juju or maas
[17:54] <TheChistoso> jcastro: (thanks!)
[17:55] <TheChistoso> jcastro: that would make sense
[17:55] <FunnyLookinHat> Given that errors are produced occasionally by bugs, is there a way to ensure that a service will always work by specifying a particular charm version ( or set of versions ) ?
[17:55] <mattrae> if services need to be related in a particular order, is there a way to tell a juju to automate a complecated rollout?
[17:57] <TheChistoso> FunnyLookinHat: how i manage that is by checking out the source and using locally-available charms
[17:57] <mattrae> cool
[17:58] <FunnyLookinHat> yeah - having the source local is... not as clean. :)  But being able to define the version w/ the charm store is a solid fix
[17:59] <FunnyLookinHat> Ah ok - so it'll cache the charm you've used
[18:00] <paraglade> jcastro: I thought that was what the control bucket was for.  caching the charms
[18:00] <jcastro> yeah
[18:00] <jcastro> I think it dumps other stuff in there too
[18:01] <TheChistoso> jcastro: hey so when are you gonna support windows in AWS? ;D
[18:01] <paraglade> before you bail you should so what happens if you kill an instance in the AWS console. :)
[18:02] <jcastro> good point
[18:02] <paraglade> s/so/show/
[18:03] <TheChistoso> lol j/k actually :D
[18:04] <paraglade> vCD support would be awesome also.
[18:04] <TheChistoso> does anybody actually do that? ubuntu on azure?
[18:05]  * SpamapS hums the song from The Little Mermaid "Poor Unfortunate Souls"
[18:05] <sarnold> TheChistoso: doubtless the azure images weren't done just for giggles :)
[18:06] <arosales> to help clearify its Juju managing Ubuntu workloads in Azure, in the same way Juju manages Ubuntu workloads in AWS
[18:06] <jcastro> yeah did I say Juju managing windows at first?
[18:07] <jcastro> let me be clear that "juju deploy sqlserver" would be awesome, it's not anything we're working on.
[18:07] <arosales> jcastro, I think your finaly statement got it correct, just wanted to throw out some additional info.
[18:07] <TheChistoso> if i deploy mysql w/ juju on AWS -- is it using EBS to back its data? ephemeral mysql storage wouldn't seem to be a good idea
[18:08] <aaronj___> that would be cool to see.
[18:09] <TheChistoso> that would be fantastic
[18:09] <saras> jcastro: could someone add rdbms interface to charm
[18:11] <TheChistoso> jcastro: you play guitar? :D you pretty good?
[18:11] <aaronj___> looking forward to it!
[18:11] <FunnyLookinHat> jcastro, marcoceppi - Great presentation! thanks guys :D
[18:11] <TheChistoso> yes -- thanks guys!
[18:12] <mgz> thanks jcastro :)
[18:12] <mattrae> jcastro: is there a schedule for this training?
[18:12] <arosales> jcastro, marcoceppi thanks
[18:12] <aaronj___> thank you
[18:12] <jcastro> thanks guys!
[18:12] <ahayzen> thanks guys :)
[18:12] <saras> bye
[18:13] <mattrae> thanks :)
[18:13] <jcastro> http://juju.ubuntu.com/survey
[18:13] <paraglade> jcastro: this would be huge to get this working: https://code.launchpad.net/~franciscosouza/juju/juju-vpc/+merge/134169
[18:13] <jcastro> ^^ feel free to pass that one
[18:13] <aaronj___> jcastro: you mentioned a robbie williamson video ... regarding openstack iirc.
[18:13] <jcastro> yeh let me find you that link
[18:14] <arosales> jcastro, marcoceppi for the "debug" charm you may want to note the juju version and environment details for reproduction next charm school.
[18:14] <marcoceppi> saras: Do you mean like a connection to Amazon's RDS service?
[18:14] <jcastro> https://www.openstack.org/summit/portland-2013/session-videos/presentation/openstack-in-production-the-good-the-bad-and-the-ugly
[18:14] <jcastro> aaronj___: ^^^
[18:14] <jcastro> that's how we roll with openstack in prod
[18:15] <marcoceppi> For those wondering which bug I was talking about, https://bugs.launchpad.net/juju-core/+bug/1172895 it's this one and it's been committed already
[18:15] <_mup_> Bug #1172895: relation-list incompatibility with pyjuju: -r <juju-core:Fix Committed by fwereade> <https://launchpad.net/bugs/1172895>
[18:15] <saras> rdbms i think
[18:15] <mattrae> jcastro: is the plan to have charm school every other week at the same time?
[18:15] <saras> nice
[18:16] <jcastro> mattrae: yep
[18:16] <mattrae> jcastro: cool thx
[18:16] <jcastro> I'm going to go make sure it's repeating on the calendar
[18:16] <marcoceppi> saras: Well, rdbms is a general name for data storage solutions. So MySQL, Postgresql, Oracle, SQLite, etc are all RDBMS systems
[18:16] <mattrae> ah cool
[18:16] <jcastro> I will also send mail reminders to the list, etc.
[18:18] <saras> DRBD
[18:18] <saras> sorry to many letters running around lose in my head
[18:19] <marcoceppi> saras: ah, the block device mirroring tool?
[18:19] <saras> yes
[18:19] <marcoceppi> So, we don't have a charm for that, I'm not familiar about it
[18:19] <marcoceppi> it's implementation but it looks like it would be something the provider implements?
[18:20] <jcastro> I think that would be a sub
[18:20] <jcastro> you would plop it on a service
[18:20] <jcastro> and it would just start replicating
[18:20] <marcoceppi> saras: we do have similar charms like Ceph that can do large continuous filesystems that are HA
[18:21] <saras> ceph can speak drbd
[18:21] <saras> so interface that you could talk to ceph over
[18:21] <jcastro> the ceph guy is usually in this room
[18:21] <jcastro> I think the question would be how does the ceph charm use drdb
[18:22] <marcoceppi> or if it uses it at all. I don't see an interface for it, but if you knew ceph and knew how drdb would work you could implement it in to the ceph charm and have it reviewed/merged in to the official one if you had other services that needed to talk to ceph via drdb
[18:22]  * marcoceppi goes to read more about it
[18:23] <jcastro> scuttlemonkey knows, he's not here right now though
[18:23] <jcastro> chttp://ceph.com/dev-notes/deploying-ceph-with-juju/
[18:23] <jcastro> as an FYI
[18:29] <saras> jcastro: so till me if I nut or should i use the node.js as starting point for the meteor.js
[18:29] <saras> then add the npm command to install is and setup the mongodb inface
[18:30] <saras> interface
[18:30] <TheChistoso> so i'd actually like to deploy ceph alongside nova compute nodes. somewhat like piston cloud's solution (if you're familiar w/ that)
[18:30] <jcastro> saras: yeah for sure
[18:30] <TheChistoso> is that possible?
[18:31] <jcastro> man, meteor looks pretty awesome
[18:31] <marcoceppi> saras: actually, meteror.js might "just work" with the nodejs charm
[18:32] <TheChistoso> how i would structure things is putting keystone/horizon/rabbitmq/mysql in a few HA controller nodes and then putting ceph, nova compute, quantum, swift/glance (swift would just be ceph) on each compute node using locally attached storage
[18:32] <TheChistoso> how easy would it be to configure the charms to enable that scenario?
[18:32] <marcoceppi> saras: http://jujucharms.com/charms/precise/node-app it's designed to "just work" for deploying node.js apps that use mongodb backends
[18:33] <saras> marcoceppi: i see that
[18:34] <saras> marcoceppi: thanks
[18:34] <marcoceppi> saras: though it might be good to just make it it's own charm
[18:34] <jcastro> https://install.meteor.com/
[18:34] <marcoceppi> since it's more of a standalone service than it is "a node app"
[18:34] <jcastro> has their install stuff
[18:34] <marcoceppi> that's entirely up to you
[18:34] <jcastro> https://bugs.launchpad.net/charms/+bug/1035003
[18:34] <_mup_> Bug #1035003: charm needed: meteor <Juju Charms Collection:New> <https://launchpad.net/bugs/1035003>
[18:34] <jcastro> So I think it'd be like a node or rails charm
[18:34] <jcastro> just for meteor apps though
[18:35] <sarnold> why on earth would they only support amd64 and i386??
[18:35] <sarnold> I thought it was javascript..
[18:35] <marcoceppi> saras: jcastro looking at it, yeah I agree, I think this is probably  it's own charm. It would be awesome as a charm
[18:35] <jcastro> yeah so a little yaml file pointing to my meteor app on github
[18:35] <jcastro> and deploy
[18:35] <marcoceppi> sarnold: some node.js modules are compiled C++, like native php extensions
[18:36] <saras> think i would use npm
[18:36] <saras> not githud
[18:36] <saras> not github
[18:36] <jcastro> yeah so from experience on the node charm
[18:36] <jcastro> people like having a switch to grab from $wherever
[18:36] <jcastro> at first we used a PPA
[18:36] <sarnold> marcoceppi: ah, so you'd have to compile yourself for othe arches, and the hilarious curl url | sh thing is just for the Big Architectures... okay. :)
[18:36] <jcastro> then some people are like, I want to use the one in the distro not some PPA, etc.
[18:37] <jcastro> so having a switch the user can deploy, defaulting to the safe distro version seems to be the defacto best practice
[18:37] <jcastro> saras: do you work on meteor or just use it?
[18:38] <saras> jcastro: no going to be playing with alot soon
[18:38] <jcastro> it looks neat
[18:38] <saras> learning mongo now
[18:39] <jcastro> on the next screencasts we will talk about platform charms
[18:39] <saras> i love idea of use mongo on server and client
[18:39] <jcastro> where instead of charming up every single app, we have a stack of platform charms people can use to deploy apps on those frameworks
[18:42] <saras> jcastro: thanks
[18:42] <jcastro> <-- off to lunch
[19:23] <ColinR> Howdy. I've been looking at implementing juju where I work and had some questions I was wondering if anyone in here could answer. First, from what I've read, juju support for multiple "bootstrap" nodes (don't know the proper term for them), is on it's way. Is there anyway to join in beta testing that feature/possibly contribute, and if so, who should I talk to?
[19:23] <bkerensa> juju generate-config -w
[19:24] <bkerensa> does nothing?
[19:25] <ColinR> Was that directed towards me bkerensa?
[19:25] <bkerensa> no
[19:25] <bkerensa> m_3 or jcastro
[19:25] <bkerensa> :)
[19:26] <ColinR> no worries, just joined the channel, so I wasn't sure.
[19:26] <bkerensa> ahh k
[19:26] <bkerensa> or marcoceppi even ^
[19:28] <ColinR> My second question is regarding chef. Is there anyone here who has experience working with chef and juju? Currently, we use Noah (https://github.com/lusis/Noah), a Zookeeper alternative, for orchestrating node interaction and chef for node configuration, but juju looks very promising if we can tie in our existing chef cookbooks.
[19:38] <jcastro> ColinR: for multiple bootstraps, we call that Juju HA
[19:38] <jcastro> For that you'd want to check out the dev mailing list: juju-dev@lists.ubuntu.com
[19:38] <jcastro> there's all so #juju-dev on freenode fore core stuff
[19:38] <jcastro> for your second question
[19:39] <jcastro> there are some people using it with chef solo
[19:39] <jcastro> as far as I know so far though, no one's really published a "here's how we reused our chef stuff with juju" yet.
[19:42] <bkerensa> jcastro: your juju doc for rackspace is no works :P
[19:42] <bkerensa> https://juju.ubuntu.com/get-started/rackspace/
[19:42] <bkerensa> what [] brackets?
[19:42] <jcastro> read the top thing
[19:42] <bkerensa> also the command you listed is depracated
[19:42] <jcastro> "This doesn't work and is really an inprogress placeholder while we figure out how to make it work, Rackspace Cloud is still in private beta so this answer is probably useless to most people. "
[19:42] <bkerensa> LOL
[19:43] <jcastro> they haven't rolled out the stuff we need to work on it
[19:43] <bkerensa> ah
[19:43] <jcastro> TLDR it's not really vanilla openstack
[19:43] <jcastro> they have committed to fix it
[19:43] <bkerensa> hmm
[19:43] <jcastro> they just haven't yet
[19:43] <jcastro> so one day, it will Just Work
[19:43] <jcastro> and from ODS I know it's pretty soon
[19:43] <bkerensa> :(
[19:43] <bkerensa> what to do with this $500 a month rackspace credit then?
[19:43] <bkerensa> pftt
[19:44] <jcastro> does it expire?
[19:44] <bkerensa> in two years
[19:44] <bkerensa> :D
[19:44] <bkerensa> but I was going to deploy a wordpress instance for testing my framework
[19:44] <bkerensa> and then put up a huge instance to do BOINC
[19:44] <jcastro> http://www.rackspace.com/blog/ramping-up-our-openstack-investment-involvement/
[19:44] <jcastro> towards the bottom
[19:45] <jcastro> We also realize that the Rackspace public cloud has a number of implementation details that are out of step with common practice in the OpenStack community. Most of these were due to either our early lockdown on features to enable our 2012 launch – or decisions designed to keep in line with our first generation Cloud Servers offering. While  we believe some variation in implementations will be inevitable, we
[19:45] <jcastro> do want to eliminate as many of these as possible to provide as much of a common OpenStack experience as we can. Along those lines, we are committed to take the following steps between now and the end of 2013. Hopefully, this will close most of the gaps that may currently exist:
[19:45] <bkerensa> ah
[19:46] <jcastro> someone had a boinc charm somewhere
[19:46] <jcastro> I don't think it ever made it into the store
[19:46] <jcastro> It'd be awesome though
[19:46] <jcastro> build your new cloud, break it in with boinc for like a week
[19:49] <hazmat> bkerensa, its not too hard to enable it re rack
[19:49] <bkerensa> ah
[19:49] <bkerensa> jcastro: also they are using the old ubuntu logo
[19:49] <bkerensa> ;p
[19:50] <bkerensa> send sladen after them
[19:50] <saras> their is guide to setup local juju for test and dev right
[19:59] <ColinR> Thanks for the answers, jcastro. Is juju-core still using Zookeeper? (I'm trying to dig through the source code now, but Go projects are a new experience for me)
[20:00] <jcastro> ColinR: mongodb
[20:00] <jcastro> saras: yeah, the only bummer right now is that the local provider only works with old juju and not new juju
[20:00] <jcastro> so you need to do an update-alternatives to switch the client back and forth
[20:01] <saras> hum
[20:01] <jcastro> it's not an ideal situation but the local provider for goju isn't finished yet. :-/
[20:01] <jcastro> but using pyju to use the local provider to write a charm should work just fine
[20:02] <saras> will just got apport to say hi by installing ppa verison
[20:02] <jcastro> since the version of core doesn't matter to the charm
[20:03] <saras> why "go" any way
[20:03] <jcastro> I think we have that on video, sec
[20:04] <jcastro> http://youtu.be/kKQLhGZVN4A
[20:05] <hazmat> ColinR, isn't noah single node and dead upstream?
[20:06] <hazmat> ColinR, re existing chef cookbooks, if you can get them to run solo mode you can have juju pass config into ohai/chef as relation data comes in
[20:07] <ColinR> yes. We've forked it and made some changes to support multiple nodes/newer redis versions, but it still needs some work
[20:08] <ColinR> Thanks, hazmat; looks like the big thing holding us back from jumping to juju is the HA support. I know it's marked high priority in launchpad, fingers crossed it won't be long.
[20:09] <hazmat> ColinR, i expect will be there this summer
[20:09] <hazmat> ColinR,  what are you using for chef ha now?
[20:09] <hazmat> s/we'll
[20:10] <TheChistoso|2> how i would structure things is putting keystone/horizon/rabbitmq/mysql in a few HA controller nodes and then putting ceph, nova compute, quantum, swift/glance (swift would just be ceph) on each compute node using locally attached storage -- how easy would it be to configure the charms to enable that scenario?
[20:10] <ColinR> Our setup doesn't rely on chef HA, though it'd be nice if it was. If we can ditch relying on chef-server and just use chef-solo (which with something like juju we could) even better.
[20:13] <hazmat> TheChistoso|2, i'm not sure if the ha charms support that.. ie. colocation.. adam_g ?
[20:14] <hazmat> ColinR, cool. incidentally zookeeper has an http api.
[20:15] <TheChistoso|2> i'd like to use a more up-to-date version of juju. the one i have (in raring) is 0.7. can/how do i upgrade? i'm using this w/ maas and so is there an issue w/ the cloud image used by maas b/c the latest juju version isn't in backports?
[20:15] <ColinR> yeah, we looked at zookeeper quite a bit, but liked the way noah handled watches more.
[20:16] <adam_g> hazmat, TheChistoso|2 the colocation of principle  services might work okay since they're deploying unrelated services to the same machines, but when you start to add HA, i could imagine the hacluster subordinates stomping on each others cluster configurations.
[20:16] <jcastro> TheChistoso|2: you can use the PPA
[20:16] <jcastro> https://juju.ubuntu.com/get-started/
[20:16] <adam_g> TBH, i've not tried collocating services to share the same underlying pacemaker/corosync cluster. it *might* work okay
[20:16] <TheChistoso|2> adam_g: is this a limitation of the way the charms configure openstack or an inherent openstack limitation? (i think the former)
[20:17] <jcastro> repo is the top line on the CLI instructions
[20:17] <serch> hello
[20:17] <jcastro> TheChistoso|2: we're working on backporting all that to raring and precise, it's just a WIP right now
[20:17] <TheChistoso|2> jcastro: PPA will work w/ machines allocated from maas?
[20:17] <adam_g> TheChistoso|2, like i said its not really an issue for the openstack services themselves (the principle services), but the underlying cluster infrastrucutre that gets setup as subordinates.
[20:18] <jcastro> TheChistoso|2: that's a good question, let me check it out
[20:18] <TheChistoso|2> adam_g: well i suppose it merits the question of -- does this setup i'm looking to do make sense?
[20:18] <serch> hello someone can explain how to release the last vercion of ubuntu from the official website
[20:19] <jcastro> TheChistoso|2: I think you would just use juju-origin:ppa in your environments yaml
[20:19] <TheChistoso|2> jcastro: i imagine it would require modifying the cloud image. i haven't done that, but i imagine it's possible.
[20:19] <jcastro> and maas will know what to do
[20:19] <adam_g> TheChistoso|2, collocating glance and keystone isn't an issue, as theres no overlap in  config files or service ports. when you add the hacluster subordinate, a corosync/pacemaker cluster gets setup underneath. there would be overlap there, between twoh acluster subordinates on the same machine
[20:19] <TheChistoso|2> jcastro: ah -- okay. i'll have to read up on the docs for that then. never done that before.
[20:20] <jcastro> http://maas.ubuntu.com/docs/quantal/juju-quick-start.html
[20:20] <jcastro> needs to be updated probably
[20:20] <ColinR> Is there a way to seed a bootstrap node's db with data from another bootstrap node?
[20:20] <TheChistoso|2> ColinR: i imagine once bootstrap HA is done, you wouldn't need to do that
[20:21] <adam_g> TheChistoso|2, i think it makes sense for the purposes of conserving hardware.  quantum-gateway might need to be hosted on a machine other than compute, since there is potential  OVS overlap
[20:22] <ColinR> TheChistoso|2, agreed. in the meantime though, I'll have a much easier time convincing the boss to use it if I can show we can recover our data from a db dump.
[20:23] <TheChistoso|2> adam_g: let me explain my scenario a bit more. i'm constructing a small private cloud (~100 machines right now, likely to grow to ~300) for testing purposes. our tests are both CPU and write intensive. so, we'd like to use ceph but need as much of the data locally available as possible.
[20:24] <TheChistoso|2> so we have a few core VM images (snapshots) that we replicate hundreds of times across compute nodes.
[20:25] <TheChistoso|2> we plan on using gridcentric's virtual memory streaming product actually to run the VMs
[20:26] <jcastro> oh, who was the guy saying they didn't have enough machines for openstack?
[20:26] <TheChistoso|2> lol for my PoC i don't have enough machines :D
[20:27] <jcastro> http://jujucharms.com/~virtual-maasers/precise/virtual-maas
[20:27] <TheChistoso|2> and i'd like to dedicate as many machines to the core problem (testing) as possible
[20:27] <TheChistoso|2> i have 16 machines i can use for the PoC
[20:27] <TheChistoso|2> lots of HA-capable OpenStack implementations don't require 28+ to work
[20:28] <TheChistoso|2> mirantis' fuel package can do it w/ 3-4 nodes for a small-medium size implementation
[20:28] <TheChistoso|2> rackspace private cloud as well
[20:28] <TheChistoso|2> piston cloud's solution has all openstack services running on all nodes
[20:30] <saras> juju no like me
[20:33] <TheChistoso|2> btw, is there an environment variable i can use to specify where the environments.yaml file is?
[20:33] <adam_g> hazmat, does current service colocation (jitsu deploy-to) work with multiple unit services? eg, deploy rabbitmq and mysql to machine 1. add-unit rabbitmq. where does rabbitmq/1 end up?
[20:34] <hazmat> adam_g, new machine
[20:34] <adam_g> TheChistoso|2, interesting.  does collocating ceph storage nodes with compute nodes equate to locally available storage, though?
[20:34] <TheChistoso|2> adam_g: not necessarily :D
[20:35] <adam_g> hazmat, so, would it be possible to add-unit rabbimq, add-unit mysql and have both second units on the same machine?
[20:35] <hazmat> adam_g, deploy-to and the integrated --force-machine in core.. only apply to the initial unit deployed.. when scaling up those services they go to new machines
[20:35] <hazmat> adam_g, thats been discussed as part of a deploy --with=service
[20:35] <TheChistoso|2> i'm actually rather new to ceph -- what i actually would like is to have certain snapshots automatically replicated to every compute node for fast VM creation
[20:36] <saras> any one will help me troubshoot why juju install in going wrong
[20:36] <hazmat> its not done though, but its where the force machine stuff is probably heading.. its likely spelled as a constraint
[20:36] <TheChistoso|2> saras: what are you seeing?
[20:36] <adam_g> TheChistoso|2, FWIW, when you boot an image/snapshot, its cached locally on the compute node the first time its booted. the second time, its booted from the local copy
[20:36] <hazmat> saras, what's the problem?
[20:37] <hazmat> and which version
[20:37] <TheChistoso|2> saras: i've recently been through this myself
[20:37] <TheChistoso|2> adam_g: in my case i'm going to have copies upon copies of the same VM -- not rebooting the same VM necessarily.
[20:38] <adam_g> TheChistoso|2, there was talk at openstack summit a couple weeks back of moving from machine/image snapshots being  entire copies of images to differential snapshots
[20:38] <TheChistoso|2> most our VMs run constantly, but when we update our core image, it needs to be replicated to all compute nodes and the VMs restarted from the new (snapshotted) image
[20:38] <TheChistoso|2> adam_g: gridcentric's VMS product effectively does that now. but w/ memory as well. so you can snapshot a running VM and then stream it to compute nodes.
[20:39] <saras> dpkg: error processing /var/cache/apt/archives/juju-core_1.10.0-1~1189~precise1_amd64.deb (--unpack):
[20:39] <saras>  trying to overwrite '/usr/bin/juju', which is also in package juju 0.5+bzr531-0ubuntu1.3
[20:39] <saras> dpkg-deb: error: subprocess paste was killed by signal (Broken pipe)
[20:39] <TheChistoso|2> my issue w/ that is when we update our image, we'll have a barrage of requests for the new snapshot all at once. a "VM snapshot storm" if you will....
[20:39] <jcastro> "dpkg -i --force-overwrite /var/cache/apt/archives/juju-core_1.10.0-1~1189~precise1_amd64.deb" should fix that, but please wait for someone to confirm that.
[20:42] <TheChistoso|2> for starters, though, i'll just get a basic openstack cloud up and running :D i've done it a couple of times using the mirantis fuel and then the mirantis fuel-web package and then w/ rackspace private cloud as well. but i really, really want to use juju for this. :D
[20:44] <saras> i think that worked thanks
[20:44] <saras> i think that worked
[20:45] <saras> that worked i think
[20:46] <TheChistoso|2> jcastro: a small recommendation for http://maas.ubuntu.com/docs/quantal/juju-quick-start.html -- i ran into this issue myself...please add a note that when configuring the "maas-server" entry, that you'll want to include the port number in the URI. I know that's fixed these days, but w/ juju that comes OOTB and following just the quick start, it can be an issue.
[20:47] <TheChistoso|2> (sorry -- that is, when configuring "maas-server" in environments.yaml)
[20:51] <saras> how do i till juju to use local
[20:52] <jcastro> TheChistoso|2: got it, I'll submit a fix
[20:52] <TheChistoso|2> saras: do you mean to use local charms?
[20:52] <saras> yes
[20:53] <TheChistoso|2> there's a couple different ways. i prefer to set the JUJU_REPOSITORY env var to the directory where they reside.
[20:53] <TheChistoso|2> then to deploy you'd say something like juju deploy local:<service>
[20:55] <TheChistoso|2> jcastro: adam_g: i have the issue that my hardware is not homogeneous. so i have a preference for which machine might get peeled off from maas for use in deploying services. actually what i have is a preference on machine characteristics and not the machine itself.
[20:57] <TheChistoso|2> i know i can instruct juju to use a specific machine. but i really don't care which machine -- i care which "class" of machine it is.
[20:57] <saras> saras@saras-Bonobo-Performance:~$ juju deploy local:precise/node-app-6
[20:57] <saras> error: environment has no access-key or secret-key
[20:58] <saras> why is this not liking me
[20:58] <TheChistoso|2> my thought was to use separate maas environments and then deploy using one or the other. the issue is that i'm not sure if juju knows about different environments and can make them work together (i doubt it).
[20:58] <TheChistoso|2> saras: you just need to specify that value in your .juju/environments.yaml file
[20:59] <saras> TheChistoso|2: how do i do that for local
[20:59] <TheChistoso|2> are you using lxc containers?
[20:59] <saras> sure how
[20:59] <saras> sounds fun
[21:00] <TheChistoso|2> :D what are you using juju with? AWS? OpenStack? MAAS?
[21:00] <saras> local
[21:00] <saras> i was hoping
[21:01] <TheChistoso|2> just to be clear -- you have locally-defined charms and you're using lxc?
[21:01] <saras> um hum
[21:01] <saras> i would love to use lxc
[21:02] <TheChistoso|2> i haven't setup lxc before so i'm not much help. but first, provide the contents of your ~/.juju/environments.yaml (pastebin it)
[21:03] <TheChistoso|2> you said you're using 1.10 IIRC -- which release? precise? raring?
[21:03] <sarnold> (be careful not to pastebin secrets from that file..)
[21:03] <saras> i am so lost now
[21:03] <saras> https://plus.google.com/hangouts/_/c3ccfdfd6addbbd723b6ac93fa6be15a236e1a07?authuser=0&hl=en
[21:03] <saras> help
[21:05] <saras> is their not a way to get feet wet with out cloud provider
[21:06] <sarnold> saras: I used local provider with juju when I was getting started, and a bit when developing new charms. It seemed about the same complexity to me as the ec2 provider..
[21:07] <saras> sarnold: what do i need in the envirment file to do that
[21:07] <saras> as the getting start guide does not cover it
[21:07] <saras> arrrrrrrrrrrrrrggg
[21:08] <saras> Open `~/.juju/environments/yaml` in your text editor and uncomment out the sections you need for your cloud provider.
[21:08] <saras> is what the getting started guide tell me
[21:09] <saras> their should be way to setup local to test stuff before you play spend money on cloud guys
[21:09] <saras> or am i crazy
[21:10] <TheChistoso|2> that's what lxc and maas can do for you
[21:10] <TheChistoso|2> (well maas or virtual-maas)
[21:10] <saras> then why are they not in the getting started guide
[21:10] <sarnold> saras: iirc, juju generate-config -w should create a template ~/.juju/environments.yaml file for you to edit
[21:11] <saras> yes it is
[21:11] <sarnold> saras: you don't -need- the template, but it is a bit helpful
[21:11] <TheChistoso|2> saras: the getting started guide was a bit difficult, i found, to get started. :D for one, i installed raring and didn't realize it was an older version and juju generate-config didn't exist.
[21:12] <saras> yes i had the same issue
[21:12] <sarnold> older versions I htink just used juju bootstrap to make the config .. yo'ud run juju bootstrap twice, once to create your template, edit that, then run it again to create the headnode..
[21:12] <sarnold> saras: http://pastebin.ubuntu.com/5630403/
[21:12] <sarnold> saras: something like that in your ~/.juju/environment.yaml
[21:14] <sarnold> jcastro: hey, many typos abound in "Take 5 minutes to get to know juju" https://juju.ubuntu.com/get-started/   --- there's a ~./juju (yes, ~./)  and the markdown `` don't appear to do anything :) and the ~/.juju/environments/yaml -- with extra /  :)
[21:15] <jcastro> ok I can fix those
[21:15] <TheChistoso|2> so i tried to run add-apt-repository (in raring) and the command wasn't available. so i sudo apt-get install python-software-properties and it still doesn't work. anybody know which package that lives in?
[21:15] <jcastro> slow down :)
[21:17] <sarnold> software-properties-common: /usr/bin/add-apt-repository
[21:17] <sarnold> jcastro: thanks :)
[21:20] <TheChistoso|2> i'm not getting juju 1.10 -- still on 0.7. I ran the following:
[21:20] <TheChistoso|2> sudo apt-get install software-properties-common
[21:20] <TheChistoso|2> sudo add-apt-repository -y ppa:juju/pkgs
[21:20] <TheChistoso|2> sudo apt-get update && sudo apt-get install juju charm-tools
[21:20] <jcastro> you need juju-devel as the ppa
[21:20] <jcastro> and it's juju-core
[21:20] <jcastro> that'll get  you 1.10
[21:20] <sarnold> ooh juju-core is at 1.10? nice work :)
[21:20] <jcastro> sarnold: man, that's like the third or fourth time I've used markdown on a HTML webpage
[21:21] <sarnold> (that's a confident-sounding round number. :)
[21:21] <sarnold> jcastro: I keep finding myself typing markdown in irc.
[21:22] <saras> any willing to jump into hangout help me out
[21:23] <TheChistoso|2> jcastro: tyvm -- that's much better :D
[21:23] <jcastro> I need to step out for the weekend, I'll be back in a few hours though
[21:23] <TheChistoso|2> now how about the agents on the maas-provided machines?
[21:23] <jcastro> I just realized that made no sense
[21:24] <TheChistoso|2> lol :D
[21:24] <TheChistoso|2> few days?
[21:25] <sarnold> jcastro: have fun for the weekend, for the few hours :)
[21:27] <saras> i get some of the novacut guys in hangout so beat our heads on this one toughter
[21:27] <saras> have fun
[21:33] <TheChistoso|2> so i upgraded juju and ran juju status. i now get: error: no CA certificate in environment configuration. that's needed now-a-days?
[21:45] <TheChistoso|2> hazmat: i'm seeing "error: no CA certificate in environment configuration" -- any thoughts?
[22:04] <TheChistoso|2> juju bootstrap now gives me: error: no tools available
[22:09] <sarnold> TheChistoso|2: do you still have the 0.7 pyjuju stuff installed alongside the 1.10 gojuju?
[22:09] <TheChistoso|2> sarnold: i must have
[22:09] <TheChistoso|2> i'm updating charm-tools right now
[22:14] <TheChistoso|2> juju sync-tools isn't working either
[22:17] <TheChistoso|2> $ juju sync-tools -v 2013/05/03 15:16:20 INFO environs/ec2: opening environment "juju-public" 2013/05/03 15:16:20 ERROR failed to initialize the official bucket environment 2013/05/03 15:16:20 ERROR command failed: environment has no access-key or secret-key error: environment has no access-key or secret-key
[22:17] <TheChistoso|2> whoops, let me try that again...
[22:18] <TheChistoso|2> $ juju sync-tools -v
[22:18] <TheChistoso|2> 2013/05/03 15:16:20 INFO environs/ec2: opening environment "juju-public"
[22:18] <TheChistoso|2> 2013/05/03 15:16:20 ERROR failed to initialize the official bucket environment
[22:18] <TheChistoso|2> 2013/05/03 15:16:20 ERROR command failed: environment has no access-key or secret-key
[22:18] <TheChistoso|2> error: environment has no access-key or secret-key
[22:18] <TheChistoso|2> sarnold: any thoughts?
[22:19] <sarnold> TheChistoso|2: you need some environment variables to declare how to connect to your EC2 instances..
[22:19] <TheChistoso|2> i'm not running w/ EC2
[22:19] <TheChistoso|2> and this is to bootstrap w/ maas
[22:20] <sarnold> TheChistoso|2: I think maas may require the same EC2 env variables... see e.g. http://wiki.debian.org/euca2ools
[22:21] <TheChistoso|2> sarnold: that doesn't sound right to me
[22:23] <sarnold> TheChistoso|2: hrm, you're right, I see none of that mentioned here http://maas.ubuntu.com/docs/quantal/juju-quick-start.html
[22:30] <TheChistoso|2> sarnold: i found this: https://bugs.launchpad.net/juju-core/+bug/1172973
[22:30] <_mup_> Bug #1172973: sync-tools requires aws credentials be set in the (shell) environment <juju-core:Confirmed> <https://launchpad.net/bugs/1172973>
[22:31] <TheChistoso|2> sarnold: however, i set those env vars, but apparently i don't have the right values b/c i get:
[22:31] <TheChistoso|2> $ juju sync-tools -v
[22:31] <TheChistoso|2> 2013/05/03 15:31:15 INFO environs/ec2: opening environment "juju-public"
[22:31] <TheChistoso|2> listing the source bucket
[22:31] <TheChistoso|2> 2013/05/03 15:31:15 ERROR command failed: The AWS Access Key Id you provided does not exist in our records.
[22:31] <TheChistoso|2> error: The AWS Access Key Id you provided does not exist in our records.
[22:33] <sarnold> TheChistoso|2: oof. :/
[22:38] <TheChistoso|2> not sure what to do at this point
[22:40] <sarnold> TheChistoso|2: perhaps try #juju-dev ? I think the go juju folks hang out there..
[22:42] <TheChistoso|2> sarnold: thanks -- i just asked and i'll cross my fingers for an answer
[22:42] <sarnold> TheChistoso|2: good luck :)
[22:42] <TheChistoso|2> thanks :/