=== defunctzombie_zz is now known as defunctzombie
=== thumper-afk is now known as thumper
[02:58] <alicam> Anyone online who can help me setup on EC2 in Sydney (ap-southeast-2) because nothing I do is working!
[02:58] <alicam> ap-southeast-2 is not recognised as a valid region when you add it to the environment file.
=== defunctzombie is now known as defunctzombie_zz
[03:01] <alicam> When I add ec2.ap-southeast-2.amazonaws.com as the ec2-uri variable, as an override for region, it breaks the secure transport, and the auth fails.
[03:14] <thumper> alicam: I have also noticed problems with ap-southeast-2
[03:14] <thumper> alicam: sorry, don't have a fix though
[03:14] <thumper> not all regions created equal
[03:14] <thumper> sydney ec2 lies
[03:14] <thumper> in its api
[03:14] <thumper> I start an image there
[03:14] <thumper> but ask over the api if it is running
[03:14] <thumper> nope
[03:15] <thumper> but I can see it in the aws console
[03:15] <thumper> doesn't matter, api lies
[03:15] <thumper> so *sadface*
[03:15] <alicam> Sadface here too...
[03:16] <alicam> Looks like I'll try for Singapore of Nth Cal
[03:16] <alicam> Any experience with those?
[03:16] <thumper> sorry, no experience there
[03:17] <thumper> good luck
[03:42] <marcoceppi_> alicam: north California should be fine. haven't tried Singapore though
[03:47] <alicam> Thanks
[03:47] <alicam> Thanks Marco. I tried Singapore and it couldn't speak to the instance. It fired up an instance, but then lost contact with it! eird
[03:47] <alicam> Weird
[03:49] <alicam> Marco, is there any reason why I'd have trouble specifying t2.micro instances in my constraints? I don't need m1s
[03:49] <alicam> t1.micro instances, I meant, sorry!
[05:03] <alicam> marcoceppi_: is there a way to make NFS share an instance with memcached, in the same way that mysql and php-fpm can be made to share?
=== wedgwood_away is now known as wedgwood
=== alexlist` is now known as alexlist
=== agy_ is now known as agy
=== ubot5` is now known as ubot5
=== wedgwood is now known as wedgwood_away
[12:22] <ahasenack> hi guys, where in launchpad can I see the merge proposals against a specific charm?
[12:22] <ahasenack> or, in lp lingo, "+activereviews"?
[12:23] <jamespage> ahasenack, all of the official charm branchs are owned by ~charmers
[12:24] <jamespage> but http://jujucharms.com/review-queue is the definitive source
[12:24] <jamespage> jcastro, sorry about all of the openstack merge proposals
[12:24] <jamespage> :-)
[12:24] <ahasenack> jamespage: hm, ok, but I wanted to get an email when a MP was put up for review against a specific charm
[12:24] <jamespage> adam_g and I will work through those (i.e. we won't merge our own proposals)
[12:25] <jcastro> jamespage: I wasn't complaining
[12:25] <jcastro> I was more like "dang, impressive"
[12:25] <jamespage> jcastro, :-)
[12:25] <ahasenack> jcastro: hey, is there a charm school today?
[12:26] <jcastro> yeah
[12:26] <jcastro> I'll post the reminder to the list
[12:29] <ahasenack> ok
=== wedgwood_away is now known as wedgwood
[13:36] <jcastro> sidnei: scale of 1 to 10 how much work is that zero downtime for haproxy?
[13:36] <jcastro> sidnei: that link was awesome btw.
[13:36] <jcastro> http://www.igvita.com/2008/12/02/zero-downtime-restarts-with-haproxy/
[13:36] <jcastro> ^^ for those interested
[13:37] <dpb1> Is there a workaround-type-replacement for debug-hooks missing in juju-core?
[13:37] <dpb1> I want to run commands interactively in the context of a hook on a machine to test out what is happening.
[13:47] <sidnei> jcastro: im considering other alternatives mentioned elsewhere, let me dig a couple more links
[13:49] <jcastro> dpb1: I'm not aware of any at this time
[13:49] <jcastro> marcoceppi_: ^^^
[13:49] <sidnei> jcastro: http://blog.balancedpayments.com/payments-infrastructure-suspending-traffic-zero-downtime-migrations/ is the other one i had in mind, it requires haproxy 1.5 though
[13:50] <sidnei> http://labs.animoto.com/2010/03/10/uptime-with-haproxy/ is a variation of the first one
=== wedgwood is now known as wedgwood_away
=== wedgwood_away is now known as wedgwood
[15:32] <marcoceppi_> dpb1: not that I know of yet
[15:32] <dpb1> Hey marcoceppi_, :(
[15:34] <marcoceppi_> dpb1: I haven't really tried yet, A lot of my testing is "hook broke, ssh in, patch file, resolved --retry, check again"
[15:35] <dpb1> marcoceppi_: ya, same.
[15:36] <dpb1> marcoceppi_: I was thinking at one point of trying out some kind remote debugging, but then I... slapped myself.
[15:37] <marcoceppi_> dpb1: it'd be nice if core added that again *severe winking towards #juju-dev members*
=== defunctzombie_zz is now known as defunctzombie
[15:38] <dpb1> marcoceppi_: I sent an email to juju-dev, hopefully that gets some attention. :)
=== defunctzombie is now known as defunctzombie_zz
[15:59] <jcastro> http://ubuntuonair.com/ is up to date, we'll star the charm school in about 2 minutes!
[16:03] <jcastro> ok let's begin
[16:03] <jcastro> if you want to ask questions during the charm school then feel free to do so here
[16:03] <jcastro> you can follow the video stream on http://ubuntuonair.com/
[16:03] <jcastro> today we're doing How to Write a Charm Part II
[16:04] <jcastro> and Mark Mims is going to show us a charm from scratch using what we call a platform charm -- this example will be node.js!
[16:05] <evilnickveitch> \o/
[16:08] <marcoceppi_> woo who!
[16:15] <arosales> charm schooling :-)
[16:59] <jcastro> m_3: you appear to have frozen
[17:04] <arosales> jcastro, m_3 thanks!
[17:07] <m_3> tmux-driven-presentations... whoohoo!
=== wedgwood is now known as wedgwood_away
[19:25] <blabla> hi, can someone please point me to the last Jono Bacon Q&A?
[19:25] <blabla> can't find the last videos in the youtube channel
[19:33] <jcastro> for juju?
[19:34] <jcastro> anyone ever get an error similar to this
[19:34] <jcastro> error: Get https://s3.amazonaws.com/juju-e4698ca2abde47dddd405d1f595745e596f/provider-state:
[19:35] <jcastro>  remote error: handshake failure
[19:35] <jcastro> I'm doing a bootstrap and it's alternating between that and error: The specified bucket does not exist
[20:29] <marcoceppi_> jcastro: get them all the time
[21:30] <ahasenack> hm, I'm not understanding why relation-get needs a unit id
[21:30] <ahasenack> in a config-changed hook, I understand I need a -r <relation_id>
[21:31] <ahasenack> isn't that enough? the relation ids are between services, not units
[21:31] <ahasenack> i.e., the r_id between server/0 and client/0 is the same as between server/0 and client/1
[21:31] <ahasenack> meaning in my mind that whatever relation values have been set, the relation id is enough to read them
[21:32] <ahasenack> it wouldn't be different if I did relation-get .... client/0 or client/1
[21:33] <sarnold> ahasenack: would it provide any utility for client/N to determine which of server/0 or server/1 to prefer for its queries/requests?
[21:34] <ahasenack> sarnold: the get command ("question") isn't sent to the server unit, it's "juju" that answers
[21:34] <ahasenack> if I understood the question correctly
[21:34] <sarnold> ahasenack: sorry, I'll back up a bit..
[21:34] <ahasenack> 2013/05/31 21:19:57 INFO worker/uniter/jujuc: running hook tool "relation-get" ["-r" "test-r:1"]
[21:34] <ahasenack> 2013/05/31 21:19:57 DEBUG worker/uniter/jujuc: hook context id "andreas-client/1:config-changed:1660150820142556165"; dir "/var/lib/juju/agents/unit-andreas-client-1/charm"
[21:34] <ahasenack> 2013/05/31 21:19:57 INFO worker/uniter: HOOK error: no unit id specified
[21:34] <ahasenack> that was the error
[21:34] <ahasenack> this is a config-changed context
[21:35] <ahasenack> I added several units before, I printed the relation-ids and they are all the same for all units
[21:35] <ahasenack> meaning, the relation id is between services, not units
[21:35] <sarnold> ahasenack: some clients / protocols (memcached? httpd?) allow clients to query a server and when providing a service, you might wish to spread out which servers get the queries.. I'm curious if requiring the unit id would allow writing config files to say "this client should prefer server/0 before server/1", and another client to configure "server/1 before server/0"
[21:36] <ahasenack> sarnold: ok, I see. What I understood is that it doesn't matter which unit I put in that command, the answer will be the same
[21:36] <ahasenack> hence, why do I have to put the unit in there at all
[21:37] <ahasenack> unless with relation-set I can scope things for a specific unit...
[21:37] <ahasenack> hm
[21:37] <sarnold> ahasenack: memcached isn't a perfect example, since most front-end libs will do something slightly better anyway.. I was just wondering if the intent was to make this feature possible.
[21:37] <ahasenack> let's run a hook again to get relation-set --help
[21:37] <sarnold> hehe
[21:39] <ahasenack> hm, no, relation-set only accepts at most a relation id
[21:39] <sarnold> heh
[21:39] <ahasenack> while
[21:39] <ahasenack> 2013/05/31 21:38:36 INFO andreas-client/1: usage: relation-get [options] <key> <unit id>
[21:39] <ahasenack> so, I don't know
[21:39] <ahasenack> relation-get prints the value of a unit's relation setting, specified by key.
[21:39] <ahasenack> (that was from --help)
[21:40] <ahasenack> that help doesn't seem accurate, as I can call relation-get without any parameters in a relation hook
[21:40] <ahasenack> neither key nor unit are mandatory in that context
[21:41] <ahasenack> mailing list to the rescue
[21:42] <ahasenack> at best I'll know the answer just before hitting "send"
[21:42] <sarnold> :)
=== sidnei` is now known as sidnei