[00:26] <lifeless> hazmat: dc thing?
[00:27] <imbrandon> power out all over
[00:27] <hazmat> lifeless, just a thunderstorm with strong winds, dc == washington, dc
[00:27] <hazmat> not data center :-)
[00:27] <lifeless> hazmat: I figured that, but its taken out the whole city grid?
[00:28] <hazmat> lifeless, quite a bit of the suburbs, like 3.5 million were without power across several states
[00:28] <hazmat> within the city itself only about 65k
[00:28] <lifeless> thats a fair number; does the US have a fragile power grid?
[00:28] <imbrandon> heh,  whats causing the ec2 east too
[00:28] <hazmat> but theirs a tech center an hr west of the city, that has data centers for most of the tech industry that took out the local grid
[00:29] <hazmat> for ec2 i imagine a backup generator didn't do its job in one of the azs
[00:29] <imbrandon> lifeless: not really , just bad storm
[00:30] <imbrandon> hazmat: yea likely
[00:30] <hazmat> lifeless, debatable :-) but its an unusual event, hurricane force winds 80mph on a storm
[00:30] <hazmat> afaik this is the largest non hurricane outage for this area
[00:31] <hazmat> larger than some of the severe snow storms we had a few years back as far wide spread outage
[00:32] <imbrandon> wow
[00:32] <hazmat> lots of trees down all over, i saw like 10 down, and 2 cars crushed walking my dog this morning
[00:32] <imbrandon> pics!! jk
[00:33] <hazmat> imbrandon, http://www.cnn.com/2012/06/30/us/dc-storm-damage/index.html
[00:33] <hazmat> imbrandon, the internet has your back :-)
[00:33] <imbrandon> kinda funny how instagram and a few others are cactching hell for not having geo redundant stuff
[00:33] <imbrandon> heh
[00:34] <imbrandon> oh wow
[00:35] <lifeless> meep
[00:35] <imbrandon> In neighboring Virginia and Maryland, seven people were killed by fallen trees, authorities said.
[00:35] <imbrandon> wow
[00:36] <imbrandon> says new storms may head in tonight too
[00:37] <imbrandon> you good where you are hazmat ?
[00:37]  * imbrandon hopes marcoceppi is cool too, he is in that area somewhere
[00:38] <hazmat> imbrandon, yup we're good, its not that bad for most
[00:38] <imbrandon> cool
[00:39] <hazmat> just finished adding a few new goodies to juju-jitsu env import/export to a json file, and service machine placement when deploying
[00:39]  * hazmat switches into review mode
[00:39] <imbrandon> sweet, was getting tired of adding placement:local to the env.y and rembering to remove it
[00:40] <imbrandon> brb afk a few , i'll likely be on most of the evening if you need a guiney pig tho
[00:42] <hazmat> imbrandon, they've both been tested by others but feedback welcome
[07:04] <nathwill> upgrade-charm looks at revision number to determine "newer"? trying to test upgrade-charm hook and it's not picking up my local charm as newer...
[11:29] <marcoceppi> Lost power for a bit, but we're okay here
[18:37] <_mup_> juju/trunk r547 committed by kapil@canonical.com
[18:37] <_mup_> [trivial] adjust readme to point to external docs
[22:28] <hokka> i would like to have multiple local environments. i understand i can specify different directories, and probably need to hack juju code to allow for a custom parameter -- to specify the name of the network, 3rd octet for the network. any other catches?
[22:41] <lifeless> hokka: the network name shouldn't matter AIUI,
[22:41] <lifeless> hokka: unless you want really detailed control
[22:42] <hokka> lifeless: i want to be able to deploy services attached to different networks. then I will be able to control exposure of these networks to the outside world. this is because "expose" is not particularly fine grained.
[22:43] <hokka> lifeless: i thought libvirt uses network name to distinguish between bridges to connect a container to. maybe i'm wrong, need to check.
[22:49] <hokka> i have serious doubts the name of the network has no meaning, it is used extensively throughout the code http://bazaar.launchpad.net/~juju/juju/trunk/view/head:/juju/providers/local/network.py
[22:51] <hokka> could anybody please give me a hint how IP address allocation is managed in juju with local provider? do I need to take care that as well if I want multiple networks?
[23:04] <lifeless> hokka: I'm pretty sure its done via dnsmasq
[23:04] <lifeless> hokka: I'm not sure what expose does locally, I would have expected it to let just traffic to the one port through.
[23:04] <hokka> lifeless: thanks, seems to be so
[23:05] <hokka> anyway, I think it is would be nice to have different networks, will try to implement it
[23:07] <lifeless> we should really have a true cloud API that happens to drive LXC.
[23:07] <lifeless> would be more consistent, reusable, etc.
[23:43] <hokka> lifeless: what do you mean by 'true cloud API'?
[23:47] <hokka> i've implemented optional 'net-name' and 'subnet' config options and they are used to create a new libvirt network just fine. but i'm stuck at the stage of injecting the new network name into lxc container config file. can't figure out how to access the config of the current environment from within UnitContainerDeployment._get_master_template(). any clue?
[23:57] <lifeless> hokka: I mean a JSON HTTP based API that will bring up containers, destroy them, provide blob storage and support image uploads.
[23:57] <lifeless> it would take care of restarting containers on host reboot
[23:57] <lifeless> and firewall / network rules etc
[23:57] <lifeless> probably be best built on top of libvirt.