=== natefinch-afk is now known as natefinch === thumper is now known as thumper-afk === thumper-afk is now known as thumper [08:30] Hello Juju World! [09:45] Hi! jamespage: are you here? [13:27] if anyone is interested, charm and charm-tools are now snapped up in edge for testing. [13:27] `snap install charm --edge` [13:29] marcoceppi: sweet [13:29] marcoceppi: was there any issues with the macaroon storage path and such? [13:30] not when you PLUGinto home [13:30] git was the biggest problem >_> [13:30] rick_h_: https://github.com/juju-solutions/charm-pkg/blob/master/snapcraft.yaml [13:31] rick_h_: however, putting the macaroon data in $SNAP_DATA_PATH is a good idea [13:31] very cool marcoceppi [13:49] stub, question for ya with the apt layer [14:05] cholcombe: yo [14:06] stub, i know it's super late there. I'm running into some trouble getting the config.yaml correct with the apt layer. I followed the readme closely but it won't build or deploy [14:06] stub, i'm falling back on adding the apt sources in my reactive.py which isn't really the best [14:07] Got a pastebin? [14:07] sure one sec [14:09] stub, https://gist.github.com/cholcombe973/e1405d70ba4921178e0908d7c4a6ba1c [14:10] cholcombe: at first glace that looks fine. What happens when you deploy it? [14:10] nothing it seems. it doesn't setup either of the sources in the /etc/apt/sources.list file [14:11] when i include the same thing in the python reactive.py file it works ok [14:11] cholcombe: Does your layer.yaml declare its using layer:apt, and its all been rebuild and upgraded? [14:12] stub, yeah i have this in there: includes: ['layer:basic', 'layer:apt', 'interface:ceph-client'] [14:13] The indentation seems wonky - it doesn't parse as yaml [14:14] oh yeah i fixed that [14:14] i recreated the file to show you what it looked like [14:16] I'd need to see the logs to diagnose further [14:16] stub, ok no problem. i can pack them up [14:16] stub, do you have a working example i can take a look at? [14:17] The PostgreSQL charm uses the apt layer. Or are you after a mojo spec that configures it? [14:18] postgres charm sounds fine. i'll check that out :) [14:18] either one. i'm looking for how the config.yaml was setup [14:18] prob the mojo spec then [14:18] telegraf-charm might be better [14:18] ok [14:18] launchpad.net/telegraf-charm [14:19] awesome thank you [14:21] cholcombe: In your logs, you should see 'Initializing Apt Layer', soon after which it will call charmhelpers.fetch.configure_sources() (not sure if that gets logged) [14:22] One easy way to test the layer is being bootstrapped - set the 'package_status' config to 'bogus'. It should put your unit into a blocked state until you give it a valid value. [14:23] ok [14:53] cholcombe: oh, that is your config.yaml you posted. [14:54] cholcombe: You are setting install_sources and install_keys to strings, but in config.yaml options are mappings. You want to be setting options: install_sources: default: , not options: install_sources: [14:56] https://git.launchpad.net/telegraf-charm/tree/config.yaml [14:56] (I was thinking that was a deployment config for some reason) [14:57] anybody knows where is jamespage? or where I can find him? maybe he is on vacation? === natefinch is now known as natefinch-afk [15:09] magicaltrout: Are you able to join the meeting? [15:13] stub, thanks! [15:14] cholcombe: I'm surprised juju let you deploy it at all [15:14] stub, juju is lazy haha [15:32] stub, i updated the gist. is this more like what the apt layer expects: https://gist.github.com/cholcombe973/e1405d70ba4921178e0908d7c4a6ba1c [15:37] that doesn't look right [15:38] https://gist.github.com/stub42/0438059a79dd1cf7d020fa8ad039b3dc [15:42] gah... just revised it. I had bad yaml in one of the strings [15:42] * stub wishes again we had richer data types than strings [15:42] stub, i'll try that :). Yeah this yaml is miserable [15:43] I don't mind yaml. Its having to encode yaml into strings that get embedded in yaml :) [15:43] lol === mup_ is now known as mup === mup_ is now known as mup [16:03] stub, nope that doesn't work either. nothing is added to the /etc/apt/sources.list file [16:07] cholcombe: How about /etc/apt/sources.list.d ? [16:07] nope that's blank also [16:08] Do you see the 'Initializing Apt Layer' message in the logs? [16:09] i do yeah [16:09] Output from gpg after that? [16:09] nope. lemme gist it [16:10] https://gist.github.com/cholcombe973/7438f1a25fe2da1a1c6c3a1014c67898 [16:10] There isn't much logging here, as it is just calling out to charmhelpers [16:11] right [16:12] Have a look at your generated charm's config.yaml, and see if charm build included your updated defaults [16:13] stub, my bad i forgot to run charm build again. let me try and deploy this again === frankban is now known as frankban|afk [16:22] stub, ah yes now we're getting somewhere. invalid gpg armor header :) [16:47] cory_fu: For testing the new jdk setup we have in the bigtop base layer, which bundle is the most up to date for testing? Is it https://github.com/juju-solutions/bundle-bigtop-processing-mapreduce, or one of the others? [16:52] Is that still a thing? We need to delete that repo [16:53] petevg: https://github.com/juju-solutions/bigtop/tree/master/bigtop-deploy/juju/hadoop-processing [16:53] cory_fu: aha. Thx. [17:24] cholcomb: sup [17:32] cholcombe: sup [17:33] cholcombe: what plans have you for vault_charm? [17:35] cholcombe: I think we need to combine forces on the vault thing [17:36] cholcombe: https://jujucharms.com/u/jamesbeedy/consul/1, https://jujucharms.com/u/jamesbeedy/consul-agent/6, https://jujucharms.com/u/jamesbeedy/vault/8 [17:36] they all need readme, icon, and tests :-( [17:37] the^ consul, consul-agent, vault ha stack deploys successfully :-) [17:38] cholcombe: I've put alot of work into it recently .... lets chat when you get a minute [17:49] bdx, i'm trying to get the vault charm to hook up with consul so i can store cephx keys on it [17:49] bdx, do you have a bundle i can deploy of that whole setup? [17:53] cholcombe: omp [18:02] cholcombe: https://gist.github.com/jamesbeedy/102d1bbcfe5dbf6227f014348026d0a3 === natefinch-afk is now known as natefinch [18:14] bdx, sweet thanks [18:14] cholcombe: one existing caveat to the vault charm is that non-leader units error when you request a token though [18:14] http://paste.ubuntu.com/23065296/ [18:15] bdx, i see [18:15] cholcombe: because they all try to request the token [18:15] but only a singleton request need be made [18:16] yeah [18:16] that should be a leader only thing [18:17] yeah ... is there an 'is_leader()' call of some sort that you know of? [18:18] bdx, yeah. are you using layers or old school? [18:18] layers [18:19] there's a leadership layer i remember seeing on interfaces.juju [18:20] yeah, vault is already using it [18:21] cory_fu, do you have any experience with the apt layer and custom sources? I can't get it working :-/ [18:22] cholcombe: A bit. The most common thing I've seen go wrong is that, since config options can't be nested and thus it uses an embedded yaml string, the indentation on the value is really finicky [18:22] cholcombe: this might help there too -> https://github.com/jamesbeedy/layer-kibana/blob/master/config.yaml [18:23] bdx, yeah i'm having trouble with the gpg key. it's complaining about an invalid header but i know it's valid [18:23] cory_fu, yeah the value is hard to get right [18:23] cholcombe: Can you point us to your config.yaml? [18:23] bdx, i'll try to copy your example [18:23] sure [18:24] cory_fu, bdx https://gist.github.com/cholcombe973/e1405d70ba4921178e0908d7c4a6ba1c [18:24] cholcombe: Your indentation is off on your GPG key [18:24] you need another space [18:24] ok [18:25] so 2 spaces in [18:25] Told you it's super finicky. It's really annoying to deal with, and setting it on the CLI is even worse [18:25] I somewhat feel like those params should be options instead of configs [18:25] yeah. we need a better way than this [18:25] This will get way better when config options support the schema syntax that action params do [18:25] aah [18:26] ha [18:27] Unfortunately, that's a breaking change and since it doesn't look like it's going to be in 2.0, I'm not holding my breath. [18:27] Unless the core devs can find a way to make it backwards compatible, which would be rather a bit more work [18:28] 11th deploy is a charm? lol [18:28] * cholcombe crosses fingers [18:28] :) [18:30] petevg: Any objection to me rebasing and squashing BIGTOP-2476-zookeeper? [18:30] No conflicts [18:30] @cory_fu: other than it never being a good idea to edit the history of a branch that multiple people have checked out :-) [18:31] The Bigtop people upstream don't seem to like to use github, including github's convenient "squash this when I merge" feature, though. [18:31] Yeah, except that the Bigtop reviewers have specifically asked for nicely squashed PRs [18:31] So, it's probably fine to just squash it now. [18:32] bdx, yeah so there should be an is_leader() function. just requires a little searching [18:32] Right. We could stop using the GitHub <-> Jira integration and submit patches that we manually update instead, but I personally don't find dealing with rebases all that painful [18:32] cory_fu: yeah. Go ahead. I have some outstanding changes against my local branch, but they're not difficult to recreate. [18:32] git stash is your friend. ;) [18:34] cory_fu: In the past, I've had people do really horrible thing with merges when they tried to "fix" a squashed, which is why I try to avoid them. I won't do bad things with it, though, so we should be okay. [18:35] *things, *squashed branch [18:35] cory_fu: what we really need to do is get the dang PR merged so that we can stop merging other things to it. [18:37] cholcombe: yeah, just found it. thx [18:37] bdx, cory_fu thanks! That got it to install finally [18:38] np! nice! [18:39] cholcombe: I'll keep you posted with my vault progress, going to get this mult-token-request thing squared away asap [18:39] bdx, awesome [19:30] what do people typically do for charms if the deb package being installed has an ncurses setup interface that gets invoked while installing? coreycb you might know this [19:32] cholcombe, I'm not sure, but I think that's frowned upon for deb packaging. for openstack packages, one of the reasons we have separate core packages from Debian, is because Debian uses debconf prompts to configure things. [19:32] frowned upon for Ubuntu that is [19:32] yeah the openattic package has debconf prompts. i don't know how to disable them [19:33] or how i can give it a config file or something [19:34] cholcombe, It looks like you can disable debconf - http://stackoverflow.com/questions/4401431/disable-prompts-while-installing-a-debian-package [19:35] coreycb, interesting. thanks [19:35] cholcombe, good luck [19:35] coreycb, ugh. it's always something [19:35] yep [19:57] is there a way to access auth_url specified in environment.yaml file used for bootstrap, in charm hook? [20:16] hi all.. is there a way to access auth url specified in environment.yaml file from charm hook? [20:39] mayurisapre: no, the charms are disconnected from the environment they're deployed in [20:45] marcoceppi : okay..is there a way to inject these values in deployed charm? [20:45] mayurisapre: you can have them set via a configuration option on the charm [20:45] mayurisapre: may I ask what you're charming up? [20:46] i am deploying in openstack env [20:46] and need these values for further operations in charm hooks [20:47] what is the configuration option you just suggested? [20:51] marcoceppi : ohh sorry .. are you talking about config.yaml file provided in charm? [20:52] mayurisapre: yes [20:53] yes.. i thought of that option but if I could get auth_url that is more suitable in my case === natefinch is now known as natefinch-afk === deref is now known as alex-deref [21:35] cholcome: https://github.com/jamesbeedy/charm-vault/blob/master/reactive/vault.py#L56 [21:35] cholcombe: ^ thats all it took [21:35] bdx, correct [21:36] pumped [22:16] bdx: I don't have context, but could you use `@when('leadership.is_leader')` [22:16] instead of an if block [22:16] also, love seeing progress on the vault charm [22:16] it's something I've heard a few people ask for already [22:44] marcoceppi: thanks for pointing that out, -1 me for not seeing that state decorating the function directly above ;/ === natefinch-afk is now known as natefinch