=== jcsackett_ is now known as jcsackett === yo61_ is now known as yo61 [01:05] fuzzy_: still wrestling with mongodb? [01:09] lazyPower: yes [01:09] Following, https://maas.ubuntu.com/2012/11/30/lets-shard-something/ using the command "juju add-relation mongos:mongos-cfg configsvr:configsvr" I get "ERROR service "mongos" not found". What am I doing wrong? [01:09] do you have a service named 'mongos' deployed? [01:10] not that I see [01:10] that's why the howto doesn't make sense [01:12] fuzzy_: are you trying to deploy the sharded/replicated cluster? there is a bundle for it [01:12] that howto looks really old [01:12] https://jujucharms.com/bundle/mongodb/5/cluster/ [01:12] take care to note that there are 3 replicated units per shard/configsvr [01:13] thats 13 total units listed in that bundle [01:13] thank you === scuttle|afk is now known as scuttlemonkey [01:14] fuzzy_: the mongodb charm is notoriously finicky as well. I've put some elbow grease into it in the past - if you run into any issues with it please file bugs here: https://bugs.launchpad.net/charms/+source/mongodb [01:15] i'm just going to reroll everything and try one more time [01:18] ok. I typically have to do that when i botch a mongodb deploy as well, start over from scratch [01:18] there appears to be a race condition in teh charm I haven't tracked down on one of the roles. [01:18] sorry about the weak experience with the charm :( MongoDB has been a problem child for me when triaging/maintaining [01:19] I'm open to any suggestions on how we can make it a better experience, any documented cases / examples will definately help as we move forward with the next iteration. [01:19] i'm curious why python-yaml ins't part of a machine bootstrap [01:20] I've run into this on 2 substrates - DigitalOcean and Linode [01:20] its standard on AWS, HPCloud, and Joyent [01:20] i figured that is sort of lsb stuff [01:20] ah [01:20] i think they are using modified cloud images, but i have no data to back up my claim [01:21] I've had to go back through and add the python-yaml dependency to any of the python based charms as i run across them testing on DO [01:21] you hooked the do api [01:21] is it fully automated yet? [01:21] mostly - its a phaux provider [01:21] https://github.com/kapilt/juju-digitalocean [01:21] because linode does have an api [01:21] and it's not hard to hook [01:22] mhm, i've mentioned this approach to you before [01:22] i just didn't know how hard the juju side of it would be [01:22] i know you did [01:22] i just lost the link [01:22] well that plugin is about all i can reference for you - i'm not familiar enough with core to make a recommendation. [01:22] do you use it to admin your do boxes? [01:22] however hazmat has all teh goods when he's around. he's the author of that plugin. [01:22] i do. I've got ~ 6 nodes running on DO right now [01:23] hmmmmmmm [01:23] (including the api server) === axw__ is now known as axw [01:25] damn that drives a hard bargin [01:27] are you using digital oceans private network to do the deed? [01:27] lamont: [01:27] ack [01:27] lazyPower: [01:27] I'm watching ;) no need to ping [01:27] Yeah, it automatically adds the private networking to the droplet. [01:31] *sigh* I hope this is worth it [01:31] * fuzzy_ crosses fingers [01:33] lazyPower: see if you got any digital ocean promo codes for new accounts so you can get hooked up with this [01:34] i do actually, 1 moment [01:34] https://www.digitalocean.com/?refcode=b6ac387b6f36 - $10 free credit with a new account [01:44] ok I think I got a do account now [01:44] awesome [01:45] awww crap [01:45] lol [01:45] stupid do [01:45] ? [01:45] this might not actually work [01:46] remember when they screwed up nyc2 for all those weeks last year? [01:46] mhmm, the juniper upgrade [01:46] well i moved everything off of them [01:46] but one of the droplets kept running until it racked up a bill [01:47] whoops [01:47] and I told them no way I was paying it, I wasn't even using the box it was a left over from an idle [01:47] now they just took $5 out of paypal and locked my account [01:47] before i could make a droplet [01:47] oh nope there it goes [01:47] whooo [01:48] ok i'm in business [01:48] the DO guys are really reasonable. [01:49] Every time i've reachedo ut or support i get an answer in an hour or less, and 90% of the time its resolution. [01:49] *out for [01:49] actually nope [01:49] i'm locked [01:49] lets seee [01:50] but now i'm limited to 1 droplet [01:51] I'd open a ticket and explain whats going on [01:51] they should crank you up to 5 instances for your first month or so [01:52] I just did [01:52] now I wait === scuttlemonkey is now known as scuttle|afk [02:42] https://github.com/charms/mongodb when I try juju deploy mongos i get failure [02:43] weird it took that time [02:53] lazyPower: http://hastebin.com/dadetadepu.sm [02:53] it claims to be running, but i can't seem to find the exposed console for mongo [02:58] yea i don't get it, netstat on that box shows it's listening to 28017, i made sure to disable it's firewall and i still can't see it remotely [03:00] http://hastebin.com/patadakepo.hs [03:00] makes no sense [03:03] So reading the juju log on mongos/0 i get alot of permission denied. I think that might be because of ufw [03:03] i'm going to tear it down and retry without ufw and ssh configured differently === uru_ is now known as urulama === MasterPiece_ is now known as MasterPiece === MmikeM is now known as Mmike === MmikeM is now known as Mmike === CyberJacob|Away is now known as CyberJacob [07:10] yea it was ufw fighting me i got a mongo console lazyPower === CyberJacob is now known as CyberJacob|Away === wesleyma` is now known as wesleymason === Spads_ is now known as Spads === Spads_ is now known as Spads === jam1 is now known as jam === Spads_ is now known as Spads === scuttle|afk is now known as scuttlemonkey === fabrice is now known as fabrice|lunch === roadmr is now known as roadmr_afk === lborda is now known as lborda_ === lborda_ is now known as lborda__ === fabrice is now known as fabrice|coffee === fabrice|coffee is now known as fabrice === ianunruh_ is now known as ianunruh === tdc_ is now known as tdc === roadmr_afk is now known as roadmr === makyo_ is now known as Makyo [15:56] gnuoy`, nudge - https://code.launchpad.net/~james-page/charms/trusty/nova-cloud-controller/cluster-sync-fix/+merge/234475 [15:56] ack === seelaman` is now known as seelaman === ev__ is now known as ev === gnuoy` is now known as gnuoy === sarnold_ is now known as sarnold [17:07] Review Queue is going down for an upgrade [17:08] Review Queue is going down for an upgrade === roadmr is now known as roadmr_afk === hazmat` is now known as hazmat === CyberJacob|Away is now known as CyberJacob === roadmr_afk is now known as roadmr [20:12] Hey marcoceppi since we have automated testing now, we should start requiring all new charms to have tests. [20:14] mbruzek, that's what I thought the policy was... I didn't realize it hadn't been more widely disseminated [20:14] mbruzek: that's not a bad idea, we should propose this policy to the list [20:14] jose, sorry about that :[ [20:15] There's a bunch of previous talks about it on the list, I'll round them up and make a new post [20:16] whit: don't worry, not a prob :) === CyberJacob is now known as CyberJacob|Away [20:49] Should we consider unit testing sufficient? [20:50] mbruzek lazyPower jose ^? et al [20:50] what would you define as unit testing? [20:51] anything that tests the hooks without being deployed [20:51] I would be good with something saying 'hey! I deployed mycharm successfully, even though I haven't accessed it it didn't error while deploying+relating!" [20:51] would that be unit testing? [20:52] so, some charms have python unit tests [20:52] marcoceppi, I don't see how we can ask authors to write python tests if they don't know it. [20:52] would that be enough to satisify new policy for all chams need tests [20:53] mbruzek: I'm saying, is unittesting enougth to satisfy testing or should we required functional/integration testing as well? [20:53] marcoceppi, could they write a bash deployment or a bundle [20:53] that's not unit testing, that's fucntional testoing [20:53] and that's basically what we think of testing today, a la bundle/amulet tests [20:54] marcoceppi, I think we should unit test at minimum but HIGHLY suggest having a bundle or amulet test that deploys a working unit. [20:54] I have to leave for uni. Giving some input later. [20:54] jose: thanks o/ [20:55] mbruzek: right, that's my question, is unittesting enough to satisfy testing requirement in policy [20:56] yes. [20:56] I don't like only unit tests because I still may not know how to deploy said charm. [20:56] cool, I'll wait for feedback from jose and lazyPower and draft teh policy [20:56] mbruzek: that;'s a readme issue [20:56] and nackable based on that alone [20:56] yes. === Guest31361 is now known as wallyworld === urulama is now known as urulama-afk === CyberJacob|Away is now known as CyberJacob [21:46] marcoceppi: i think unit tests or integratino tests at bare minimum. i've been treating it as an either/or [21:47] marcoceppi: my reasoning is we still have a ton of bash charms in the store, and bash hasn't exhibited any ground breaking unit test frameworks. [21:58] +1 lazyPower === CyberJacob is now known as CyberJacob|Away === ayr_ton is now known as ayr-ton === mup_ is now known as mup