=== arosales_ is now known as arosales === fuzzy_ is now known as Fuzai === CyberJacob is now known as zz_CyberJacob [09:01] gnuoy`, could you give me a review of https://code.launchpad.net/~james-page/charms/trusty/rabbitmq-server/network-splits/+merge/247384 [09:02] I've manually run lint, amulet and unit tests [09:15] jamespage, I don't see the link between network splits and the new ceph.create_pool line [09:15] gnuoy`, there is none other than I noticed it was broken [09:15] gnuoy`, I can do that as a separate MP if you like [09:15] jamespage, no, that's fine. [09:17] jamespage, approved [09:21] gnuoy`, I've also switched osci to use next for rabbitmq-server [09:22] thanks [10:00] marcoceppi_, gnuoy`, btw I backported the latest juju-deployer release into the stable PPA [10:28] hi [10:34] gnuoy`, https://launchpad.net/charms/+milestone/15.01 [11:08] hi all [11:08] i would like to run my django tests over several (say 100) servers [11:09] i was considering deploying ec2 instances and make each of them run a part of the tests [11:09] a friend of mine told me about juju [11:11] so my need is basically to create 100 clones of the same instance, run a slightly different command-line on each of them and aggregate results [11:47] jamespage, could I get a review of the mps associated with Bug #1403132 when you have a moment please? [11:47] Bug #1403132: hacluster default ports conflict between openstack charms

Charms Collection):New>

[11:52] gnuoy`, +1 that looks fine === sfeole` is now known as sfeole [13:39] gnuoy`, all of the nrpe stuff has landed right? [13:40] jamespage, almost, I think there was a mongodb mp that someone else was reviewing and I ceilometer-agent slipped through the net but I'm hoping to have that done rsn [15:14] arg, juju test doesn't respect $JUJU_HOME [15:15] must have a hardcoded ~/.juju ? [15:50] bloodearnest: we've been superceeding juju test with bundletester === roadmr is now known as roadmr_afk [15:51] lazyPower, I assumed that was for testing bundles? [15:51] That was the original target, it executes everything that is in /tests that's chmod +x however [15:52] and sniffs makefile targets to run linting + an implicit charm proof [15:52] its a parity tool for use with CI, as thats our exclusive tool for CI test running [15:57] lazyPower, where can I get it? [15:57] bloodearnest: pip install bundletester - its not a package in the repos yet [15:57] we have that as a target for this/next cycle. [15:58] ack [16:00] bloodearnest fwiw in the next few weeks when charm-tools is released, bundletester will replace juju-test, so running juju test will execute bundletester under the hood [16:03] lazyPower, marcoceppi_ : and does bundletester respect $JUJU_HOME [16:03] ? [16:03] bloodearnest: it should [16:05] pipsi install bundletester works nicely, much better output [16:06] does it leave the env intact on failure? [16:07] it can be configured to do either/or === roadmr_afk is now known as roadmr [17:09] is block-storage-broker okay to use in trusty? [17:10] I've been using trusty and now want to add storage for my postgresql relation [17:10] but I see block-storage-broker is in precise. but maybe my searchfu is weeak? [17:12] s/relation/service [17:14] are there docs I can look at on how to upgrade a charm? how much work would it be for a naive user? [17:14] can I do it in a day? [17:20] * skay talks to someone about it [17:20] for a workaround I'll collect the precise repo and tell mojo to think it is trusty [17:20] so cowboy. much irresponsible [17:35] skay: the upgrade work is kind of dependent on how its installing the package. ergo: if the package exists in both trusty/precise repo's it should be a fairly short winded porting process, ensuring there are tests in the charm. [17:36] if the repo doesn't exist that metric can go up by an order of magnitude. [17:36] IIRC we only have storage-broker in precise due to the libs it's consuming - it's not using the AWS SDK last time i checked, it was using the Euca2oolset [17:36] lazyPower: I don't have time to look in to it today or over the weekend. maybe I can look in to it next week. it might be a good exercise in going through the process so that I can participate more in reviews and so on [17:36] whit: cf-weekly? [17:36] certainly - we're here to help :) === kadams54 is now known as kadams54-away [19:01] mbruzek1, arosales , in one charm hook, is there a way to get specific information from the deployer (such as hostname, ip)? something like what charmhelpers.core.host does, but for the deployer [19:02] adalbas: Hello. Let me see if I understand your question. Is there a way to get the hostname/ip from a hook? === roadmr is now known as roadmr_afk [19:03] adalbas: hello [19:04] mbruzek1, for instance, i want to add the hostname or ip to /etc/hosts of the machine where i'm deploying a charm in the service [19:06] adalbas: that can be done. What is the use case here? The host computer (that is deploying the services) can already access the VMs in the cloud? [19:08] mbruzek1, yes, but i want the vm in the cloud to have the host hostname, for i'm not using an external dns resolution [19:10] adalbas: I am not sure I follow why you would want to do this but here is how I would get that information. [19:10] juju status etcd/0 | grep public-address | cut -d: -f 2 [19:11] Where "etcd/0" is the deployed charm that you want to find the address for. [19:12] mbruzek1, it is actually the other way around. I'm deploying etcd/0 from my server (juju) and in the install hook, i want to be able to get the information from juju, so etcd/0 would have the ip information of it's deployer. [19:14] so in the hooks/install, i could add something like "echo $deployer_ip $deployer_hostname >> /etc/hosts" [19:15] adalbas: I am not sure how to get that information from the install hook. [19:16] mbruzek1, i see. i think i might need to change my approach here or have something in config.yaml. [19:17] tks [19:17] adalbas: From the deployer you could embed the IP address into the VM like this. [19:17] juju ssh etcd/0 "echo $deployer_ip $deployer_hostname >> /etc/hosts" [19:17] adalbas: I am not sure the deployed vm knows where it came from. Lets say you are deploying from your laptop. [19:18] when you bootstrap you create a Juju server on the cloud that you are using. The juju client just talks to the bootstrap node which does all the provisioning of the other vms [19:18] yes, exactly, i wasn't sure juju had a way to identify who is the deployer. [19:19] So the VMs don't know about your laptop, and I am not sure how they would get that information because the adalbas bootstrap node is doing all the work. [19:20] right. that makes sense. [19:21] Since your laptop can modify the units that sounds like something that you could do from the laptop, I will check to see if the nodes can get that information, but I kind of doubt it. [19:29] * marcoceppi_ reads back [19:30] adalbas: that's not possible withinjuju [19:30] adalbas: if you needed the hostname/ip of the machine that's executing the deploymeht [19:31] you'd have to set it as a configuration value in the charm [19:31] and the person would have to `juju set mycharm my-hostname=$(hostname)` [19:31] marcoceppi_, tks! [19:32] yes, i ll set it on the config === roadmr_afk is now known as roadmr === zz_CyberJacob is now known as CyberJacob [20:11] adalbas, latest jujus tie identity to the services/units in play [20:12] along with apis for share/unshare env which create additional pricnipals [20:12] although perhaps i mistake intent === roadmr is now known as roadmr_afk [20:55] I've got juju bootstrapped to an openstack environment rather than a local one, but when I run protect-new and then workspace-new it attempts to make lxc containers [20:55] I'm trying to figure out what step I missed === roadmr_afk is now known as roadmr === scuttlemonkey is now known as scuttle|afk === scuttle|afk is now known as scuttlemonkey [21:14] oh, I see that mojo creates a local container, but when it gets to running hte manifest juju behaves as I'd expect it to === CyberJacob is now known as zz_CyberJacob === zz_CyberJacob is now known as CyberJacob [22:24] so bundletester doesn't support --upload-tools currently eh?