=== thumper is now known as thumper-dogwalk === thumper-dogwalk is now known as thumper [02:15] Any ecosystem folks around by chance? Looking for some guidance on getting the 2.x-compatible version of juju-deployer & python-jujuclient up & running [02:20] Alternatively, can anyone point me to an up-to-date packaged version of juju-deployer & python-jujuclient? === menn0 is now known as menn0-afk === rogpeppe1 is now known as rogpeppe [07:23] Can anyone tell why I see a install hook failure here that seems to be juju itself failing with a socket error? http://paste.ubuntu.com/23057586/ === frankban|afk is now known as frankban === frankban is now known as frankban|afk === menn0-afk is now known as menn0 === frankban|afk is now known as frankban [10:05] jamespage: could you look to my question in review https://review.openstack.org/#/c/348336/6/metadata.yaml ? [10:14] hi everybody, i try to use layers in my juju charm. as i understand i need to execute `charm build` and publish result. but in this case result charm is pretty complicated(it contains code from all layers). is it possible to publish only my own layer? === danilos` is now known as danilos [13:21] gennadiy - sure, you can checkin your layer to git. If you've built a middle layer, we have a web-service to warehouse those. However in order to push to the charm store you will need to push your assembled charm. Top layers typically aren't useful without all their supporting layers code. [13:24] @lazyPower, it's clear for me when we use local charms. but i think it can be implemented for charms from store too. juju can run "charm build" during deployment process [13:25] gennadiy - right now charms are built and assembled externally to the deploy process. That sounds like a reasonable feature reuqest to bring up on the list or the issue tracker. [13:27] yesterday i tried to create custom charm with docker layer. i have added small part of code but when i built by charm i was surprised. my charm contains all code from all layers [13:27] gennadiy - anything you dont want from a lower layer can be ignore'd or overridden with a file in your top layer. [13:28] there is an ignore keyword in layer.yaml that should help you sort out what you dont want to bring forward [13:29] https://github.com/juju/charm-tools/issues/214 <- some info here [13:30] and additional info here: https://github.com/juju/charm-tools/issues/220 [13:30] stokachu: can you check this one out? http://askubuntu.com/questions/811389/openstack-install-fails-because-of-wrongly-specified-ubuntu-version [13:37] jcastro: done [13:37] thanks! [14:03] hello.. How can i unblock specific services [14:03] some have the "workload/status" blocked [14:26] BlackDex: so they tend to be blocked for some reason. for instance a web app might be blocked until it's realted to a database for the back end [14:26] BlackDex: so it's dependent upoon the application you're working with at the moment [14:28] Can any one please check this bug : https://bugs.launchpad.net/juju-core/+bug/1602572 [14:28] Bug #1602572: Handler function is not being called even after changed one of the config values of the config.yaml [14:39] geetha: ty, asked cory_fu to take a peek when he gets a chance. [14:52] Please...it's stopping me to push my charm. [14:52] thanks rick_h_ [15:14] Is there a good mechanism to install containerized openstack (ie docker) via juju? Has anyone seen a write up? [15:18] Hi all [15:22] valeech: there's a lxd deployment notes here: https://github.com/openstack-charmers/openstack-on-lxd [15:24] rick_h_: thanks! [15:28] anyone here who could help me with setting up glance with swift? Everything seems to work fine but when creating an image in glance a 500 internal server error is thrown: http://paste.ubuntu.com/23058730/ [15:29] I checked the configuration of glance and swift_store_key is present in /etc/glance/glance-api.conf [15:33] try restart glance-api [15:34] but i don't know what happens [15:34] if you relate glance to swift then all should works [15:35] it can be dependent from OpenStack version [15:35] andrey-mp, I have swift-proxy related to glance, keystone and swift-storage [15:35] ok, but I think it worked! [15:36] I've never configured glance to swift [15:36] only to cinder ) [15:37] what version are you use? [15:38] andrey-mp, trusty/swift-proxy-54, trusty/glance-251 not sure what versions those are [15:38] i mean version of OpenStack [15:40] how can I tell? [15:41] dpkg -s python-cinder [15:41] on node with cinder [15:41] or check config parameter of any openstack charm - there should be version [15:44] Is that what you're looking for? 1:2014.1.5-0ubuntu2 this if from dpkg -s python-cinder. I couldn't find any information on the charms [15:51] yeah [15:53] this is IceHouse [15:53] very old version ) [15:53] Well it's the current Trusty version of the charms [15:54] are you really want to use this version? i would try to use newer first... [15:56] version can be defined with paramter "openstack-origin": "cloud:trusty-mitaka" [15:57] Ok, I will give it a try. [15:58] looks like it is a bug in glance charm [15:58] or in the glance ) [15:59] anyway - IceHouse version is not supported now... [16:02] Ok, I'll try with a newer version :) [16:03] juju is not stable on Xenial yet, is it? === frankban is now known as frankban|afk [16:13] MonsieurBon: not juju 2.0, juju-1 is [16:14] rick_h_, that's what I meant :) [16:22] how's it going everyone? [16:23] I'm trying to write a consul-base layer to be consumed by my consul-agent and consul-server charms here -> https://github.com/jamesbeedy/layer-consul-base [16:27] having some trouble understanding how I might have the consul agent interface to the consul server using 'raft' relation stub .... it seems its service name sensitive though, so I'm having a tough time understanding how I might join services with a different name e.g. 'consul-agent' to the consul-server cluster ... [16:27] any ideas? [16:32] petevg: wasn't there an issue where zk would not honor 0.0.0.0 for clientPortAddr? is this checking that the value gets changed, or do we expect zk to actually re-bind to all interfaces? https://github.com/juju-solutions/bigtop/pull/38/files#diff-f5b8005c1652359285aac30975c6290bR67 [16:32] kwmonroe: Yep. Forgot to drop in a comment. I'm skipping the test because Zookeeper is broken. [16:32] Will drop one in ... [16:33] ahhh, roger that petevg. i recall you saying that wasn't valid until upstream zk handles it. my bad. [16:33] .. and i see the 'return' right up in my face on that test case now [16:34] bdx: how is it being 'service name sensisitve'? is something looking for an explicit 'consul-server' string? [16:36] kwmonroe: I decorated that test with unittest.skip, with appropriate note. Thx for the catch. [16:37] nice petevg! i didn't know about unittest.skip. TIL. [16:39] kwmonroe: when I `juju deploy -n 3 consul`, the consul service uses raft protocol to facilitate determining the cluster creation, state and member states ... etc ... I guess I'm wondering how the consul-agents would use the raft stub to join the cluster too [16:41] your problem here, bdx, is that i don't know how the current consul stuffs work ;) i'll poke around and see if i can offer a suggestion. [16:45] kwmonroe: awwwe shucks .... I'm might be mistaken by thinking that the consul agents need to participate in raft -> https://postimg.org/image/5rdlif72n/ [16:45] kwmonroe: np, thanks [17:43] thanks for swapping that meeting lazyPower [17:43] magicaltrout - np sorry for the scheduling conflict :) [17:43] they have to pry my protective contacts out [17:43] the sooner the better [17:46] was just telling jcastro I got asked to present at mesos usergroup london as well lazyPower [17:46] Nice, so you're going to be the traveling engineer on all this :D [17:46] i got pinged by marketing this morning about the topic page. Did you wind up syncing with someone over that? i got disconnected from that whole process over the last few weeks [17:47] hmm, we should have given you shirts when you were in london [17:47] i did have a big long chat with someone about it off the back of your email [17:47] i'm not sure what happened after that [20:49] kwmonroe: I just updated https://github.com/juju-solutions/layer-apache-bigtop-base/pull/39 -- it had a stray reference to charms.unit, which isn't release yet (I changed the reference back to point at bigtop_harness). [20:50] cool petevg === natefinch is now known as natefinch-afk [21:00] question on subordinate charms/relations ... [21:02] so I'm working with the consul service, I have a need for consul-server and consul-agent server charms [21:02] blah .. consul-server and consul-agent charms * [21:07] the problem I'm having is this [21:10] how can I make a subordinate charm install software on a primary when related to one service, and just do something else when related to another service (not install software on primary) [21:11] I've been looking at cinder-ceph, and nrpe as examples [21:13] oooh, I think I see .... you have to use 'scope: container' [21:45] hello everyone.. [21:46] how to bootstrap in openstack environment with JUJU 2.0? [21:48] mayurisapre: is it an openstack setup for you? or something you've setup? [21:48] i have openstack env [21:49] i just confused with all add-credential, config.yaml, credential.yaml, cloud.yaml files [21:50] i am trying to add openstack cloud with add-credential command with credential.yaml file [21:50] it is giving me an error [21:50] mayurisapre: https://jujucharms.com/docs/stable/clouds#specifying-additional-clouds is the user end side [23:50] Can anyone tell why I see a install hook failure here that seems to be juju itself failing with a socket error? http://paste.ubuntu.com/23057586/ [23:50] If I try to debug-hooks, I see "failed to connect to server" [23:50] it drops me in a session, but the debug-hooks session is never started when resolved --retrying