=== kadams54 is now known as kadams54-away [00:59] jose: there's a lot of variables there - i would take notes for our wednesday call [01:00] ack [01:00] jose: thing is, we may or may not have a good answer there. we run into the same papercuts [01:00] that's fine === axw_ is now known as axw === kadams54 is now known as kadams54-away === thumper is now known as thumper-dinner [09:35] gnuoy: ping [09:35] apuimedo, hi [09:35] Hi [09:36] gnuoy: I'm getting some errors running tests on charm-helpers [09:36] File "/home/celebdor/code/midonet-charms/mem-manager/.venv2/lib/python2.7/site-packages/charmhelpers/core/services/helpers.py", line 157, in __init__ [09:36] super(HttpRelation).__init__(self, *args, **kwargs) [09:36] TypeError: must be type, not HttpRelation [09:37] that super statement does not look py2 nor py3 compliant to me [09:38] apuimedo, it looks ok to me, I haven't been having problems running the unit tests. let me retry [09:39] the old one used to be: [09:39] super(Child, self).__init__(*args, **kwargs) [09:39] the new one is supposed to be [09:40] super().__init__(*args, **kwargs) [09:41] apuimedo, lp:charm-helper tests are passing cleanly for me [09:41] mmm [09:41] python2, 3 or both? [09:42] apuimedo, running the test makefile target which runs both [09:42] in any case, shouldn't it be: super(MysqlRelation).__init__(self, *args, **kwargs) [09:42] if it's the MysqlRelation class? [09:44] apuimedo, I have no idea what is in you charmhelpers/core/services/helpers.py given it seems to be different from trunk so it's hard to say [09:44] * your [09:45] gnuoy: it's retrieved with pip [09:45] automatically with my Makefile [09:46] apuimedo, it seems not to be trunk as trunk tests are passing [09:47] gnuoy: check out the version of pypi [09:47] apuimedo, are you on revno 341 / [09:47] ? [09:47] (after all, it's what's being pulled with the default charm setup.py [09:48] it was 322 [09:48] so that seems old [09:48] apuimedo, I think we may be on cross-purposes somehow [09:48] yes, probably a new pypi release should be done [09:49] gnuoy: or it should be based on charm-helpers/stable [09:50] note it was updated on 17th of this month [09:51] apuimedo, for the openstack charms the charmhelpers dir is populated by branching directly from lp:charm-helpers into the hooks directly [09:52] s/ hooks directly/hooks directory/ [09:52] I know. My concern is that when you do `charm create` and it pregenerates some setup.py that goes to pip for getting the helpers [09:52] it will get 0.2.3 currently [09:52] which has this issue [09:53] so. Should I expect a fix on the pypi version or should I branch stable in my repo? [09:53] apuimedo, ok, that makes sense, I thought your were trying to land a change to charm helpers [09:54] gnuoy: no. Just alerting of a problem in the distributed version [09:54] sorry if I was not clear enough [09:55] apuimedo, no problem at all, thank you for reporting the issue. [09:55] gnuoy: you're welcome [09:56] gnuoy: so I guess it's better that I pin the stable version on my tree [09:56] for now [09:57] apuimedo, I would follow the same path as the other openstack charms. create a charm-helpers-hooks.yaml pointing at lp:charm-helpers , list the modules you need and create a sync make file target [09:57] ok ;-) [10:08] good. It worked fine with the pinning [11:16] rick_h_: done. https://github.com/juju/docs/pull/311 === Spads_ is now known as Spads === Spads_ is now known as Spads === kadams54 is now known as kadams54-away === Spads_ is now known as Spads [14:25] hello guys [14:26] can I run juju over a net that does not have internet connection? [14:26] cos I need juju to deploy some basic services without internet access [14:26] in some subnets [14:31] Muntaner: yes and to do so you will need your own simplestreams services. [14:31] jrwren, I already have them [14:32] jrwren, the problem is that I can't give the floating IPs to those machines [14:32] since they don't have a direct internet access [14:32] (it is a private openstack cloud) [14:33] jrwren, so, I can't connect to the machines, and can't even bootstrap [14:35] Muntaner: I don't know. [14:36] Muntaner: ignoring juju, if you start a nova instance do you have connectivity to it? You'll need that. [14:36] jrwren, yes, but the VM does not have access to the intetnet === kadams54-away is now known as kadams54 [14:53] jamespage, If you have a sec could you take a look at the two tiny mps associated with Bug 1435902 pls? [14:53] Muntaner: internet is irrelevant. connectivity from where you are running juju to the openstack environment is important. [14:53] Bug #1435902: services error deploying to next due to bad mtu setting [14:55] jrwren, since I'm launching the bootstrap on a subnet that has no internet access, if I don't use floating IPs I can't "talk" to the machines. So, the bootstrap hangs at "attempting to connect at ... " [14:57] gnuoy, +1's [14:57] Muntaner: I won't pretend to understand. Sorry :( [14:57] Muntaner, that machine that you're doing the juju deployment from is outside of the openstack cloud ? [14:57] jamespage, thanks [14:57] jrwren, don't worry, I know it's a "strange" configuration :) [14:57] gnuoy, maybe I can explain it better with an image [14:58] gnuoy, http://i59.tinypic.com/2jevkwp.png [14:59] I'm trying to bootstrap juju on the purple subnet [14:59] now I'm giving to that subnet internet access via the router... but in a normal use, I don't want to do that [15:08] can anyone answer a few questions about juju and maas? [15:14] We are looking at juju and maas to solve cm problems in a physical environment. My big question is do I have to dedicate a machine to manage juju for every environment I want to deploy [15:15] for instance we have a project that uses tomcat and mysql on a single machine and another that uses apache, django and mysql on two machines [15:16] I realize that I may have to write the charms to deploy those services in that configuration, but do I need a server dedicated to managing each of those? [15:16] and would I need yet another set of servers to manage the test and development servers? [15:17] if i am getting you correctly you would need to have a several environments and switch between those [15:18] if you are by dedicated machine to manage juju referring to the machine on which juju client will reside [15:18] im not sure about state servers [15:19] yeah that is what I was wondering if I needed a dedicated machine for managing juju [15:19] for each environment [15:20] we have a several environments and each environment has it's own juju state server [15:21] but again in our case it makes sense [15:21] perhaps you could have several containers and in each container state server running and place all containers on a single physical server [15:22] or several vms running on a single physical server [15:23] but again someone more knowledgeable might have better answer :) === natefinch_ is now known as natefinch [15:27] jamespage, thanks [15:28] I am just getting started with all of this so I am not sure what the best way to go about it is. The idea of having state servers running on vm's is an good suggestion, of course that means I would need a more robust maas environment to manage the different networks [15:29] or possibly just use the second NIC in each server to manage them all from a single internal network [15:29] true that [15:30] at this point I am still playing with both maas and juju on some old laptops and virtual box vm's at home with a linksys router running the subnet [15:30] still having DNS issues too [15:31] maas team thinks they will be fixed if I upgrade to 1.7 which is my next task when I get back to the house [15:31] im using both maas and juju in virtual environments [15:31] can't help much with physical environments setup [15:32] have you ever had problems with logging in to a server provisioned by maas? Sometimes it seems like my keys aren't going across with the build even though they are in the maas gui [15:32] nope [15:32] I think I am going to write a blog post about the next trip through the setup... [15:33] but i had similar problems when juju agent is not running or in odd state [15:33] what version of maas are you running now? [15:34] I haven't even installed juju on this go around yet === kadams54 is now known as kadams54-away [17:37] Charmers, here's a list of reviews currently in the queue that need a charmer's attention: http://pastebin.ubuntu.com/10671316/ [17:43] Hi all -- how can I find out the JUJU_UNIT_NAME while doing 'juju run'? [17:49] sinzui, No space left on device [17:50] :( [17:53] dpb1: you should be able to `echo $JUJU_UNIT_NAME` when doing juju-run [17:56] marcoceppi_: no dice. :( nothing JUJU in the environment (I really thought I've used that before, but can't find anything documented). [17:57] dpb1: depending on what you're doing and how patcient you can be, juju actions are in 1.23-beta1 which gives you more structure around juju-run [17:58] dpb1: nothing $JUJU_ in there? That sounds broken. I could have sworn juju run ran in full hook context? [17:59] rick_h_: same. :( echo $JUJU_UNIT_NAME gives nothing, trying 'set, env, compgen -v', all show nothing 'JUJU_' [18:02] dpb1: are you doing juju run against the unit or the machine number? [18:02] marcoceppi_: --all, let me try some variation [18:02] dpb1: ah, that might be it [18:02] dpb1: I know that juju run --machine will just use SSH, while --unit will do it as a hook context [18:02] all may be the same way as --machine [18:02] w.t.h. [18:03] :) [18:03] * dpb1 checks [18:03] --service has the JUJU_ ones [18:04] nice job marco [18:04] marcoceppi_: now, I can confirm that ssh is not involved, but it certainly is different [18:05] I can kinda see why juju run --machine does not run in a hook context [18:05] I guess --all is shorthand for all machines, not all units [18:05] --all (= false) [18:05] run the commands on all the machines [18:05] very interesting [18:06] thx, that helps [18:07] dpb1: yeah, I'm going to raise a bug with core, it shouldn't be too hard to have --machine use the machine-agent like it does the unit-agent [18:07] but you still wouldn't get things like JUJU_UNIT_NAME since if you --to you'd have conflicting values [18:07] marcoceppi_: yes, I can understand. would be nice to have --service 'all' or something [18:07] yeah [18:08] --unit all too [18:08] I guess --all-units or similar [18:08] dpb1: yeah [18:09] marcoceppi_: if you have that bug, let me know, I'll subscribe the appropriate people === kadams54-away is now known as kadams54 [18:09] (when you have...) [18:11] dpb1: https://bugs.launchpad.net/juju-core/+bug/1435999 [18:11] Bug #1435999: juju run should have an --all-units option, --all is ambiguous and confusing [18:12] marcoceppi_: danke [18:39] Update someone else's chrams in the juju charm store | http://askubuntu.com/q/600802 === brandon is now known as Guest51625 [19:13] Juan Negron or Tom Haddon | May I ask one of you a quick question regarding Haproxy and SSL === Guest51625 is now known as web === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === weblife is now known as web [20:07] marcoceppi_: https://code.launchpad.net/~paulgear/charms/precise/nrpe-external-master/fix-duplicate-rsync-entries/+merge/250248 === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 [20:27] * web would greatly appreciate some insight about what I may be missing when trying to use SSL based node.js application with Haproxy. I am using the following hooks ( http://paste.ubuntu.com/10672089/ ) and cant understand the error coming from the reverse server ( http://paste.ubuntu.com/10672447/ ). GUI shot: http://ibin.co/1vxjz6hLgSMJ [20:29] jose: did you see? :) [20:29] wat? wat?! [20:30] oops [20:30] sorry [20:30] lp|sprint: what's going on?! [20:30] jose: http://review.juju.solutions/ - look at those numbers [20:31] lp|sprint: you. all. are. awesome. [20:32] it was all aisrael, cory_fu, and marcoceppi_ [20:32] kudos to them, great job guys! [20:32] wow the review dash is nice and organized now and days [20:32] ^5 guys [20:33] web: yeah, we've been working on upping the ante on UX [20:34] lp|sprint: you know that if you guys need me, I'm here to give you a hand [20:34] so, any merges that have approve community and need charmer merging, please pass them on to me and I can take care of them [20:35] jcastro: https://code.launchpad.net/~jorge/charms/trusty/kibana/update-defaults-to-4/+merge/250379 [20:37] web: i'm trying ot pull up your pastebins but it looks like our network is tanking here at the sprint :( [20:41] lp|sprint: Thank you I appreciate it. No rush I have a feeling I am off on my configuration parameters. I also just notices a note about Haproxy on Tusky. Gonna try a precise based charm for now. I have set haproxy up plenty but never with SSL. [20:42] trusky [20:42] IIRC ssl with haproxy was touchy with the charm atm. aisrael has some experience there i think. [20:42] It seems to be stuck in an infinite loop of some sort because the connection doen't timeout [20:43] lp|sprint: that was rabbitmq. I haven't touched haproxy today [20:43] aisrael: i was thinking back w/ the reddit bundle. weren't you working with haproxy as the reverse proxy server? [20:44] I sent Juan and Tom a message also since their the maintainers [20:45] lp|sprint: yeah, that's been an issue wrt complex configurations, but I think a merge just landed that may help with that. [21:14] aisrael: Thanks for the heads up, there are a couple relating to ssl for merging. https://code.launchpad.net/charms/trusty/+source/haproxy [21:21] stokachu, openstack-install is basically a wrapper around juju-deployer, correct? [21:21] mwenning: uh no [21:21] ok. [21:21] stokachu, elucidate please [21:22] juju-deployer is only used with landscape autopilot as thats what they use to deploy [21:22] otherwise is our own stuff [21:23] seems all the newest ssl merges were made. Not sure if I caught the merge or not. Cross fingers and tries again. [21:24] 6 hours is what I mean by just merged [21:24] stokachu, thx. [22:25] SSL Node.js Charms with Haproxy | http://askubuntu.com/q/600880 === brandon is now known as web [23:06] * web having trouble. Feels he is overlooking something. http://askubuntu.com/q/600880/88546 [23:07] Isn't the room bot supposed to announce new questions??? [23:08] web: it did, about a half hour ago :) [23:08] okay must have timed out for a min