[04:27] marcoceppi: juju-log issues are covered by https://bugs.launchpad.net/juju-core/+bug/1274460 [04:27] Bug #1274460: juju-log vs. command line length limits

[04:27] (and there are similar bugs for the other tools, which share the problems, apart from relation-get which has been fixed already) === JoshStrobl is now known as JoshStrobl|AFK === CyberJacob is now known as zz_CyberJacob [10:24] Good morning [10:24] Does somebody here use the juju digital ocean plugin? [10:25] https://github.com/kapilt/juju-digitalocean [10:25] I saw some comment from lazyPower on one of the issues [10:25] for some reason `juju ssh` does not work in my digital ocean environment [10:26] apuimedo: what does it do ? [10:36] axino: from poking around [10:37] I saw that it sshs to the bootstrap machine [10:37] and from there does `nc` [10:37] the first part worked, I can ssh to the bootstrap env [10:37] but the nc part would just leave me waiting forever [10:39] apuimedo: can you reach the second instance from the bootstrap node ? [10:40] axino: I can ssh to it [10:40] but I don't know its pass [10:40] axino: are you using it frequently? [10:41] apuimedo: I'm not :) [10:43] axino: what do you use, an OSt env? [10:44] apuimedo: yes, OS and a bit of AWS [10:44] apuimedo: can you run "juju --debug ssh $whatever_you're_sshing_to ? [10:45] axino: how well does the AWS one work for deploying OSt? [10:46] apuimedo: I'm sorry, I don't get your question [10:48] axino: I mean, does the AWS environment work well for deploying OpenStack charms? [10:50] apuimedo: I never deployed OpenStack charms to AWS [10:50] ah, ok [10:51] jamespage: have you? [10:58] gnuoy: I guess you use the internal OSt to launch openstack charms, right? [10:59] apuimedo, ost? [11:00] OpenStack [11:01] apuimedo, we use ost on ost and then ost on baremetal [11:01] cool === JoshStrobl|AFK is now known as JoshStrobl [14:27] jcastro: Ubucon LatAm almost starting! [14:28] glad to see you made it! [14:28] people were, uhhh, worried you didn't make the flight, heh [14:28] yeah!! i were looking after the ride but I didn [14:29] didn't find any sign with my name [14:29] don't know what happen :P [14:29] but hey! I'm here in one pice [14:29] jose: is making an awesome work :) === jaywink_ is now known as jaywink [14:50] jamespage: hey. frankban just released new deployer and also new juju-gui, which now allows you to deploy v4 bundles (so the new format you've been working with Makyo). Bundle deploy almost works, fails due to hook/config problem, no longer machine placement. [14:51] jamespage: and v3 -> v4 translation has landed, but that requires new charmstore release, which should happen next week [14:52] urulama, jamespage: also quickstart fixes to deploy v4 bundles with placement will land next week, for now the deployer should work, v0.5.1 released on Pypi and the juju stable PPA [16:28] marcoceppi: ping [16:45] someone is having problems with python dependencies? http://pastebin.com/qX0c98F0 [16:46] I have almost the same problem with every charm deploy [16:46] help? [16:55] http://pastebin.com/qX0c98F0 help ? [16:55] sebas5384: looking [16:55] marcoceppi: thanks o/ [16:56] sebas5384: can you ssh into the machine and run this: [16:56] sudo apt-get -y install python-apt python-launchpadlib python-tempita python-yaml [16:56] sebas5384: apt-get is having a problem installing one of those packages [16:57] yes [16:57] let me see [16:57] hmmm yeah couldn't even download the mysql for the mysql charm [16:59] http://pastebin.com/DaztURCf [17:00] marcoceppi: ^ [17:00] it seems there's definitvely an apt problem [17:00] sebas5384: well, there ya go. Looks like you've got a proxy or networking issue [17:00] is this on local provider or a cloud? [17:00] 403 Forbidden [17:00] local [17:00] somethign is up with the networkign wherever you are [17:01] the archive is reachable here [17:01] maybe is something related to firewalls [17:01] yep definitively is that [17:01] marcoceppi: I'm going to do a demo tomorrow in the ubuconLA [17:01] and I wanted to show the local provider [17:02] sebas5384: is this at the conference's wifi? or the hotel? [17:02] or somewhere else? [17:02] conference wifi [17:02] eek [17:02] i think i must deploy locally in the hotel [17:02] jose: if you're around (you're probably mad busy) can you help sort this? [17:03] marcoceppi: jose is at the stage in this very moment [17:03] hehe [17:03] i'll talk with him [17:03] sebas5384: haha, well he should be able to help hopefully [17:03] definitively! [17:03] thanks for your help marcoceppi :) [17:03] marcoceppi++ [17:04] we should have some karma bots here :P [20:12] ping wolsen [20:12] pong beisner [20:12] wolsen, hey so in kilo and later, rmq doesn't allow guest amqp connections unless coming from localhost [20:13] wolsen, so i've got tests that do all sorts of cool validation ... for combos up to kilo. [20:13] beisner, mmm was it just kilo? I thought that was a change in rmq version 3.something iirc [20:13] beisner, ahh right [20:13] wolsen, then i have a work-around in the tests to just force-enable guest from anywhere at kilo and later. [20:13] wolsen, but... [20:13] wolsen, that clobbers ssl configs, since it's all in the same funky lil erlang stanza [20:14] wolsen, ref: https://www.rabbitmq.com/access-control.html [20:14] wolsen, tldr: the rmq charm needs to grow a config option to set loopback_users to [], while preserving all of the mangling happening for ssl configs in that file. [20:15] wolsen, in order for me to include ssl tests that is. [20:15] beisner, what about granting access for a specific user? oh those are made over the relations right? [20:15] wolsen, it looks to me like that file is only currently touched by the charm when ssl is on. [20:15] beisner, oh [20:15] wolsen, yep, rock:hardplace [20:16] beisner, yep... and enabling the option risks security :/ [20:16] wolsen, right, but the charm is untestable (across our whole support matrix) without either adding that option [20:16] or [20:17] doing more serious mangling of configs in the test, outside of charm config options [20:17] which i try not to have to do [20:17] beisner, that's fair [20:18] wolsen, the new test is this: [20:18] beisner, fwiw, I'm not opposed to adding the option, just need to have the appropriate amount of "do not use this in the real world" [20:18] deploy precise-icehouse through vivid-kilo [20:18] 3 rmq units [20:18] send amqp msg to each unit, 1 at a time, and check that the message can be read from all other units [20:19] check that all units are represented IN all units cluster_status [20:19] so it's more of a core cluster/messaging functional check that works on the whole horizon of combos (sans ssl). [20:20] i have a whole load of helpers coming out of this too, which will make it easy to add the other scenarios [20:21] wolsen, i wonder if we can borrow creds from something else that does have perms to talk over the wire? [20:21] and use that in pika instead of guest [20:22] beisner, should would be nice to be able to set a mock relationship here - so you could create it that way! [20:23] beisner, we probably could borrow them if you really wanted to - seems like quite a kluge though [20:23] beisner, I think its probably better to enable the guest account for what is being tested [20:24] wolsen, i was just running with the guest@guest pika thing the other tests were using. i'll look at adding a separate user explicitly who can talk. but if that involves that funky stanza thing, we'll still be clobbering ssl charm activities. [20:24] wolsen, the charm isn't plumbed for that file currently, except for in the ssl enablement pieces, so it doesn't look like that will be non-trivial. [20:25] s/non-trivial/trivial/ bah the double negatives [20:25] ie. the ssl charm config options are a bit antisocial wrt the file we need to touch [20:43] marcoceppi: the mediawiki scalable bundle is giving errors in the haproxy and mediawiki install hooks http://pastebin.com/8gXkdhKz [20:43] deploying in GCE [20:43] I'm gonna test it in aws later [20:45] sebas5384: It looks like that ppa doesn't exist anymore [20:45] hmmm [20:46] so the charm is not updated? [20:46] maybe because is precise/ [21:13] wolsen, got it sorted. the test is just adding a testuser with necessary privs, should work independently of that file and the ssl erlangy stuff. [21:13] beisner, sweet - that's even better === JoshStrobl is now known as JoshStrobl|zzz [21:20] tvansteenburgh: was there any clarity on how frankban's branch fixed the memory issue? [21:30] * beisner => eow [21:43] hi juju team, how do I set proxy's on bootstrap nodes? I tried 'http_proxy: blah' in environments.yaml and it said "unknown config" [22:12] mwenning: what version of juju? [22:12] hazmat: tvansteenburgh is out for the next week [22:13] marcoceppi, 1.24.3-trusty-amd64 [22:13] mwenning: try http-proxy [22:13] and https-proxy [22:13] mwenning: you can also set it on an exiting environment with juju set-env [22:18] marcoceppi, thx . in 30 minutes I'll know if that fixed it ;-) [22:21] marcoceppi, is there a way to tell juju to just abort everything and quit? [22:23] mwenning: ctrl + c? [22:23] marcoceppi, nope, juju says something like 'waiting for bootstrap to finish' and keep right on going until all the timeouts finish. [22:24] mwenning: you could maybe do juju destroy-environment --force [22:24] marcoceppi, nope, juju ignores that if it's in the middle of a bootstrap [22:25] \o. [22:34] \o/ , Bootstrapping Juju machine agent [22:41] mwenning: huzzah!