UbuntuIRC / 2014 /11 /25 /#juju.txt
niansa
Initial commit
4aa5fce
=== CyberJacob is now known as CyberJacob|Away
=== kadams54 is now known as kadams54-away
=== kadams54 is now known as kadams54-away
=== kadams54-away is now known as kadams54
[01:50] <hatch> Has anyone asked for a config field in which there are a select few options which users could choose from? Which could be represented in the GUI as a select/dropdown?
[01:52] <rick_h_> hatch: yea, enums basically
[01:52] <rick_h_> hatch: it's part of the discussion around using jsonschema to represent config options
=== kadams54 is now known as kadams54-away
=== kadams54-away is now known as kadams54
[01:52] <hatch> ok - atm I am just going to give them the possible options in the description but that just feels ripe for issues
[01:52] <hatch> typos and whatnot
[01:55] <rick_h_> hatch: yea
[01:55] <hatch> it would be even better if those enum options could enable/disable subsequent config options :)
[02:33] <hatch> when running debug-hooks on a unit the tmux session that opens is in the home directory of the unit not the charm as shown in the docs - where can I find the actual hooks?
[02:34] <hatch> $CHARM_DIR is blank
[02:37] <hatch> ahah I need to actually trigger the hook externally
[02:37] <hatch> looks like we need to update these docs :)
=== _thumper_ is now known as thumper
=== scuttlemonkey is now known as scuttle|afk
[04:04] <noodles775> hatch: Yeah - it could be clearer that none of the juju env is available until you're in a hook context (ie. config-get and other juju binaries)
=== erkules_ is now known as erkules
[04:17] <hatch> yeah just a small wording change will help with that - I'll try and remember to submit a PR for that this week
=== kadams54 is now known as kadams54-away
=== CyberJacob|Away is now known as CyberJacob
=== CyberJacob is now known as CyberJacob|Away
=== CyberJacob|Away is now known as CyberJacob
=== CyberJacob is now known as CyberJacob|Away
=== psivaa-holiday is now known as psivaa
[13:38] <gnuoy> Am I missing something or is charmhelpers now broken on trusty since the py3 support mp?
[13:41] * gnuoy goes to investigate further
[14:14] <marcoceppi> gnuoy: there's a bug already
[14:14] <marcoceppi> gnuoy: https://bugs.launchpad.net/charm-helpers/+bug/1395378
[14:14] <gnuoy> marcoceppi, ah, ok. thanks
[14:14] <mup> Bug #1395378: latest python3 additions seem to break with trusty python-six <Charm Helpers:In Progress by stub> <https://launchpad.net/bugs/1395378>
[14:15] <stub> gnuoy: If you want, check out if lp:~stub/charmhelpers/py3-2 sorts things for you
[14:16] <gnuoy> stub, thanks
[14:18] <stub> I can't get the venvs built in a precise VM :-(
[14:22] <mhall119> marcoceppi: does juju work against Canonistack? I was told it did, I can bootstrap on it but `juju status` never returns
[14:22] <marcoceppi> mhall119: it does
[14:23] <marcoceppi> lazyPower may have some insight on this
[14:23] <marcoceppi> mhall119: you have do to like proxies and shuttle and stuff
[14:23] <lazyPower> mhall119: do you have your VPN setup?
[14:23] <lazyPower> you wont get anything back from juju status util you've activated your VPN or canonistack-sshuttle - i highly recommend the VPN route
[14:25] <mhall119> lazyPower: no, I don't
[14:32] <stub> gnuoy: did that work btw? I'd like to know what series you deployed too.
[14:33] <gnuoy> stub, I haven't tried as yet tbh
[15:09] <gnuoy> jamespage, got a sec for https://code.launchpad.net/~gnuoy/charm-helpers/haproxy-singlenode-mode/+merge/242790 ?
[15:11] <jamespage> gnuoy, looks reasonable
[15:11] <jamespage> is this for the 'run haproxy always' stuff?
[15:12] <gnuoy> jamespage, it is
[15:17] <stub> tvansteenburgh: yes, apt-get as an import side effect is disgusting. There are some others hiding in charm-helpers too...
[15:20] <jamespage> gnuoy, +1
[15:35] <gnuoy> jamespage, thanks
[15:36] <jamespage> gnuoy, https://code.launchpad.net/~james-page/charms/trusty/quantum-gateway/stable-vlan-flat-support
[15:36] <jamespage> I'd like to get that landed asap
[15:37] <gnuoy> jamespage, err, I thought I already had. sorry about that
[15:37] <jamespage> gnuoy, you acked and I landed next
[15:37] <jamespage> thats for stable
[15:37] <gnuoy> jamespage, +1
[15:38] <jamespage> gnuoy, thanks
=== scuttle|afk is now known as scuttlemonkey
=== scuttlemonkey is now known as scuttle|afk
=== scuttle|afk is now known as scuttlemonkey
=== kadams54 is now known as kadams54-away
[16:42] <sayon> hey there, i am wondering if i can upgrade from juju 0.7 to the newer juju-core, without having to re-bootstrap the whole environment again?
[16:45] <marcoceppi> sayon: that's a great question
[16:47] <marcoceppi> sayon: as you may be aware a lot has changed in the latest releases of juju, let me find out if there is a migration path
[16:47] <sayon> marcoceppi: i know, right :)
[16:47] <lazyPower> oh wow 0.7 to 1.20.x? woo thats a huge update.
[16:48] <sayon> lazyPower: yea, i tried to upgrade once juju-gui was introduced but i could not figure out how to do it without rebootstrapping
[16:49] <marcoceppi> sayon: so we've since introduced a "juju upgrade-juju" command but I don't think that will work on an environment that's as old as 0.7
[16:49] <hazmat> sayon, there's isn't anything supported from pyjuju/zookeeper to juju-core/mongodb
[16:50] <sayon> marcoceppi: it won't i already tried, there is no upgrade-juju command :(
[16:50] <sayon> hazmat: i thought so
[16:51] <hazmat> sayon, capturing the topology from juju status > and capturing the data in env and then rebootstrapping and restoring by hand .. its a pretty manual process.. juju-core does support in place upgrades.
=== kadams54-away is now known as kadams54
=== kadams54 is now known as kadams54-away
[16:51] <sayon> i am running in alot of trouble since i upgraded my maas installation on ubuntu 12.04
[16:52] <hazmat> sayon, doh..
[16:52] <hazmat> sayon, was that recently?
[16:53] <sayon> hazmat: it has been some month i guess, everything worked just fine till i wanted to add another machine
[16:53] <hazmat> sayon, there was a security fix that got pushed for maas in the last few weeks that broke pyjuju .. pyjuju has been eol'd for a few years.
[16:53] <sayon> hazmat i guess thats what i am facing right now
[16:53] <hazmat> sayon, if you want to keep the env, i'd recommending apt-pinning maas to the older version
[16:54] <hazmat> the security fix was around the object storage capabilities of maas, and is only relevant if maas was exposed directly to untrusted users which is not common.
[16:54] <sayon> hazmat, tried that already but i am running into problem since the database structure seems to have changed alot
[16:55] <hazmat> ugh
[16:55] <sayon> when downgrading it fails to convert from the newer to the older database version
[16:56] * hazmat tries to dig up details
[17:00] <hazmat> sayon, do you have the error when adding a new machine to the environment?
[17:00] <hazmat> ie from log
[17:00] <hazmat> it looks like the change was specifically around how charm urls are handed to units
[17:01] <hazmat> sayon, also what version of maas is the newer one?
[17:01] <sayon> well when i add a new machine it gets commisioned in maas and is the ready
[17:01] <sayon> but "juju status" won't list the new machine and may not acuire it for charm installations
[17:02] <sayon> hazmat: its maas 1.2+bzr1373+dfsg-0ubuntu1~12.04.5
[17:06] <hazmat> sayon, interesting.. thats a different issue.. the commit i was referencing hasn't landed in 1.2 series yet... do you know what the other version of maas was.. apt-cache policy maas should have it
[17:07] <sayon> hazmat: wow, seems like it was 0.1+bzr482+dfsg-0ubuntu1
[17:11] <hazmat> sayon, hmmm.. so nutshell i'd recommend a bootstrapping that env again using the juju stable ppa on precise https://launchpad.net/~juju/+archive/ubuntu/stable?field.series_filter=precise
[17:11] <hazmat> juju-core can self-update itself..
[17:12] <hazmat> i'm asking on #maas re best version/ppa for precise
[17:12] <sayon> hazmat: that means reinstallation of all machines used with juju charms, right?
[17:13] <hazmat> sayon, sadly yes
[17:14] <hazmat> that version of maas was ancient, and pyjuju has been eol'd for a while
[17:15] <sayon> hazmat: ok, i think i will then do a complete reinstallation with the new LTS version of ubuntu
[17:15] <sayon> and not stick with precise for any longer
[17:16] <sayon> but thanks alot for your help and support hazmat and marcoceppi!
[17:18] <hazmat> np
=== scuttlemonkey is now known as scuttle|afk
=== scuttle|afk is now known as scuttlemonkey
=== scuttlemonkey is now known as scuttle|afk
=== scuttle|afk is now known as scuttlemonkey
[19:03] <skay> avoine: I'd like to be able to run collectstatic, any objection to the idea of having it happen during install and upgrade?
[19:04] <skay> avoine: also, there is the ansible branch you have, and I'm wondering if most all of hte python code in hooks.py is going away and it will turn in to a lightweight shim around ansible calls?
[19:05] <avoine> skay: The idea was to use a subordinate for that: http://bazaar.launchpad.net/~patrick-hetu/+junk/django-contrib-staticfiles/files
[19:07] <avoine> skay: also Simon proposed to use dj-static: https://code.launchpad.net/~bloodearnest/charms/precise/python-django/trunk/+merge/235430
[19:07] <avoine> skay: yes, migrating to Ansible is the long term plan
[19:09] <skay> for that dj-static branch -- I would leave whether or not someone uses dj-static up to the project
[19:09] <skay> but the dollectstatic command still needs to be called, and if it should be handled as a subordinate, maybe document it in the readme?
[19:10] <skay> I happen to be using dj-static right now, and I handle the dependencies in my project's requirements file
[19:10] <skay> so python-django installs it for me accordingly
[19:11] <skay> on another note, when find_django_admin_cmd fails to find the admin, would you want sys.exit(1) there?
[19:12] <skay> I'm not clear on when things should fail silently or not, but that one seems like maybe it shouldn't fail silently
[19:14] <avoine> skay: can you give me the line number?
[19:15] <avoine> skay: Simon's patch would be better since it is not recommended to use django.contrib.staticfiles in production
[19:18] <skay> avoine: http://bazaar.launchpad.net/~patrick-hetu/charms/precise/python-django/pure-python/view/head:/hooks/hooks.py#L206
=== rcj is now known as Guest93149
[19:22] <avoine> I think I did that to catch the cases where django is not installed
[19:22] <avoine> skay: do you have a case where this should not append?
[19:23] <skay> avoine: I'll try to dig up what was happening to me. somehow the method was returning None and then the charm was attemping to call None syncdb etc.
[19:24] <avoine> ouch
[19:24] <skay> avoine: I'm not sure how much you want to dig in to it. I was working on a branch of this to allow for pip extra args and was doing --no-index --file-files and had hte wrong version of something, if memory serves
[19:24] <skay> avoine: so it didn't install django properly? I thought perhaps that should put the unit in to an error state earlier than it did
[19:24] <skay> it ended up looking like an error with the pgsql relation rather than earlier
[19:25] <skay> so, the method allows for the django-admin command not to be found, and doesn't fail. perhaps it should
[19:25] <avoine> skay: no because you could have installed django from a debian package so I'm not checking if it exist earlier
[19:26] <skay> avoine: oh that makes sense.
[19:26] <skay> I figured there was a reason
[19:26] <avoine> skay: yeah the pgsql will also fail if django is not installed
[19:26] <avoine> *relation
=== roadmr is now known as roadmr_afk
=== rcj is now known as Guest31019
=== roadmr_afk is now known as roadmr
=== scuttlemonkey is now known as scuttle|afk
=== fuzzy_ is now known as Fuzai
=== Guest31019 is now known as rcj
=== rcj is now known as Guest68049