[02:07] <rick_h_> hm, so my charm installs build-essential but failing today with #1247299 so wondering if others hit this and wondering if this then makes it a juju issue since lxc doesn't consider it one?
[02:07] <_mup_> Bug #1247299: apparmor blocks cgroup-lite from mounting <cgroup-lite (Ubuntu):Invalid> <https://launchpad.net/bugs/1247299>
[05:13] <Azendale> So, I'm trying to "juju ssh" to a LXC container, but it is just trying to connect to the host computer (which doesn't have the ubuntu user, and therefore the key is rejected). Is there a way around this problem?
[05:15] <Azendale> juju version is 1.16.2-saucy-amd64
[20:50] <Azendale> https://bugs.launchpad.net/charms/+bug/1247636 would someone be able to look at this bug and tell me if my fix to it is a good/bad idea? Should that function even be being called?
[20:50] <_mup_> Bug #1247636: etherpad lite fails to deploy, install hook running get relation <Juju Charms Collection:New> <https://launchpad.net/bugs/1247636>
[22:52] <marcoceppi> Azendale: you can use juju ssh <unit> btw, juju ssh <machine> doesn't work with LXC
[22:53] <marcoceppi> Azendale: also, charmhelpers is part of lp:charmhelpers so you might want to consider patching that there instead
[22:54] <Azendale> marcoceppi: Thanks. (I'm trying to remember what I asked, but I think I was having trouble with getting LXC machines to start. But I got LXC to work eventually, and it was really nice. Fast, too, compared to installing an OS in a VM).
[22:55] <Azendale> marcoceppi: The change I made on the bug I filed was just to get it to work. But I'm really new to charm stuff (but used to python), so I don't know exactly what is the bug.
[22:55] <marcoceppi> Azendale: well, you can't (and shouldn't) be able to call relation-get from install hook
[22:56] <Azendale> marcoceppi: I'm not sure if that function should not be called in the first place in an install hook, or if the function should just return None
[22:56] <marcoceppi> so it seems the charm is doing something wrong
[22:57] <marcoceppi> Azendale: but there's nothing wrong with building in guards. So charmhelpers is a set of scripts designed to solve common problems that charm authors have. One problem is making writing charms in python easier. So that's why you have decorators and methods in python that streamline communication to juju commands
[22:57] <Azendale> marcoceppi: So, you're saying maybe fix the bug in two places, and then it's less likely to come up again in some other charm
[22:59] <marcoceppi> Azendale: the real problem is in the install hook, it shouldn't be calling that, and if it is, it needs to guard when results aren't available
[22:59] <marcoceppi> Azendale: the change to charmhelpers is a nice addition, not really a bug fix per se
[22:59] <marcoceppi> Azendale: but still
[23:01] <marcoceppi> Azendale: actually, having it fail like this when a relation_id isn't found is probably a good thing. Silently continuing when the method is called out of place could lead to weird results
[23:02] <Azendale> marcoceppi: ok, thanks for the advice. I'll probably see what I can find as far as a proper fix then (sorry I'm so non-committal, I'm just not sure when/if I'll have time to try to understand the charm hook and see where to disable getting the relation)
[23:04] <marcoceppi> Azendale: that's fine, I'm happy to help get you on the path to committing!
[23:15] <Azendale> marcoceppi: Thanks, I appreciate it