[06:29] <bigjools> jtv: can I make a suggestion about your tmpdir fix
[06:30] <bigjools> can we put a base name in the tmpdir so we know what it's there for
[06:30] <jtv> Sure.  There was one before, but I forgot.
[06:31] <bigjools> yeah
[06:31] <bigjools> I just realised it was there before, hadn't got that far
[06:32] <bigjools> oh for side-by-side diffs in LP
[09:21] <bigjools> jtv: can I archive off your "done" cards on kanban?
[09:21] <jtv> bigjools: yes please.
[09:21] <bigjools> done.
[09:42] <rvba> bigjools: allenap: jtv: gmb: I just filed https://bugs.launchpad.net/maas/+bug/1366726 about the nodes not getting a static IP address.
[09:49] <jtv> Looking.
[09:50] <jtv> rvba: you mentioned that there was a race condition, right?
[09:53] <rvba> jtv: well, it seems it doesn't happen all the time.  But once the other CI breakage will be fixed, we will know more about the frequency of the problem.
[09:54] <jtv> So more a placeholder than a full bug report for now?
[09:55] <rvba> jtv: indeed
[09:55]  * jtv nods
[09:55] <rvba> jtv: I thought it was critical enough to file a bug as early as possible.
[09:57]  * gmb loves finding Blaswell code that’s untested. Because, Giggles!
[10:24] <gmb> allenap: Moar halp with twisted pls. Whyfore is this happening http://pastebin.ubuntu.com/8289294/ with this diff: http://paste.ubuntu.com/8289292/
[10:29] <kickinz1> \o
[10:30] <kickinz1> Hello, how do remove achitectures from maas (to retrieve only i386 and amd64) ?
[10:30] <kickinz1> So import can be faster?
[10:31] <kickinz1> and /var/lib/maas could be smaller (ppa version: 1.6.1+bzr2550+2551+294~ppa0~ubuntu14.04.1)
[10:38] <gmb> allenap: Never mind. Holy duplicated useFixture batman!
[10:43] <kickinz1> using maas $Profile boot-source-selections read $Cluster_uuid 1
[10:44] <kickinz1> had to take cluster_id from webgui, didn't found any other way to get the UUID from cli
[10:44] <kickinz1> and it tells me 'Not Found'.
[10:45] <kickinz1> ok got it sorry,
[10:45] <kickinz1> need to maas $profile boot-source read $cluster_uuid, thne it gives you its "id"
[10:45] <kickinz1> bye
[10:59] <jtv> kickinz1: the CLI is horrible — definitely on the list of things we want to improve.
[11:01] <kickinz1> jtv: thanks for answering, yes cli is not super usable, but I had a conf with arches='*', so near 50Gb of data to download... Maybe it could be insterresting to allow selection of arches within web interface for clusters?
[11:03] <jtv> kickinz1: definitely something we want — our problem is time!
[11:03] <kickinz1> jtv, I understand, not a critic ;)
[11:04] <jtv> No worries. Glad you got this one sorted out.
[11:04] <kickinz1> jtv, first time I used maas was a long time ago, just when it was renamed from Orchestra to Maas, so I can tell that work has been done meanwhile !
[11:05] <jtv> Oo!
[11:05] <jtv> That's a while back.
[11:06] <kickinz1> Yes and replayed with it lately as I changed from company...
[13:17] <gmb> allenap: Halp again… I’m hitting this: http://paste.ubuntu.com/8290409/ with this diff: http://paste.ubuntu.com/8290398/. Now, I know that that’s happening because the function is timing out waiting for a connection to the (Mock) region… BUt I don’t know why *that*’s happening.
[13:17] <allenap> gmb: I’ll take a look.
[13:18] <gmb> allenap: Thank you.
[13:19] <gmb> allenap: In turn, I’ll review your branches.
[13:19] <allenap> gmb: Thanks!
[13:27] <allenap> gmb: Don’t use the Clock() first.
[13:28] <gmb> allenap: Okay, what should I do instead? I only use the Clock() as far as creating the ClusterClientService is concerned, and I only need that once I’m in create_node… Should I just do that there?
[13:28] <allenap> Clock is useful for writing tests without the reactor, where you can control the advancement of time and observe what happens, but it essentially detaches code from the reactor.
[13:28] <gmb> Ahh.
[13:29] <gmb> allenap: So can I just use the reactor in create_node and not have to do anything in the test to make it all sane?
[13:29] <gmb> i.e.:
[13:29] <gmb> client_service = ClusterClientService(reactor)
[13:29] <gmb> ?
[13:29] <allenap> gmb: I’m not sure you need the client service anyway; the MockLiveBlah will be arranging for that already.
[13:29] <gmb> ORLY?
[13:30] <gmb> I’ve missed something somewhere…
[13:30] <gmb> allenap: Oh, d'oh!
[13:30] <gmb> getRegionService()
[13:37] <gmb> *Client
[13:37] <gmb> Right, new failures. All good.
[13:40] <allenap> gmb: http://paste.ubuntu.com/8290586/ (needed to provide a response for the CreateNode stub)
[13:40] <gmb> allenap: Fab, ta
[13:43] <gmb> allenap: Remind me, will get_cluster_uuid actually work from within PServ? I know the API credentials stuff doesn’t, but that should, shouldn’t it?
[13:54] <allenap> gmb: It depends on the packaging I think, if pserv is invoked with CLUSTER_UUID set :-/ Not a good answer I know.
[13:54] <gmb> GAAAAH.
[13:55] <gmb> That’s exactly what I was hoping you wouldn’t say :)
[14:36] <allenap> gmb: Actually, it must be okay; lots of other RPC calls use the UUID.
[14:37] <gmb> allenap: Yeah, I saw the same in plugin.py. Cool.