[03:00] <bigjools> jtv: what else did you need to know?
[03:01] <jtv> What exactly it is we're not doing.  Multiple worker packages?
[03:02] <bigjools> jtv: right, single worker per rack
[03:02] <jtv> Regardless of arch?  Good.
[03:02] <bigjools> indeed
[03:02] <bigjools> it's a little mad to have multiple ones
[03:03] <bigjools> but I suspect that thought arose due to intimate lack of knowledge of our new architecture
[03:04] <jtv> Nicely put
[03:05] <bigjools> s/intimate lack of/lack of intimate/
[03:06] <jtv> Shame.  You had a nice pattern going there.
[03:07] <jtv> Oh, about the anonymous-metadata issue: my preference is to have a config setting (dev/demo, not production) that would open up a mirror to the metadata tree keyed by MAC address: /metadata-by-mac/<version>/<mac>/…
[03:08] <jtv> The way I see it, the MAC address is about the most reliable knowledge the node has at the beginning of the provisioning stage, and conversely MAAS knows all its MAC addresses, so it makes a good key.
[03:10] <bigjools> jtv: I asked Francis for confirmation on this bug
[03:11] <bigjools> ie. why was it escalated
[03:11] <bigjools> and we'll fix that problem now rather than one that doesn't exist (yet)
[03:13] <jtv> What is the existing problem and what is the non-existing problem?
[03:17] <bigjools> exactly what I want to clarify
[03:19] <jtv> So… put this one on ice until we have an answer?  Or start doing work that'll be needed either way?
[03:20] <jtv> Actually, one thing that needs to be done either way is proper coffee.
[03:21] <bigjools> jtv: if you can see common work, by all means do it
[03:22] <jtv> but
[03:25] <jtv> no but?
[08:07] <allenap> Morning.
[08:08] <jtv> Hi there allenap
[08:09] <allenap> \o jtv
[13:54] <cheez0r> criminy, on a freshly loaded MaaS server node I can't even get cobbler system dumpvars to run
[13:56] <cheez0r> "cannot marshal None unless allow_none is enabled"
[14:52] <allenap> Daviey: Do you have experience with iPXE? James Page was saying that it's Good Stuff.
[14:53] <Daviey> it is very good stuff.
[14:53] <Daviey> And i do.
[14:54] <Daviey> allenap: The 'maas without controlling infrastructure' story, was initially going to be ipxe based.
[14:55] <allenap> Daviey: I'm not familiar with that story, unless I know it under a different name.
[14:58] <Daviey> allenap: user doesn't have control over dhcp
[14:59] <allenap> Daviey: Ah, right. Is there any reason to not use ipxe for all scenarios?
[15:00] <Daviey> allenap: doesn't work on arm
[15:00] <allenap> Daviey: Blast.
[15:06] <Daviey> it would make me very happy if it did
[15:06] <Daviey> would provide a nice abstraction
[15:06] <Daviey> and support iscsi out of the box
[18:12] <Aaton> so anyone have a good write up for setting up maas in a virtualbox test env.  Setting up the network right is the main thing I'm looking for
[18:31] <robbiew> Aaton: marcoceppi might
[18:32] <marcoceppi> Aaton: try this: http://marcoceppi.com/2012/05/juju-maas-virtualbox/ it's not a perfect experience, but you can get MaaS running
[18:56] <hazmat> marcoceppi, nice!
[20:12] <Aaton> thanks!
[20:30] <marcoceppi> Aaton: it's kind of a chronicling, so I'd recommend reading the whole thing through before starting, half-way through I got smart and figured a bunch of things out. Wiki page for VirtualBox should be done this weekend
[22:25] <Aaton> marcoceppi: yea, just read it over. so did you create a second interface. I want to connect to the bridged interface but boot the other servers inside virtualbox. I don't have control of the dhcp server.
[22:30] <marcoceppi> Aaton: I didn't create a second interface, I used a VM to put the MaaS main node on
[22:30] <marcoceppi> If you install the main node as a VM, then install maas-dhcp ( didn't have access to my offices dhcp server, but it worked anyways ) it *should* work
[22:39] <Aaton> marcoceppi: I'll be giving that a try