[07:20] <rvba> bigjools: with your fix in https://code.launchpad.net/~julian-edwards/maas/utopic-syslinux-location/+merge/223851 we should finally have a working Utopic package.  (The other blocker got resolved yesterday.)
[07:36] <rvba> bigjools: can you confirm that you saw a *static IP* in the forward zone file?
[07:56] <rvba> bigjools: I'm testing the new static DHCP stuff and I'm confused: my commissioning node gets a static IP.
[10:43] <bigjools> rvba: thanks for landing the fix. Yes I saw a static IP for a CNAME.
[10:43] <bigjools> rvba: static IP for commisioning is either a bug or a feature depending on your PoV
[10:43] <bigjools> :)
[10:44] <rvba> bigjools: well, looks like the lander is down (I just redeployed the lander now) so it's not landed yet.
[10:45] <rvba> bigjools: Yeah, I figured out the CNAME thing eventually: even static mappings end up in the leases file… so they are parsed and stored in the DB… eventually.
[10:45] <rvba> bigjools: oh, it just got landed
[10:45] <bigjools> yeah static leases in there is a feature too
[10:45] <bigjools> not that worried about those
[10:45] <bigjools> in fact it drives the DNS at the moment :)
[10:46] <rvba> bigjools: well, have a look at my email, I found myself in a weird situation were I had conflicting IP addresses (an allocated node had a *dynamic* IP).
[10:46] <bigjools> rvba: I skimmed it, it's too late and I am too tired to take it all in
[10:47] <rvba> bigjools: sure, no worries.  It's just good to be able to QA this…
[10:47] <bigjools> rvba: yes, thank you for doing some QA
[10:48] <rvba> welcome
[10:48] <rvba> Starting a CI run now…
[11:23] <allenap> jtv: You’ve got some conflicts in https://code.launchpad.net/~jtv/maas/dual-stack-network-helpers/+merge/223869
[11:23] <allenap> jtv: I’ll review that after lunch if it hasn’t been taken by then.
[13:36] <rvba> Reviewer needed! https://code.launchpad.net/~rvb/maas/fix-bootloader-path/+merge/223920
[13:45] <rvba> allenap: gmb: Anyone up for a tiny review? ^
[13:45] <gmb> rvba: Sure
[13:45] <rvba> Ta
[13:51] <wrale_> on 1.5.1+bzr2269-0ubuntu0.1: i'm running supermicro nodes.  i manually typed in the IPMI information which I know to be correct via ipmiview (SM's tool).  node says ready and has IPMI v2.0 config, yet start but is greyed out.. is this normal?
[14:11] <jtv> wrale_: if Start is greyed out in the UI, there should be a tooltip to tell you the reason.
[14:11] <jtv> For me it's usually "need to register an SSH key or you won't be able to access the node."
[14:12] <jtv> That reminds me — blake_r, jhobbs: we inhibit the Start node action in the UI if the user has no ssh keys registered... Does that need to be smarter and look at SSL and/or SSH keys depending on the chosen OS?
[14:13] <blake_r> jtv: The node should fail to start if missing sslkeys and windows
[14:13] <blake_r> jtv: meaning not allowed*
[14:15] <blake_r> jtv: quick question
[14:15] <blake_r> jtv: would it be okay to do this
[14:15] <blake_r> jtv: http://paste.ubuntu.com/7674825/
[14:17] <jtv> Looking...
[14:17] <jtv> But note that "fail to start" is different from "inhibit."  If you don't register an SSH key, and then you browse to a node's page, the Start button will be greyed out and a tooltip will say that you need to register an SSH key.
[14:18] <jtv> blake_r: it's ugly but in the context I think it makes sense.
[14:18] <jtv> (Your paste, that is)
[14:19] <blake_r> jtv: i understand, having a tooltip that says register sslkey would be good
[14:20] <blake_r> jtv: tahnks
[14:20] <jtv> ISTM if you want to start a Windows node, you can use either type of key; if you want a regular node, you need an SSH one.
[14:21] <roaksoax> hey guys, if we were to add a way for users to add custom power methods, is that soemthing we would want to do?
[14:21] <roaksoax> jtv: ^^
[14:21] <jtv> AIUI, no.
[14:21] <roaksoax> rvba: ^^
[14:21] <blake_r> roaksoax: yes
[14:21] <roaksoax> like "Add power method"
[14:21] <jtv> We want them all in the codebase, so we can change the API, right?
[14:21] <blake_r> roaksoax: look at my plugin architecture
[14:21] <roaksoax> blake_r: branch?
[14:21] <blake_r> roaksoax: doc
[14:22] <roaksoax> blake_r: doc?
[14:22] <roaksoax> :)
[14:22] <roaksoax> jtv: right, so if we were to do a plugin architecture such as the one blake_r designed in the dcoument. Why would we not want to do that?
[14:22] <roaksoax> jtv: what are the benefits of having it so
[14:22] <jtv> Orders.  :)
[14:22] <roaksoax> jtv: what are the disadvantages?
[14:22] <roaksoax> jtv: just exploring cause someone on canonical was requesting whether we can do that
[14:23] <jtv> Right.  My recollection is that we want to keep all power methods in a single codebase, to keep development unified.
[14:23] <gmb> rvba, jtv, allenap: Can haz review: https://code.launchpad.net/~gmb/maas/clean-ip-ranges-ipv6/+merge/223936
[14:23] <jtv> I'll take it.
[14:24] <gmb> Thankee squire
[14:24] <rvba> roaksoax: Like jtv said, I think we want to keep all the code in MAAS itself if possible.  But the plugin architecture allows external packages to contribute power method so it's technically possible.
[14:30] <wrale_> jtv: brilliant.. thanks!
[14:31] <jtv> wrale_: was it the SSH key?  :-)
[14:31] <allenap> roaksoax: Mark wants it all in MAAS, so that we manage QA, via CI and the like. Having a hook point for adding new power methods is, IMO, about helping the user to help herself; if we make it easy for someone to make a new power method, that’s a happy user. However, they may be unhappy when we change the API and their code breaks...
[14:32] <wrale_> indeed!
[14:32] <jtv> :)
[14:32] <wrale_> strange UI feature
[14:32] <allenap> roaksoax: I think we should have a hook, but not something that can be added via the UI. As in, only a developer could make it. That way there’s a greater chance that it can be upstreamed to us.
[14:33] <allenap> roaksoax: And, iirc, that’s what blake_r’s doc suggests.
[14:33] <wrale_> i'd rather fail fast and find i can't login than greying out this box
[14:33] <wrale_> hmm maybe that's not fail fast.. either way :)
[14:33] <jtv> Well theoretically, being able to see the notice _before_ you click the box should be faster.
[14:33] <wrale_> ^ yup
[14:33] <jtv> The question is whether the tooltip is obvious enough.
[14:34] <jtv> Maybe we should just show that message under the button.
[14:35] <jpds> jtv: You around?
[14:36] <jtv> Hi jpds
[14:36] <jpds> jtv: I'm having issues with a fresh install of MAAS and gmb said that you might be able to help.
[14:36] <jtv> ?
[14:37]  * gmb also named rvba, just to spread the misery
[14:37] <rvba> heh
[14:37] <jpds> Oh, sorry.
[14:37] <gmb> Also allenap, first mate of HMS MAAS
[14:37] <gmb> Basically anyone who’s not me, then…
[14:38] <jpds> I'm getting this error: http://paste.ubuntu.com/7674927/
[14:38] <jtv> Good to know who our friends are
[14:39] <jtv> jpds: I think that means there's a problem with the credentials string.
[14:40] <jtv> The oauth module probably just happens to validate the "consumer" part of the string first, finds a fatal problem there, and says "invalid consumer" instead of "this is a recipe for mussels à la mode, not a valid credentials string."
[14:42] <gmb> jtv: This is on the first PXE boot of a node, which is what’s confusing me, and why I refered tit to you guys. :)
[14:42] <jtv> Ooo
[14:42] <allenap> jtv: I believe we added the is_recipe_for_mussels_a_la_mode() validator a while back.
[14:42] <jtv> Then might it be the old clock-skew problem?
[14:42] <gmb> D'oh!
[14:42] <gmb> Didn’t think of that.
[14:45] <jtv> jpds: I thought we'd fixed it, but it might be something like "node's hardware clock is wildly out of sync with the server's clock."  It upsets oauth.
[14:46] <jtv> Maybe you're not getting NTP?
[15:13]  * jpds returns.
[15:13] <jpds> jtv: Well, ntp's on the box. Offset: -0.030.
[15:14] <jpds> Oh, I got a machine in MAAS.
[15:20] <jpds> jtv: It's still happening.
[15:28] <gmb> allenap: When you say “It might be worth allocating from the fc00::/7 range too; http://www.simpledns.com/private-ipv6.aspx has an explanation why.” do you mean “instead of just from entirely random address space”?
[15:29] <allenap> gmb: Yeah.
[15:29] <gmb> Ok
[16:54] <allenap> gmb: I’ve got an extremely little bit of bad news for you. I really like the way you’ve fixed getRandomIPv6Address(), but there’s a bug: randint() can return either endpoint, one of which is network.size, and network[network.size] results in IndexError. On average it’ll only happen once every 2658455991569831745807614120560689152 calls though, so it
[16:54] <allenap> really is a *very* small bug.
[16:57] <wrale_> i wish #juju were as responsive as #maas
[18:53] <wrale_> how can i disable swap partition creation during the node install process (during maas provision)?  do i need to edit the preseed files?
[18:53] <wrale_> swap is killing my installs to a 240GB ssd on a server with 256GB ram
[19:29] <blake_r> wrale_: if you use fast-path installer
[19:29] <blake_r> wrale_: no swap partition is created
[19:29] <blake_r> wrale_: there is a button on the node view to enable it
[19:29] <blake_r> wrale_: its normally enable by default
[19:43] <wrale_> thanks blake_r .. i'll try this
[20:32] <wrale_> blake_r: that worked.. thanks again.  any idea how i might get the juju status command to get its dns lookups from the maas running on the same machine?
[20:32] <blake_r> wrale_: set your first dns as the maas ip
[20:33] <wrale_> blake_r: cool.. will do.. thank you
[20:33] <blake_r> wrale_: dont forget to set your upstream dns address on the maas settings page
[20:35] <wrale_> hmm.. can't seem to find that.. will look for docs
[20:35] <wrale_> oh.. nvm. .found it
[20:41] <wrale_> is it normal for my secondary interfaces (enabled with dhcp and dns in maas) to remain unconfigured after booting a node (juju bootstrap) with the fast installer?
[21:19] <AskUbuntu> Can Ping but Cannot SSH to Openstack VM Instace | http://askubuntu.com/q/486151