[15:21] allenap: howdy!! so the cluster_host thing didn't really work unfortunately :( [15:31] roaksoax: Hi there. Because of the get_managed_interface() thing? [15:35] allenap: yeah, it wont get the address of the cluster [15:37] roaksoax: Look at the implementation of get_managed_interface(). You can use a similar approach to getting the address of the cluster, i.e. get any interface of the cluster. [15:44] allenap: right, but the thing is that it seems that if we are not managing any interface, it wont return the address [15:44] allenap: so from what I udnerstood, if we dont configure DHCP/DNS, then it wont return the address [15:44] but we still need itr [15:44] regardless of whether we manage it or not [15:45] roaksoax: http://pastebin.ubuntu.com/5588071/ [15:46] allenap: ack, i'll look into that [15:46] roaksoax: Yeah, get_managed_interface() won't work then, but I meant for you to follow its approach. Afaik, all interfaces are recorded. [15:46] allenap: yeah, the difficulty is finding out what interface it PXE booted from [15:51] allenap: but the difficulty is really finding out what interface it PXE booted from. [15:51] allenap: and thanks btw :) [15:57] roaksoax: Okay, so I could update lp:~allenap/maas/preseed-cluster-host to find an interface on the cluster that's on the same network as the node. I guess that would be better? === kentb is now known as kentb-afk === kentb-afk is now known as kentb [22:53] bigjools: let me know when around :) [23:02] roaksoax: o/ [23:03] bigjools: howdy!! so I think i found an issue with maas and nodegroups :( [23:03] go on ... [23:04] bigjools: so while trying to work on the cluster_host thing, I first did this:http://pastebin.ubuntu.com/5589024/ to basically obtain all the interfaces of a cluster, instead of the 'managed' interface [23:04] (extending what allenap did) [23:05] roaksoax: that won't do that [23:05] it's returning the first [23:05] bigjools: right :) [23:05] bigjools: i guess I should have said get any of the interfaces [23:05] there's a better way of doing it as well :) [23:05] but go on [23:05] bigjools: but anyways, that's not really the issue though [23:06] bigjools: the thing is that I installed 2 cluster controllers [23:06] one in the region, and one external [23:06] 1 first enlisted/commissioned node01 with external cluster in 192.168.123.3. Then node02 with local (in region) cluster in 192.168.123.2 [23:07] bigjools: and then I noticed this: http://pastebin.ubuntu.com/5589031/ [23:07] bigjools: both nodes seem to be in the same nodegroup, when they enlisted/commissioned in a different nodegroup [23:08] bigjools: note that the local cluster running in the region was registered to maas *after* the external [23:08] but I;ve tested the other way around (local cluster registered first, then external), and the adddress returned there is 192.168.123.2. [23:08] let me lool, I think your code is buggy [23:08] look [23:09] this leads me to believe that all nodes are said to belong to the same nodegroup [23:09] check in the DB [23:10] bigjools: http://pastebin.ubuntu.com/5589038/ [23:10] bigjools: unless my logic is incorrect, how else would I check to what nodegroup the node belogs to [23:11] hmmm [23:11] bigjools: this is what I did to get the cluster_host btw: http://paste.ubuntu.com/5589041/ (in get_node_preseed_context) [23:12] roaksoax: multi clusters were tested in the lab and worked ok, so either something regressed recently or your setup is not doing what you think it is [23:13] are you sure your nodes are on separate lans? [23:13] and cluster controllers [23:14] bigjools: my nodes are in the same lan, but they do register themselves with a different cluster controller. I tell it where to PXE boot from so I know they are pxe booting from different clusters [23:15] roaksoax: are you sure? :) [23:15] bigjools: 100%. It is a controlled environment and DHCP is managed by libvirt, which tells it where to pxe boot from [23:15] let me re-test [23:16] roaksoax: if you disable the DHCP that you think it is using, does it still boot? [23:16] bigjools: nope [23:16] ok [23:17] bigjools: it is a clean environment with no DNS/DHCP being managed by maas [23:17] nor dhcp is running in any nore [23:17] roaksoax: ok file a bug [23:17] not sure we'll get to it right away quite frankly [23:18] right [23:18] this does affect the FPI though === kentb is now known as kentb-out [23:31] we're getting the FPI done [23:31] roaksoax: what's the cutoff for raring? [23:32] bigjools: FF is this thursday :) [23:32] yay [23:32] when you say "final" ... [23:32] bigjools: so I'm planning on uploading and then just doing FFe's for what's important [23:32] bigjools: so yeah we have like 1 extra month to request FFe;s [23:33] ok