[00:59] <Rastin> Hi
[00:59] <Rastin> I have a question about MAAS
[01:00] <Rastin> does eanyone know the nodes of must must be connected directly to internet? or the use MAAS Server to connect to the internet?
[01:07] <sfeole> Rastin: i just happened to look over here, while trolling through my irc channels,  the nodes to not need to be directly connected to the internet,  They MUST be accessible via the maas cluster where that may be.   You can pass a HTTP / HTTPS proxy to your maas nodes so that they can access the internet, download tools, etc
[01:09] <bigjools> Rastin: they need a little access but it can be via the proxy installed on the region controller
[01:10] <jhobbs> you can run totally offline on MAAS 1.7, but it's not released yet. you do need to setup offine ubuntu archive and image streams mirrors
[01:10] <bigjools> they just download a few packages when installing
[01:10] <bigjools> what jhobbs said :)
[01:11] <Rastin> but i get can not connect to the proxy server error
[01:11] <Rastin> what i missed in network connection of cluster and region controller
[01:11] <Rastin> ?
[01:15] <Rastin> do I have to do some steps in network configuration of Cluster/Region COntroller?
[01:15] <Rastin> The node can not connect to the internet, but MAAS server can
[01:16] <bigjools> unless you're using 1.7, it's best to set up ipforwarding on the maas server
[01:18] <Rastin> i havet to setup ip forwarding?
[08:59] <jtv> allenap: some background on the pserv.yaml-rewrite bug...
[08:59] <jtv> Configuring an IPv6 address there won't work yet, and that's a documented limitation.
[09:00] <jtv> BUT when people do do that and realise their mistake, the bug stops them from reverting.
[09:00] <allenap> jtv: Okay, worth fixing for that then.
[09:01] <jtv> I think so.  I wish it would also magically IPv6-enable provisioning etc, but that's a bridge too far.
[12:22] <mwenning> k2, can you send me the google hangout link?
[12:43] <allenap> rvba: https://code.launchpad.net/~allenap/maas/respawn-cluster-until-it-works--1.7/+merge/238709 is ready to go into Utopic's package branch. It's already been reviewed into trunk, so I'm just looking for a rubber-stamp into 1.7.
[12:45] <rvba> allenap: cool
[12:51] <mburleigh> hello, MaaS newbie question. I have a SuperMicro server that shows a 0 CPU count, 0 MB ram. Where do I start to find out why its not reporting correctly. And does these actually mean anything in regards to deployment?
[13:41] <jhobbs> mburleigh: have you commissioned the node yet?
[14:37] <mburleigh> jhobbs: no, will that re-collect that information?
[14:38] <jhobbs> mburleigh: it will collect the information yes
[14:52] <mburleigh> jhobbs: commissioned and still cpu and mem are 0
[14:53] <mburleigh> and in ready state
[14:54] <jhobbs> mburleigh: that's interesting - what version of MAAS?
[14:54] <mburleigh> i guess commissioning is not completing, because it is still an option for this node. I have two other, different servers that get the information correctly.
[14:54] <jhobbs> mburleigh: you can commission a node multiple times
[14:55] <mburleigh> hmm
[14:55] <jhobbs> mburleigh: is there lshw output in a "Discovery Data" section at the bottom of the node's page?
[14:58] <mburleigh> discovered as a switch, not server ;-)
[15:01] <gmb> allenap: Asked for your thoughts over on https://bugs.launchpad.net/maas/+bug/1382575
[15:04] <jhobbs> gmb: what good are django's transactions if they don't commit prior to returning info to the user?
[15:04] <jhobbs> transactions per request that is
[15:05] <jhobbs> i wonder if claim_sticky_ip is broken the same way
[15:05] <gmb> jhobbs: There are 11 more words in your question than you really needed :)
[15:05] <gmb> jhobbs: That’s entirely possible.
[15:05] <jhobbs> haha
[15:07] <jhobbs> adding a transaction.commit() in node.acquire(), after the save, doesn't fix this
[15:14] <gmb> jhobbs: Arse. Then we probably need to lock the table.
[15:14] <gmb> Or at least the row
[15:14] <gmb> Haven’t done that myself in MAAS before… allenap: any advice?
[15:23] <mburleigh> jhobbs: ok, so the switch info is always there, but there is no additional data (like other servers have), what next?
[15:25] <jhobbs> mburleigh: what version of MAAS is this?
[15:28] <mburleigh> 1.5.4
[15:30] <mburleigh> should I upgrade to 1.7?
[15:31] <jhobbs> no
[15:32] <jhobbs> mburleigh: there should be a link on your node's page: "Commissioning Output"
[15:32] <jhobbs> mburleigh: sorry, it will be something like "8 output files" under the header "Commissioning Output"
[15:32] <jhobbs> mburleigh: can you go there and see if there is lshw output?
[15:36] <mburleigh> doesn't exist, that node only has 6 output files
[15:37] <jhobbs> mburleigh: i don't know then - it sounds like a bug to me - please file one. have you tried deploying to the node?
[15:38] <mburleigh> modaliases contains: dmi:bvnAmericanMegatrendsInc.:bvr2.0:bd01/11/2013:svnSupermicro:p nX9DRi-LN4+/X9DR3-LN4+:pvr0123456789:rvnSupermicro:rnX9DRi-LN4+/X9DR3-LN4+:rvrREV1.20A:cvnSupermicro:ct3:cvr0123456789:
[15:39] <mburleigh> jhobbs: yes, how relavant is the cpu and mem count in maas, besides constraints?
[15:39] <jhobbs> mburleigh: that's all it's used for
[15:40] <mburleigh> ah, good to know. of course I can manually edit these and still constraints, yeS?
[15:40] <mburleigh> still use constraints
[15:40] <jhobbs> mburleigh: yes
[15:40] <mburleigh> thx
[15:43] <allenap> gmb: You could use a new DatabaseXactLock; see maasserver.locks for a hint.
[15:43] <gmb> jhobbs: ^^
[15:44] <gmb> allenap: ta
[15:45] <allenap> gmb, jhobbs: That'll help address this issue, but there are going to be others.
[15:45] <gmb> Oh yes there are
[15:45] <allenap> We need a more general solution.
[15:45] <gmb> Anyway, I’m juggling three conversations here, so let me back out of this one, much as I’d love to be involved.
[15:46] <allenap> Okay.
[16:03] <jhobbs> allenap: yeah, DatabaseXactLock worked