=== marlinc_ is now known as marlinc [04:14] hello everyone [13:28] roaksoax: Hi, I tried to use FPI using http://pad.ubuntu.com/maas-fastpath-install-preseed but it did not work… I used a different method than the one described in the first 3 lines to set the 'use-fastpath-installer' tag but this part worked (I checked in the UI that the tag was applied to my nodes). Can you double check the script? [13:29] (my nodes were installed using the normal installer, I did not get the FPI to work) [13:39] roaksoax: I'm really wondering how this could work because the url referenced is "http://"+cluster_host+"/MAAS/static/images/[...] and, when talking about the cluster next to the region, this does not exist. [13:39] roaksoax: I guess I'm really missing something here :) [13:40] rvba: 1. dis u add support for saucy to maas? 2. does the preseed renders correctly? 3. are you deploying with juju? 4. change /dev/vda to /dev/sda or where you want ubunty to be ibstalled [13:41] rvba: are you using raring for the maas server? [13:41] rvba: that *does* work and exists [13:42] rvba: we have it running at one of our labs [13:42] roaksoax: 1. not yet 2. checking… 3. yes 4. /dev/vda? /dev/sda? where? 5. yes, raring [13:42] rvba: in the preseed [13:42] 4 [13:42] 4 in the preseed [13:44] roaksoax: I'm sure it works… but I suspect the setup we have in the lab differs from your setup. [13:45] rvba: doesnt matter cluster_host renders based on the cluster for which the node belongs to [13:46] hum, the preseed looks like the normal d-i preseed. [13:46] rvba: then it is not using the right one [13:46] rvba: maybe the wsy u added the tag [13:46] I see the tab on the node's page. [13:47] tag* [13:47] rvba: right but it has special cases [13:48] rvba: check the node.py model for expkanayion [13:50] roaksoax: arg, indeed, there is something weird with that tag. [13:52] rvba: yeah gavin did it in a way that wont work if theres definition or sonethibg [13:53] roaksoax: right, that's the weird part :) [13:55] indeed :) [13:55] >>> [n.should_use_fastpath_installer() for n in Node.objects.all()] [13:55] [True, True, True, True] [13:55] roaksoax: but I don't think that's the problem. ^ [13:57] that method, should_use_fastpath_installer(), is not really used anywhere… === kentb-out is now known as kentb [14:00] rvba: i know but traditional installer is [14:00] so if thats false then the fpi should be used [14:01] I really don't understand how this all fits together. [14:01] rvba: what exactly? [14:03] roaksoax: I got it now: my file /usr/share/maas/preseeds/generic does not contain the {{if node.should_use_traditional_installer() }} trick! [14:04] rvba: plop :) it shoulf if you are using raring [14:04] http://paste.ubuntu.com/5674126/ [14:04] thats weird though [14:06] roaksoax: I think the package from the daily ppa has a pb. [14:06] pb? [14:06] roaksoax: see http://paste.ubuntu.com/5674126/ [14:11] rvba: that doesn't seem to be using the latest branch cause the generic template should have the should_use.... [14:11] rvba: and i need to add packaging support for fpi to the latest packaging [14:11] branch [14:13] bahn. [14:13] bah. [14:13] i'm seeing an issue i think where if i start a node form the cli, then it gets powered on, but it tries to pxe boot to 'poweroff' template. [14:14] and in the web ui, it has 'start node' as an action [14:14] smoser: the node is not sllocated to the user then [14:15] ut is not being allocated* [14:16] i allocate it with that user from cli [14:16] and then start it [14:16] allocate works [14:16] goes from unallocated to allocated [14:16] (state goes from 4 to 6) [14:16] smoser: maybe pxe got disabled? [14:16] ? [14:17] i'm watching it pxe boot [14:17] roaksoax: the fpi stuff is not in the 1.3 branch (the branch for raring)! That's the reason why it is not in the raring daily package. [14:17] as in the node is node.pxe = False [14:17] racedo: yeah the fpi is in trunk [14:17] err [14:17] "netboot": true, [14:17] rvba ^^ [14:17] rvba: use thr package from raring [14:18] rvba it is not in 1.3 becsuse in rarin (the release) it was a patch thsg ensbled fpi because at the time fpi wsdnt gonna be merged upstream [14:19] roaksoax: ah ok. [14:19] GAH [14:21] smoser: but if the ui shows it hsd 'dtart node' as an action the it means it wont pxeboot because node is not stsrted [14:23] roaksoax, its user error [14:23] to some extent [14:23] @#$%ing annoying [14:23] maas-cli maas nodes acquire system_id=node-28272b2c-b844-11e2-bca5-d4bed9a84493 [14:23] that will happily acquire you a *different* system! [14:23] bug 1074317 [14:23] bug 1074317 in MAAS "cmdline acquire and start is difficult" [Medium,Triaged] https://launchpad.net/bugs/1074317 [14:25] smoser: yeah i think that suport should be added to work with systems using its name not its uuid [14:26] roaksoax, no. thats not the problem [14:26] acquire *only* works by name [14:26] rarg, now the preseed template cannot be rendered. [14:26] and if you specify 'system_id=SOME_SYSTEM_ID' it will happily give you a different one. [14:26] rvba thats etherpad messing with unicode [14:26] just as if you say: acquire FOOOWHIZ=excellent [14:26] it just gives you a system. [14:27] i see [14:27] basically 'system_id' is not anythign special to acquire [14:27] rvba: found snother bug with cpu discovery btw [14:27] roaksoax: ah, you're right! [14:28] smoser: yeah i see now. but apsrt from that we should be using sydtem nsme rathet than system uuid [14:29] rvba: ill get you a lshw xml file on monday [14:29] roaksoax: ok, cool. [14:30] ok then I'm off today ... so of there's nothing else y'all need me for I'm off:) [14:50] rvba: btw... re on dhcp email. i thi k is due to the latest changr madr by julian to packaging branch [15:07] just filed two bugs [15:07] bug 1181265 and bug1181263 [15:07] bug 1181265 in maas (Ubuntu) "acquire followed quickly by start may not start system" [Undecided,New] https://launchpad.net/bugs/1181265 [16:26] is there a prescribed way to perform a release-upgrade on a unit?