=== matsubara-dinner is now known as matsubara [03:57] Guys, the MaaS server really needs two ethernet interfaces?! Even if it is not the default gateway of the nodes?! [04:00] why do you think it needs two interfaces? [04:02] Weel, I think I misunderstood this: https://help.ubuntu.com/community/UbuntuCloudInfrastructure -> "...has all nodes with two network interfaces..."... Sorry for the buzz... [04:02] * bigjools reads [04:03] huh, I don't know why it says that, I run it with one interface on my server/nodes [04:05] Damn... I'm trouble to find a good network topology for MaaS -> Juju -> OpenStack... I'll try it again... [04:05] BTW, I also liked this: http://people.canonical.com/~gavin/docs/lp%3Amaas/orientation.html#setup [04:05] KISS [04:05] lol [04:05] All my nodes have two gigabit eth... [04:07] I'm thinking about enabling the BOND interfaces before deploying my MaaS nodes on top of it (BOND)... What do you think?! My network will be very busy when in production... [04:07] I fixed the wiki [04:07] * bigjools has to run, back later [04:08] Awesome! :-D [05:00] What the "Ready" status after commissioning means?! I think that "Ready" means "ready to use" but, it is the beginning of the Installation process... Is that normal? [05:01] Also, how can I manually edit my node's harddisks partitions?! The MaaS node installation just do "guided partiutioning" but I do not like it. Any tips? [05:04] Now my newest node is "allocated to root" but, still installing... [05:05] When I click "start node", MaaS starts the node automated-installation... Have no idea if this is right... [05:05] I'll start from scratch again... Damn... =( [05:05] This is hard... [05:09] ThiagoCMC: Ready means "in the pool of commissioned nodes and ready for users to allocate and deploy" [05:10] ThiagoCMC: generally, you want to use juju with maas [05:10] makes life much easier [05:13] Mmm... Okay... Gotcha! [05:13] And about node's harddisk partitioning? I want to do it manually. [05:13] Is that possible? [05:16] Or Juju/OpenStack expect it (HD partitions) the way MaaS deploy it? I'm familiarized with nodes that have: two HDs -> raid1 -> LVM -> lv_swap / lv_root / lv_vps1_disk0 / lv_etc... [05:52] Where can I find the IP of my fresh MaaS node? [05:53] I'm using managed DHCP/DNS and using my MaaS DNS server to resolve but, it does not resolve by names... :-/ [07:02] So, there is only one way to figure out the IP of my node: "grep -i MAC /var/log/syslog" ??? [07:02] on maas server? [07:14] ThiagoCMC what about /var/lib/misc/dnsmasq.leases ? [07:17] I'm using Ubuntu 12.10... There is no dnsmasq installed... [07:18] Also, this isn't a expected solution, right?! [07:19] Seems more or less like a "workaround"... I would like to know the IP of a node by MaaS CLI or GUI interfaces... Not this... =( [07:28] i havent used 12.10 nor do i know if there s a more convenient way. [07:28] hmmm, what about juju status ? It should list all the machine and their names [07:28] at least it does on 12.04 [07:32] I still do not have touched Juju... [07:32] I'm learning about MaaS first... [07:33] But tks! I'll take a look into it now... [07:33] The point is, somebody told me that MaaS can be used without Juju, so, I'm trying it... [07:33] Also [07:34] Did you know if it is possible to add a already installed Ubuntu Server 12.04 within MaaS (not installing "through" it)?! [07:42] hmm, i think you need juju to deploy nodes. (at least i never did deployment in another way with maas) [07:42] you can add a node manually if you know its ip, but i do not know how to have it in ready state without letting maas reinstall stuff on it from scratch [07:43] when i added node manually, they ended up in enlisted state i think, so i still had to comission them (wich involve booting and installing stuff on the box) [07:44] Mmm... Okay! Thanks for the info... [07:45] I'm a bit concerned about MaaS HD partitions and RAID1... It just do guided-partitioning... Which sucks... [08:20] hey, guys, I'm getting this error after restarting a make run session: setlock: fatal: unable to lock /run/lock/maas.dev.database: temporary failure [08:20] any idea? [08:23] I just deleted all /run/lock/maas* files and it works now.. [08:34] dimitern: that has been my experience. I filed a bug on it, which had the comments 'the locks should go away on reboot', but that has been broken in my experience. [08:50] allenap: bug #1066668 [08:50] Launchpad bug 1066668 in MAAS "No indication if cluster controller is failing to connect" [High,Triaged] https://launchpad.net/bugs/1066668 [08:50] jam: Ta! [08:56] rbasak: here is what juju status gives me: http://paste.ubuntu.com/1283078/ [08:57] rbasak: I'm interested in what you have instead of dns-name: 192-168-10-10.master [08:58] rvba: I'm just bootstrapping a node now [08:58] rvba: will be about 20 minutes [08:58] Ok, thanks. === jtv1 is now known as jtv [09:14] rvba: http://paste.ubuntu.com/1284658/ [09:15] rvba: something must be fundamentally different with the way I'm doing it [09:15] rbasak: you;re not using MAAS' DNS service. [09:15] rvba: I'm installing the maas package only (and juju), and have an external dhcp server [09:15] that'll be it [09:15] Thanks for checking rbasak. [11:51] fun, `rm -rf db; make sampledata` -> http://pastebin.ubuntu.com/1284868/ [11:54] that did not happen earlier today [12:01] rvba: is this a known issue with a workaround?^ [13:22] mgz: the dev fixture data relies on the fact that the master nodegroup has id 1. Which should happen when creating the db from scratch. [13:22] mgz: wait Problem installing fixture 'src/maasserver/fixtures/dev_fixture.yaml.bz2' [13:22] Why dev_fixture.yaml.bz2 ? [13:22] mgz: are you modifying the dev fixture in that branch? [13:39] mgz: I'm getting the same problem on trunk. I'm investigating. [13:48] thanks rvba, confirms it's not just my local setup [13:48] mgz: I found the problem. MP coming up. [13:55] mgz: care to review? https://code.launchpad.net/~rvb/maas/fixture-error/+merge/130120 [14:01] ThiagoCMC: you could customize the partioning by modifying the preseed data [14:02] rvba: I decided to pick up https://bugs.launchpad.net/maas/+bug/1027720 [14:02] Ubuntu bug 1027720 in MAAS ""maas-nodegroup-worker" is declared in two places" [Low,Triaged] [14:02] rvba: so I just set it to in progress and I can begin? [14:02] dimitern: yep [14:02] dimitern: that's a good one to get started ;) [14:04] rvba: ok :) I have no idea yet, but I'll dig in and ask if needed [14:04] rvba: looking [14:04] ta [14:04] one question, the 'null' behaviour is specified where? [14:05] because the testing factory.make_node behaviour is different without a nodegroup to what I guess the migration stuff does? [14:06] mgz: yeah, as I say on the MP, the ensure_master method will take care of the nodes without a nodegroup. [14:06] evilnick_: another question/task we should document: how do I modify the partioning done by MAAS [14:06] see ThiagoMC questions earlier today in the logs: http://irclogs.ubuntu.com/2012/10/17/%23maas.html [14:07] evilnick_: the answer lies in changing the preseed data, but you'll need to chase one of the server guys to get some actual example configuration changes [14:07] evilnick_: smoser, roaksoax or Daviey would probably be able to answer that [14:08] when does that happen though? [14:08] flacoste: ok, thats a great help. I shall check the logs and add it to my list. [14:08] because if I just make sampledata then look in the db, the nodes don't have a linked nodegroup [14:09] mgz: indeed, that happens in src/maasserver/start_up.py [14:10] rvba: src/maasserver/models/nodegroup.py does not contain any reference to maas-nodegroup-worker or I'm missing something.. [14:11] mgz: maybe we can add a call to ensure_master when make sampledata is run. [14:12] dimitern: you're right, this was apparently cleaned up in r853. [14:13] rvba: :) I figured something like that and started looking at the history to confirm it [14:13] I think this way is fine, generally sampledata is followed make run or something else for like testing, and that does do the population [14:13] mgz: Yes, that was my thought. [14:15] rvba: approved. [14:15] rvba: so I commented that and changed the status to Fix Committed [14:15] dimitern: cool, that was a very easy one ;). [14:16] dimitern: I changed is to fix released because r853 has been released already. [14:16] s/changed is/changed it/ [14:16] ok cool [14:17] I'm picking up #984117 then [14:18] dimitern: that involves JS stuff. I'd suggest going after 1061870 (which also is High priority) [14:19] But it's really up to you. [14:19] rvba: after you mean after I do #984117 or instead do #1061870 [14:20] dimitern: I'd suggest doing the API stuff first unless you really like JS :). [14:20] rvba: I actually really like JS :) not used YUI though, mostly jquery [14:22] dimitern: all right then. Up to you. [14:24] rvba: I'm not sure though how to push my changes, etc. when I run bzr status it shows most of the files in the working dir are changed (?!) I'm new to bzr in fact, but I used both git and hg [14:25] dimitern: that's weird, you've branched off lp:maas, modified a bunch of files… and now most of the files are 'modified' (bzr st)? [14:26] rvba: I branched of lp:maas trunk, didn't really modify anything yet, just run make (install-deps, run) [14:26] dimitern: can you paste the result of 'bzr st'? [14:26] rvba: just a sec [14:28] rvba: http://paste.ubuntu.com/1285115/ [14:28] dimitern: !? [14:29] rvba: if I run bzr diff it shows a bunch of properties changes -x to +x [14:29] heh [14:29] you're using dodgy filesystems again? :) [14:30] I was about to ask the same question :) [14:30] I remember doing chmod -R before :) [14:30] no, this time is a normal fs [14:30] in fact chmod -R ug=rwX,o=rX or something :) [14:30] because I was not sure what's not working initially with the installation [14:31] how I can revert all these changes? [14:31] bzr revert [14:32] ok :) now it looks ok 10x [14:45] rvba: is there an easier way to change node statuses instead of working directly with the models from a py shell in django? [14:46] dimitern: that is really the easiest way if you want to tweak the statuses. [14:46] rvba: running bin/maas shell right? [14:47] dimitern: make harness [14:47] dimitern: same as bin/maas shell but with the right settings loaded. [14:47] rvba: ok 10x :) [15:05] rvba: where should this information show (node_views.js) - on the front page with the wheel? [15:10] dimitern: let me have a look at that bug again first. [15:11] dimitern: well, we still don't support reserving nodes. [15:12] rvba: I saw the https://code.launchpad.net/~allenap/maas/hide-retired-reserved-bug-979902 was merged into trunk, so it should be possible now [15:12] at least to show it, if not change it from the ui [15:13] dimitern: yeah, but there is no point in showing it if it's not possible for a node to be 'reserved'. [15:13] The branch you mentioned hides the 'reserved' nodes. And then the bug says: "Once MAAS supports reserving nodes this line can be shown. " [15:13] This has not happened yet. [15:14] I see, so then probably I should pick something else? [15:14] like #1061870 you mentioned? [15:14] Yep, I'm afraid you're back to that APi bug I mentioned. That's a real one :). [15:14] ok :) no sweat [15:18] rvba: so I'll set 984117 back to nobody, but what status should I set? Confirmed? I does not seem to have Triaged in the list.. [15:19] 5~ [15:19] arg [15:20] dimitern: Triaged is in the list, just on top of 'In progress'… sure you don't see it? [15:20] rvba: once it's already In Progress it does not show it [15:21] dimitern might not be in all the correct launchpad groups still? [15:21] dimitern: I've set it back to 'Triaged' [15:21] rvba: could be, who should I ask? [15:21] 10x [15:21] not in ~launchpad yet. [15:21] I guess flacoste is your man here. [15:22] ok === hazmat is now known as kapilt [16:10] rvba: so I will be uploading a release to PPA that should be 0-day SRU'ed to quantal for you guys to test and make sure it doesn't introduce any regressions [16:10] allenap: ^^ [16:11] since I can test it, you guys will have to help with that [16:14] Okay. [16:26] matsubara: can you help there as well^^^ [16:27] roaksoax: did you use the branch that Julian created yesterday? [16:27] flacoste, sure [16:27] flacoste: in the process of doing so [16:27] awesome [16:27] roaksoax, let me know when the package is published [16:27] flacoste: network is not really a fun thing here :) [16:27] roaksoax: like most non-elmo run conference :-) [16:28] indeed :) [16:37] flacoste: what about bug #1067261 ? It hasn't been merged to the quantal branch [16:37] Launchpad bug 1067261 in maas (Ubuntu) "Web reference in UI points to wrong place" [Undecided,New] https://launchpad.net/bugs/1067261 === matsubara is now known as matsubara-lunch === kapilt is now known as hazmat [17:39] roaksoax: is the new package in the experimental ppa? I don't see it there. [17:42] rvba having trouble uploadib [17:45] k [17:45] roaksoax: that's probably an invonlatary omission by bigjools [17:53] flacoste: alright, will take care of it right before uploading to archive since it is really not a big deal [17:53] big change* [17:53] rvba: should take ~10 mins to be fully uploaded into the archive... [17:53] errexperimental ppa === matsubara-lunch is now known as matsubara [18:42] matsubara it has been uploaded not yet built [18:43] roaksoax, ok [18:43] thanks [19:12] Guys, how to disable guided partitioning during the node installation?! Tks! [19:13] I want to do it manually... [19:13] because of my "raid1 -> LVM" setup... [19:37] ThiagoCMC: you need to edit the pressed template used to tell d-i how you want your partitioning to take place [19:37] ThiagoCMC: unfortunately, i can't provide exact guidance on what you need to put in there [19:47] ThiagoCMC: This probably isn't much help right now, but I am going to find out how to do this and write up the appropriate instructions. Hopefully i'll have something tomorrow. [20:45] roaksoax: your package just finished building (I /think/). [20:46] roaksoax: not yet published apparently. [20:53] roaksoax: it seems to be stuck :/ [20:57] flacoste: any idea what's happening with the ppa package? Looks like it's stuck. [21:02] rvba: no idea [21:04] yeah stuck in publishing [21:21] flacoste, okay... thanks! [21:22] BTW, where can I find the IPs of my fresh installed nodes?! Since the managed DNS do not resolv the named that appear in MaaS web GUI... [21:24] flacoste, I think that if I use "dpkg get/set selections" I can generate my own preseed... So, MaaS just need to provide a way to import my own preseed, right?! [21:25] I'm talking about doing RAID1 -> LVM and copying it to MaaS as a template for same kind of hardware... [21:31] martinx_: no, that's a package selection, the preseed is a d-i format which includes package selection but has a special syntax to specify partitoinaing [21:33] Mmm... Okay === martinx_ is now known as ThiagoCMC [21:35] Would be very nice if MaaS provides a way to customize its preseeds (hardware related) before starting node installation... Is this kind of feature in the roadmap? [21:45] Do you guys recommends juju PPA repository even if I'm using Ubuntu 12.10 ? [21:47] As I'm reading here: https://juju.ubuntu.com/docs/getting-started.html ... ? [22:20] Guys, the hostname that I'm seeing at MaaS Web GUI is intended to be resolved (with managed DNS enabled)? I'm trying to figure out the IPs of my nodes but I can't... The only way I find my IP is looking into DHCP syslog... =( [23:27] bug #1067929 [23:27] Launchpad bug 1067929 in maas (Ubuntu) "maas-region-celeryd connects to the wrong queue." [Undecided,New] https://launchpad.net/bugs/1067929