| bigjools | bazzer: can you try the version in the PPA here instead (https://code.launchpad.net/~maas-maintainers/+archive/stable), the maas shipped with 12.04 is broken in quite a few places. | 00:01 |
|---|---|---|
| bigjools | the one in 12.04 is due to be replaced with the PPA version soon | 00:01 |
| bazzer | bigjools: huh good to know...i tried MaaS out once when 12.04 first shipped...trying to pick it up again | 00:02 |
| bazzer | bigjools: i got fed up earlier and created /var/lib/tftpboot :) | 00:02 |
| bigjools | bazzer: cool, let me know how you get on | 00:02 |
| bigjools | the latest maas in raring has a ton more features | 00:02 |
| bazzer | bigjools: it completed whatever it was up to doing on the imports...so it can't be a bad thing :) | 00:03 |
| bazzer | bigjools: i'm adding th repository now...i just thought sticking with 12.04 with it being a LTS would be the way to go for a server... | 00:04 |
| bigjools | normally true, yes :) | 00:04 |
| bigjools | the new version is also in -proposed if you want to enable it | 00:04 |
| bazzer | i (somehow) semi got it working the other day at work...nodes stuck in commissioning...clients would boot and hang complaining of iscsi targets *shrug* | 00:06 |
| bazzer | bigjools: hey that pulled in isc-dhcp-server...not using dnsmasq now? | 00:15 |
| bigjools | no | 00:15 |
| bigjools | bind9 & isc now | 00:15 |
| bazzer | sweet...i'm old school dhcp and bind...dnsmasq confusing me to no end when i'm quickly trying to do something | 00:16 |
| bigjools | yeah, dnsmasq is more a desktop solution | 00:16 |
| bazzer | yeah, i greatly hated dealing with it even if it was extremely well documented...seems too non-server-ish | 00:17 |
| bigjools | the set up of the dhcp and dns config is all done in the maas ui now | 00:18 |
| bazzer | nice | 00:19 |
| bigjools | it's all organised into a region and multiple clusters | 00:19 |
| bigjools | each cluster has its own dhcp | 00:19 |
| bazzer | thought that is how it was supposed to be? then again i tinkered with it for a short time like i said | 00:20 |
| bazzer | well that sucked...had a lockup | 00:32 |
| bazzer | bigjools: uhh, maas-import-pxe-files pulls in armhf? | 00:44 |
| bigjools | bazzer: configure what it pulls in /etc/maas/ | 01:07 |
| bazzer | gotcha | 01:08 |
| bazzer | didn't see the mention in the docs | 01:08 |
| bazzer | seem commented out to me | 01:14 |
| bazzer | ummm I just tried to enlist a node...semi success if one considers success that it booted over the network and is sitting at some login screen | 02:21 |
| bazzer | oh wait it shutdown! | 02:21 |
| bazzer | seems i got some FQDN issues :( | 02:26 |
| bigjools | if it enlisted it should show up in the UI | 02:27 |
| bazzer | yeah it is | 02:27 |
| bigjools | are you ising IPMI? | 02:27 |
| bazzer | as (;; connection timed out; no servers could be reached master | 02:28 |
| bazzer | bigjools: dunno? | 02:28 |
| bigjools | did you install maas-dhcp and maas-dns? | 02:28 |
| bigjools | if you don't know you probably aren't :) | 02:28 |
| bazzer | well dhcp yes before you gave me that ppa and i upgraded | 02:29 |
| bigjools | these are new packages | 02:29 |
| bazzer | yeah maas-dns is missing | 02:29 |
| bigjools | you will also need to visit the cluster config page and set its details up | 02:29 |
| bazzer | already did that | 02:30 |
| bigjools | cool | 02:30 |
| bigjools | you set it to manage dns as well? | 02:30 |
| bazzer | i sure did | 02:30 |
| bigjools | cool, so just install maas-dns and all should be well | 02:30 |
| bazzer | i must say this now works oh so much better than the headache i had this afternoon | 02:31 |
| bigjools | we should probably detect it's missing and warn in that page | 02:31 |
| bigjools | yeah, the default version in 12.04 is, ummm, poor | 02:31 |
| bazzer | yes that'd be good to do for sure | 02:31 |
| bazzer | lmao | 02:31 |
| bazzer | poor | 02:31 |
| bazzer | i love that statement | 02:32 |
| bigjools | it was demoware | 02:32 |
| bazzer | yeah i understand that...but i love a challenge and whatnot | 02:34 |
| bazzer | okie...trying this one more again | 02:34 |
| bazzer | oh that "semi" working default version i got set up at work...i screwed it up so bad trying to enlist nodes that it would show one and cobbler was telling me there were 10 | 02:35 |
| === AndroUser2 is now known as redlocal | ||
| bazzer | ok um how do i delete a node i accepted and its in commisioning? | 02:55 |
| bazzer | nm | 02:56 |
| bazzer | found it | 02:56 |
| bazzer | well now a new problem, juju :) | 03:24 |
| bazzer | no matching node is available | 03:25 |
| === ehw_ is now known as ehw | ||
| roaksoax | rvba: still around | 15:25 |
| roaksoax | ? | 15:25 |
| rvba | roaksoax: yep | 15:26 |
| roaksoax | rvba: so I'm gonna be separating the ipmi autodtection code away from enlist_userdata and user_data.template. However, we discussed with julian that we would like to "inject" that script | 15:27 |
| roaksoax | rvba: so any ideas of how I could use/inject that script into the templating? | 15:27 |
| roaksoax | rvba: the idea of separating the code away is to make it available in a separate package | 15:27 |
| rvba | roaksoax: well, I see two ways to do this: one solution is to read that script in MAAS and include it in the context used to render the main template. The other is to use tempita's inheritance mechanism to compose the template using the main template and the script. | 15:31 |
| roaksoax | rvba: both options would require have it installed in maas region then. Ok, I'll think about what's best and let you know :) | 15:32 |
| rvba | roaksoax: it probably needs some thinking indeed. | 15:33 |
| roaksoax | indeed | 15:33 |
| roaksoax | rvba: btw.. the piston fixes... they are not merged in upstream yet | 15:36 |
| roaksoax | rvba: do you still want me to grab those and apply them? | 15:36 |
| rvba | roaksoax: IIRC they have been there for a long time and are still not merged upstream… right? | 15:38 |
| roaksoax | rvba: yeah 1 has been there for months, the other just for days | 15:41 |
| rvba | roaksoax: if that's the case then I think we should include them as patches to avoid having to wait for upstream to include them. | 15:41 |
| roaksoax | ok | 15:41 |
| roaksoax | my point was that we usually cherry pick patches applied upstream (and these haven't) | 15:41 |
| roaksoax | but will take care of them | 15:41 |
| rvba | ok thanks. | 15:42 |
| roaksoax | rvba: btw.. https://bitbucket.org/jespern/django-piston/pull-request/25/compatibility-fix-for-json-emitter-with/diff | 15:43 |
| roaksoax | rvba: check the commet by Vladislav | 15:43 |
| roaksoax | rvba: do you think we should use what he commented instead of django.VERSION | 15:44 |
| roaksoax | ? | 15:44 |
| rvba | roaksoax: "Why soooo looong?" ? | 15:45 |
| rvba | I think the author of that patch is a core Django dev. | 15:46 |
| roaksoax | rvba: look at the diff, there's a comment in a diff with "try: import json except ImportError from ..... | 15:46 |
| roaksoax | rvba: instead of "import django \n if django.VERSION...." | 15:46 |
| rvba | roaksoax: I prefer the usage of django.VERSION tbh. | 15:47 |
| roaksoax | rvba: ok | 15:47 |
| roaksoax | rvba: uploaded. should be available in the next couple of hours | 16:20 |
| AskUbuntu | Install nodes error (could not find kernel image) | http://askubuntu.com/q/301150 | 16:50 |
| kmyst | quick question...i'm trying to enlist nodes that have two nics (eth0 is connected to the network the maas controller is on and eth1 is connected to another network with internet access) but my commissioning is failing i think since it only brings up eth0 so how do i get around that? | 19:32 |
| smoser | hey... | 20:03 |
| smoser | kmyst, it *should* be fixed to correctly load only the nic that it booted from. | 20:04 |
| smoser | hey. in 'maas-cli admin nodes list' | 20:04 |
| smoser | is there any way to get the user-name that has a node allocated ? | 20:04 |
| smoser | afaiks no | 20:04 |
| kmyst | smoser: yeah that's what i figured but the nic it boots from has no internet access so the commissioning just sits and sits then spews a ton of apt-get errors since it can't get to the internet | 20:25 |
| smoser | ah. because it doesn't bring up the other nic | 20:28 |
| smoser | ? | 20:28 |
| kmyst | right | 20:28 |
| smoser | i'd appreciate it if you opened a bug, as your use case seems quite valid to me. | 20:29 |
| smoser | or at least i'd like to see a thoughtful response. | 20:29 |
| smoser | you can probably get thorugh this by adding an entry in /etc/network/interfaces inside the ephemeral image | 20:29 |
| kmyst | which the end result is the node ultimately shuts down and goes to the ready state but well...ya know :) | 20:29 |
| kmyst | good idea | 20:30 |
| smoser | shoot.... that might not work* | 20:30 |
| smoser | probably wont. | 20:30 |
| kmyst | drat | 20:31 |
| smoser | as we have some clever code in the initramfs that dynamically writes /etc/network/interfaces | 20:31 |
| kmyst | ah | 20:32 |
| smoser | based on what BOOTIF was given/ | 20:34 |
| smoser | on the initramfs kenrel command line | 20:34 |
| smoser | let me see thoug | 20:34 |
| kmyst | k | 20:36 |
| kmyst | i just got done testing out the same thing on virtualbox | 20:36 |
| kmyst | same results | 20:36 |
| smoser | i think the easiest thing to do will be to make the commissioning user-data bring up eth1 | 20:37 |
| smoser | kmyst, what version of maas are you using ? | 20:38 |
| kmyst | uh let me check | 20:39 |
| smoser | you're also going to have to manually handle bringing up the nics in install also. | 20:39 |
| smoser | as it is also going to just bring up the nic that it sees on the cmdline | 20:39 |
| smoser | (i think) | 20:39 |
| smoser | please open a bug | 20:39 |
| kmyst | ya that's what it's doing for sure | 20:39 |
| smoser | as i honestly think this is a very reasonable setup you have. | 20:40 |
| kmyst | maas version 1.2+bzr13373-df | 20:40 |
| kmyst | i'll definitely open a bug if that's the way to go? | 20:42 |
| smoser | well, we definitely want a bug | 20:42 |
| smoser | to fix it properly | 20:42 |
| kmyst | will do then | 20:42 |
| kmyst | 1.2+bzr1373+dfsg-0ubuntu1~12.04.1~ppa1 | 20:44 |
| kmyst | from the stable ppa | 20:44 |
| smoser | ah. thats what i was looking for. | 20:46 |
| kmyst | smoser: only reason i set it up that way was i didn't care if maas ran dhcp/dns but since i already have an authoritative dhcp server on the network segment i isolated maas and the nodes and just provided the nodes with an extra nic to a lan segment that has external access | 20:47 |
| kmyst | so after experiencing the problem i fell back to testing the idea out with virtualbox...same results :) | 20:48 |
| smoser | yeah. | 20:48 |
| smoser | so one way that will probalby work... | 20:48 |
| smoser | in the ephemeral images, you can add an upstrat job that runs on 'starting cloud-init' | 20:49 |
| smoser | that just does whatever you need to do to 'ifup eth1' | 20:49 |
| smoser | (it will be more than just 'ifup eth1' as there wont be an entry in /etc/network/interfaces for eth1, but you could just append one and then ifup) | 20:49 |
| smoser | that make sense? | 20:49 |
| kmyst | completely | 20:50 |
| smoser | do that in the ephermal images, and then also make it happen in the installer | 20:50 |
| smoser | so that when it comes up it gets your handywork aplied | 20:50 |
| kmyst | so the disk.img in /var/lib/maas/ephemeral/precise/ephemeral/amd64/20121008 right? | 20:52 |
| kmyst | ls | 20:52 |
| AskUbuntu | How to preserve custom configurations across reboots when using juju and maas? | http://askubuntu.com/q/301241 | 20:54 |
| AskUbuntu | Juju deploy issue : Connection was refused by other side | http://askubuntu.com/q/301243 | 20:54 |
| kmyst | smoser: where's the installer? | 21:13 |
| AskUbuntu | sudo maas-import-isos fails | http://askubuntu.com/q/301280 | 23:40 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!