=== imthewherd is now known as cgseller_is_away | ||
dimitern | blake_r, rvba, meeting? | 13:01 |
---|---|---|
rvba | dimitern: sorry, I have a conflicting meeting. Can we postpone? | 13:01 |
rvba | dimitern: unless mpontillo can cover for me. | 13:02 |
mpontillo | rvba: I'm here... but I think the only thing we need to discuss is "the fan"... and I'd want you to be there for that, rvba | 13:02 |
rvba | mpontillo: okay, so let's postpone if that's okay with you dimitern. | 13:03 |
dimitern | rvba, fine with me | 13:04 |
rvba | dimitern: can we talk in, like 1h? | 13:04 |
dimitern | rvba, I won't manage | 13:08 |
rvba | dimitern: okay, let's pick another time then. | 13:09 |
dimitern | rvba, but mpontillo will brief you up :) | 13:09 |
kiko | negronjl, smoser, rvba: for your review, http://askubuntu.com/questions/636837/are-there-examples-of-custom-installation-scripts/636867#636867 | 14:09 |
smoser | hmm... | 14:17 |
smoser | "As background, MAAS uses cloud-init as part of its installation process; cloud-init has the concept of preseeds, which in MAAS describe actions to be run at different points in the node lifecycle. " | 14:18 |
smoser | that is odd | 14:18 |
kiko | smoser, it's because I don't know what I am talking about. could you help clarify? | 14:28 |
kiko | smoser, is it odd but factually correct? | 14:28 |
smoser | kiko, no. | 14:30 |
smoser | let me try to write something | 14:30 |
kiko | smoser, thanks | 14:57 |
smoser | kiko, http://paste.ubuntu.com/11725176/ | 14:59 |
smoser | so i started writing some information for you... so that i could then clarify the doc. | 15:00 |
smoser | but that got long winded... so you might be able to use some of that, it might be useful for your general undersanding... | 15:00 |
kiko | thanks! | 15:03 |
smoser | http://paste.ubuntu.com/11725201/ <-- just without long lines | 15:04 |
smoser | http://paste.ubuntu.com/11725206/ <-- without *any* long lines | 15:04 |
negronjl | smoser, kiko: This works | 15:05 |
kiko | smoser, I still think that first paragraph is correct; I mean, is there something factually wrong? | 15:09 |
kiko | first MAAS does use cloud-init, correct? | 15:09 |
kiko | second, cloud-init has preseeds, yes? | 15:09 |
kiko | I guess the third piece is not very correct in that the actions in the preseeds are run when booting the ephemeral environment and in the first system boot | 15:10 |
smoser | "preseeds" is a bad word. | 15:11 |
smoser | "preseed" likely implies d-i. | 15:12 |
kiko | smoser, what are they called in curtin? | 15:12 |
smoser | cloud-init doesn't really use 'preseeds'. it uses user-data. | 15:12 |
kiko | I guess I'm confused because they go into a directory called preseeds? | 15:12 |
smoser | curtin reads curtin config. | 15:13 |
kiko | hmm | 15:13 |
kiko | okay, so the yaml-ish files are curtin config? | 15:13 |
smoser | curtin reads curtin config | 15:13 |
smoser | d-i reads d-i config | 15:13 |
smoser | d-i config (and debian package install time configuration) is known as "preseed" | 15:14 |
smoser | i'd avoid the use of preseed when not talking about d-i or debian package preseeds. | 15:14 |
smoser | as that would most certainly confuse me. | 15:14 |
smoser | ie, google 'preseed linux' : https://en.wikipedia.org/wiki/Preseed | 15:15 |
mup | Bug #1465722 opened: 1.8.0 Machine details styling <ui> <MAAS:In Progress by ricgard> <https://launchpad.net/bugs/1465722> | 15:37 |
mup | Bug #1465726 opened: 1.8.0 machine details button sizes not styled dynamically <ui> <MAAS:Triaged by ricgard> <https://launchpad.net/bugs/1465726> | 15:37 |
mup | Bug # opened: 1465732, 1465735, 1465736, 1465737, 1465739, 1465740 | 15:46 |
mup | Bug # changed: 1465732, 1465735, 1465736, 1465737, 1465739, 1465740 | 15:58 |
mup | Bug # opened: 1465732, 1465735, 1465736, 1465737, 1465739, 1465740, 1465742, 1465743 | 16:07 |
mup | Bug #1465742 changed: 1.8.0 Table design styles <ui> <MAAS:Triaged by ricgard> <https://launchpad.net/bugs/1465742> | 16:10 |
mup | Bug #1465743 changed: 1.8.0 Tags spacing <ui> <MAAS:Triaged by ricgard> <https://launchpad.net/bugs/1465743> | 16:10 |
mup | Bug #1465742 opened: 1.8.0 Table design styles <ui> <MAAS:Triaged by ricgard> <https://launchpad.net/bugs/1465742> | 16:16 |
mup | Bug #1465743 opened: 1.8.0 Tags spacing <ui> <MAAS:Triaged by ricgard> <https://launchpad.net/bugs/1465743> | 16:16 |
kiko | smoser, how does cloud-init fit in with curtin and d-i? Is it that both curtin and d-i provide stuff for cloud-init to run, and the way you provide that stuff is different for curtin and d-i? | 16:17 |
smoser | cloud-init has almost nothing to do with d-i | 16:18 |
smoser | the only overlap between them is that maas installs the cloud-init package during a d-i install and "preseed"s the configuration of the maas datasource . so that on first boot, cloud-init knows where to find maas. | 16:19 |
cholcombe | maas: for block devices who does the formatting? maas or the juju charm? | 17:01 |
kiko | cholcombe, the system installation block device or additional block devices? | 17:12 |
kiko | cholcombe, for the former, the OS install done by MAAS (with di-or-curtin/cloud-init) does the formatting/resizing | 17:12 |
kiko | for the latter the charm does it | 17:12 |
cholcombe | i see yeah that's exactly what i needed to know | 17:12 |
cholcombe | i'll take that into account | 17:13 |
cholcombe | i was referring to additional block devices | 17:13 |
kiko | cholcombe, check out the ceph charm, I believe it knows how to handle additional block devices on the OST | 17:13 |
cholcombe | good point. there's some code in there to test if a block device is legit. that's prob why that exists | 17:14 |
kiko | oh, to avoid clobbering the system block device for instance? | 17:14 |
cholcombe | possibly or just to make sure it's not getting a bogus drive specified | 17:15 |
kiko | smoser, how come the user_data you supply to node start is always considered to be "late_commands" for curtin? | 18:39 |
kiko | or is that also not exactly right? | 18:39 |
kiko | ah, I think I understand now | 18:41 |
smoser | kiko, its 2 separate cloud-inits entirely. | 18:48 |
smoser | and actually 2 differnet cloud-init endpoints (or at least it used to be | 18:48 |
smoser | i tried to clarify that in in http://paste.ubuntu.com/11725206/ | 18:49 |
smoser | line 72 there. | 18:49 |
kiko | smoser, here's an updated attempt, hopefully simpler: http://askubuntu.com/questions/636837/are-there-examples-of-custom-installation-scripts/636867#636867 | 18:56 |
smoser | kiko, i'd strip "MAAS uses cloud-init as part of its installation process." as you dont really mention that in any other way, and its just confusing. the end user of maas doesn't care that maas uses cloud-init inside the installation process. | 18:58 |
smoser | also, the first paragraph somewhat implies that cloud-init is not available to the user if d-i is used for isntallation | 18:58 |
smoser | i dont want to nit-pick, you're welcome to ignore that if you think its bike-shedding | 19:00 |
kiko | smoser, yeah, I think I see your point, but it leads to more questions :) | 19:44 |
kiko | smoser, for instance, does user_data do something when you are deploying with d-i? :) | 19:44 |
smoser | kiko, it does something *after* you deploy with d-i | 19:45 |
smoser | whihc is what the end user cares about | 19:45 |
smoser | from their perspective maas just gives them a machine with that has cloud-init inside it. | 19:45 |
smoser | same as ec2. | 19:45 |
kiko | smoser, so you could use the user_data mechanism regardless of whether curtin or d-i is being used? | 19:48 |
smoser | yes | 19:49 |
kiko | oh. | 19:50 |
kiko | could I delete the "Depending on which installer you are using.." paragraph then? | 19:50 |
kiko | ah | 19:50 |
kiko | but of course, the file in /etc/maas/preseeds differss | 19:51 |
smoser | kiko, right. the mechanism for customizing installation differs between installers (they're different installers) | 19:55 |
smoser | but the mechanism for communicating with the installed Ubuntu is consistent. | 19:56 |
plars | what's the best way of importing and booting an image from an iso? | 21:31 |
plars | I've found lp:curtinator and managed to import the results of running that against an daily wily snapshot. And it shows up under "images" in maas in the "Generated Images" section | 21:32 |
plars | but I can't find where to tell it to boot that image. My preference would be to boot from the cli if that's possible | 21:32 |
plars | I'm currently on 1.7.5+bzr3369-0ubuntu1~trusty1 from the stable ppa | 21:33 |
kiko | plars! | 21:34 |
plars | Heya kiko! | 21:35 |
kiko | plars, let me check the API docs, but basically you specify a distro series name when telling a node to start | 21:35 |
plars | kiko: when importing it, I did somthing like: maas testmaas boot-resources create name=ubuntu/deskwily architecture=amd64/generic content@=/tmp/wily-desktop-amd64.iso.tar.gz | 21:37 |
kiko | oh | 21:37 |
kiko | I think you just say | 21:37 |
plars | kiko: and tried booting with something like: maas testmaas node start node-5337016e-3d56-11e4-a8dd-001cc08caad0 distro_series=deskwily | 21:37 |
kiko | maas plars node uuid start distro_series=deskwily no? | 21:37 |
kiko | right | 21:37 |
plars | kiko: right, I would have though so, but it says it's an invalid series | 21:37 |
plars | I purposely gave it a bogus series name so it wouldn't conflict with anything | 21:38 |
plars | surely that's not hardcoded to known releases though, right? | 21:38 |
plars | I wouldn't think so, I've heard of people booting centos, rhel, etc | 21:38 |
kiko | plars, correct | 21:38 |
kiko | plars, does that option show up in the UI at all? | 21:38 |
plars | kiko: well, I see the image under images, but if I acquire a node and look at the options it gives me for the image, I don't see it there | 21:39 |
kiko | that is not good | 21:39 |
kiko | plars, can you put a screenshot of /images somewhere? | 21:41 |
plars | kiko: sure, one sec | 21:41 |
plars | kiko: http://imgur.com/rPVZKP5 | 21:45 |
=== cgseller_is_away is now known as imthewherd |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!