harlowja | smoser the conversion of that json into network config, looking into that, is that network config yaml made by juju somewhere at some point? and thats why the translation? | 00:06 |
---|---|---|
harlowja | or maas | 00:06 |
harlowja | seems like https://raw.githubusercontent.com/mgagne/cloud-init-fedora-pkg/epel7/cloud-init-0.7.5-network-info-support.patch isn't really needed (or a large part of it goes away) due to that conversion and stuffs | 00:07 |
* harlowja wonders why the network json stuff wasn't == this network yaml stuff ? | 00:09 | |
mgagne | harlowja: I would like to make clear that I'm not the original author of this patch. The original version can be found here: https://github.com/jayofdoom/cloud-init-fedora-pkg/blob/master/cloud-init-0.7.5-onmetal-configdrive.patch I adapted it to support the "real" network_data.json location instead of loading it from vendor_data.json | 00:27 |
harlowja | jayofdoom, lol | 00:27 |
harlowja | nice | 00:27 |
harlowja | mgagne np, there exists some kind of ubuntu network yaml stuff that the current code in cloud-init is converting network json into | 00:28 |
harlowja | and i'm more or less trying to figure out what that yaml thing is, ha | 00:28 |
mgagne | never heard of such yaml | 00:28 |
harlowja | cause it appears to convert network_data.json ---> in memory format closer to network yaml format --> then this gets turned into a yaml file (that then gets processed by?) | 00:29 |
harlowja | aka http://bazaar.launchpad.net/~cloud-init-dev/cloud-init/trunk/view/head:/cloudinit/net/network_state.py (this thing) | 00:29 |
harlowja | the end result of that thing seems to be a yaml file that is written out (which is different from the other yaml file?) that something then processes | 00:30 |
harlowja | just not quite sure that that thing is, ha | 00:31 |
harlowja | (and maybe that thing is already in rhel/fedora?) | 00:31 |
harlowja | or maybe its just a ubuntu thing | 00:34 |
harlowja | but i'm guessing smoser knows | 00:36 |
mgagne | I'm not sure why YAML is needed. I think it's only for debug purposes. | 00:37 |
harlowja | possibly | 00:40 |
mgagne | commit message doesn't tell much about it which is unfortunate | 00:40 |
harlowja | ya, so there is a conversion into a in-memory format, which then gets processed @ http://bazaar.launchpad.net/~cloud-init-dev/cloud-init/trunk/view/head:/cloudinit/stages.py#L596 | 00:41 |
harlowja | which then activates a bunch of debian(?) specific output | 00:41 |
mgagne | it doesn't use the yaml | 00:41 |
mgagne | http://bazaar.launchpad.net/~cloud-init-dev/cloud-init/trunk/view/head:/cloudinit/net/network_state.py#L120 | 00:41 |
mgagne | debug | 00:41 |
harlowja | right, so maybe just for debug | 00:41 |
mgagne | and test http://bazaar.launchpad.net/~cloud-init-dev/cloud-init/trunk/view/head:/cloudinit/net/network_state.py#L391 | 00:42 |
harlowja | so from the looks of it, http://bazaar.launchpad.net/~cloud-init-dev/cloud-init/trunk/view/head:/cloudinit/net/__init__.py#L590 then does all the 'rendering' | 00:42 |
harlowja | but it sort of looks like debian network style rendering | 00:42 |
harlowja | which is oddly not in the debian.py distro file, lol | 00:42 |
mgagne | I guess you are looking for rhel support? | 00:42 |
harlowja | well https://raw.githubusercontent.com/mgagne/cloud-init-fedora-pkg/epel7/cloud-init-0.7.5-network-info-support.patch has some of that :-P | 00:43 |
harlowja | just there is a bunch of work that exists to do stuff with this format for ubuntu, just need to figure out where the rhel stuff goes in | 00:43 |
mgagne | yea, but actually implementation changed a lot in upstream and looks to have made it much more debian specific than needed. | 00:43 |
harlowja | yup | 00:43 |
harlowja | i blame smoser | 00:43 |
harlowja | lol | 00:43 |
mgagne | now something hacky looks much more appealing ;) | 00:44 |
harlowja | lol | 00:44 |
harlowja | or refactor time | 00:44 |
harlowja | i'll bug smoser when he gets back | 00:44 |
* harlowja unsure how much of http://bazaar.launchpad.net/~cloud-init-dev/cloud-init/trunk/view/head:/cloudinit/net/__init__.py is really debian specific | 00:44 | |
harlowja | seems like a larger part of it | 00:44 |
harlowja | why can't the distros just agree on a network format, lol | 00:45 |
mgagne | harlowja: I'm not sure why *anything* distro specific is in that file. should be in distro | 00:46 |
dmsimard | harlowja: I can get you a Mitaka config drive if you'd like (for rhel networking support) | 13:08 |
=== rangerpbzzzz is now known as rangerpb | ||
harlowja | dmsimard that'd be sweet | 16:30 |
harlowja | i can prob get one, but gotta check if the openstack we have is making that network json file | 16:31 |
harlowja | more samples i have the better :) | 16:31 |
dmsimard | harlowja: https://dmsimard.com/disk.config | 16:32 |
harlowja | thx | 16:34 |
harlowja | i gotta start ripping apart the cloud-init net stuff soon | 16:34 |
harlowja | (cause it seems ubuntu specific) | 16:35 |
harlowja | i mean debian | 16:35 |
smoser | harlowja, it does not have a rh renderer | 16:55 |
smoser | but we want one | 16:55 |
harlowja | smoser ya :-P | 16:56 |
smoser | and i have some other changes tat are voing to be necessary | 16:56 |
harlowja | just unsure what is debian specific and what isn't, ha | 16:56 |
smoser | relly need to change some things more to do this right. | 16:56 |
harlowja | ?? | 16:56 |
smoser | the datasource searching needs to chnage some really. | 16:57 |
harlowja | :) | 16:57 |
smoser | here. i'll point you at a branch in a bit with some docs on what i want to change. | 16:57 |
harlowja | cool | 16:58 |
harlowja | smoser but ya, datasources are ummm, interesting | 17:01 |
harlowja | ha | 17:01 |
harlowja | smoser do u want parallel searching, multiple datasource merging, something else :-P | 17:23 |
harlowja | or just redo the whole thing, lol | 17:26 |
harlowja | smoser whats up with the net stuff having references to 'Curtin' ? | 17:29 |
harlowja | whats 'Curtin'? | 17:29 |
smoser | curtin is the curt installer | 17:40 |
smoser | harlowja, | 17:40 |
harlowja | is curt a person? | 17:40 |
harlowja | lol | 17:40 |
harlowja | curt cobain | 17:40 |
smoser | curtin (the curt installer) is a "fast path" installer designed to install Ubuntu quickly. | 17:40 |
smoser | It is blunt, brief, snappish, snippety and unceremonious. | 17:40 |
harlowja | are u in marketing now? | 17:41 |
harlowja | lol | 17:41 |
smoser | harlowja, the goal is to move all that network stuff out into its own library | 17:41 |
harlowja | that'd be nice | 17:41 |
smoser | but under time pressure we just copied from curtin to cloud-init | 17:41 |
harlowja | has time pressure been fixed? | 17:41 |
harlowja | like if i do some rhel stuff, should i just put it in curtin? | 17:41 |
harlowja | (or maybe i can find someone from RH, lol) | 17:42 |
smoser | harlowja, heres a crappy dump of my thoughts http://paste.ubuntu.com/16222645/ | 17:47 |
mgagne | smoser: FYI, I opened the bug: https://bugs.launchpad.net/cloud-init/+bug/1577982 | 17:57 |
smoser | mgagne, yea, i saw. i reproduced . | 17:57 |
mgagne | smoser: let me know if you need access to a test instance | 18:13 |
mgagne | I can find a way to make that happen | 18:13 |
smoser | harlowja, if you do rh stuff do it in cloud-init | 18:31 |
smoser | and then we can move it there. | 18:31 |
harlowja | k | 18:31 |
smoser | did you read that paste above ? | 18:31 |
harlowja | ya, looked over a little :-P | 18:31 |
smoser | what do you think ? | 18:32 |
smoser | i think the consolidation of dsmode (basically making it go away) is a big simplification | 18:32 |
smoser | and also then a network datasource (such as openstack metadata) can look locally and say "yes, i'm the datasource". | 18:33 |
smoser | (by looking at something in dmi information) | 18:33 |
smoser | even though it does not have access to the network net. | 18:33 |
harlowja | ya, seems fair to me | 18:35 |
dmsimard | oh, and when you do that "rh stuff", let me know and I can test it :p | 18:44 |
=== rangerpb is now known as rangerpbzzzz |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!