hamalq | falcojr: we are trying to keep the default removing select_hostname from rhel will cause issues with default behaviour | 00:26 |
---|---|---|
falcojr | I wasn't suggesting changing default behavior | 00:27 |
falcojr | I'll write up a diff | 00:28 |
hamalq | the idea if we remove the select_hostname from rhel ( which prefer fqdn ) it will be back to distro select_hostname ( which prefer hostname) | 00:33 |
hamalq | and freedbsd the same which will cause the behavior change | 00:34 |
hamalq | falcojr: ^^ | 00:34 |
falcojr | hamalq: commented on the PR | 00:40 |
hamalq | how is that change different than what i already changed | 00:44 |
hamalq | ```if util.get_cfg_option_bool(self._cfg, "prefer_fqdn_over_hostname",` will not have the CFG since it send to cloud.Cloud right | 00:45 |
hamalq | falcojr: done tthe change u requested although the main ticket was suggesting rmoving the select_hostname from rhel.py and was refused | 00:59 |
falcojr | That's because it changed the functionality to do something different. This doesn't do that in absence of the config option, does it? | 01:05 |
falcojr | If it does, then ignore my suggestion 😁 | 01:06 |
hamalq | ok i will trust on u on that | 01:07 |
hamalq | anyway the change is there | 01:07 |
=== cpaelzer_ is now known as cpaelzer | ||
=== hjensas|afk is now known as hjensas | ||
mbarroso | hello all; is it possible to specify multiple data sources ? we have a use case for kubevirt where we are injecting ssh-keys over ConfigDrive metadata but would like to use NoCloud networkData (the format is a lot nicer imo). | 10:09 |
smoser | mbarroso: it is not possible. | 13:26 |
smoser | my first thought was that cloud-init 'net-convert' could convert from network-config-v1 or v2 (netplan) to openstack. | 13:29 |
smoser | (which woudl allow you to just have it do the conversion for you) | 13:29 |
smoser | but net-convert does only outputs eni,netplan,sysconfig. its not full A<->B. | 13:30 |
mbarroso | smoser: thanks :) but other than the conversion, would it be possible for cloud-init on bootup to pick up data from 2 different data sources ? e.g. get the userdata (or metadata) from one of them, and the networkdata from another ? | 13:44 |
mbarroso | rharper: sorry to ping you personally, but do you happen to know if it is possible for cloud-init to pick up data from 2 different data sources on vm bootup? e.g. get the userdata (or metadata) from one of them, and the networkdata from another ? | 16:18 |
blackboxsw | mbarroso: it's not really in the general architecture of cloud-init, some datasources have grown the ability to "see/collect" metadata from two different potential sources (like flat files in a /var/lib/cloud/seed directory) but generally the metadata is taken exclusively from one discovered "source" . | 16:28 |
blackboxsw | So trying to merge 2 separate datasource metadata and prioritize which metadata overrides other data is probably a fairly complex feature to design in cloud-init proper. | 16:29 |
blackboxsw | right now most datasources hard-code where they grab user-data, network-data and vendor-data. There isn't a configuration primitive defined that allows us to override where to grab the data "outside" of the discovered datasource logic which is unique to each datasource | 16:30 |
mbarroso | blackboxsw: thanks for your reply. Guess we'll have to port the network config to the openstack format when we use that ssh injection feature. thanks again for the reply. | 16:42 |
hamalq | blackboxsw: falcojr and me fixed the integration test problem can i get +1 https://github.com/canonical/cloud-init/pull/859 | 17:20 |
blackboxsw | hamalq: great thanks for working on this. let | 17:24 |
blackboxsw | let's see if we have any loose ends here. | 17:24 |
hamalq | sure thanks | 17:24 |
rharper | blackboxsw: for approved PRs, do you just click the squash and merge button these days? I | 19:31 |
rharper | looking at https://github.com/canonical/cloud-init/pull/866 | 19:31 |
rharper | just doc change, but wanted to check before I "push the button" | 19:31 |
blackboxsw | rharper: I realize we hadn't doc'd that process. For I merge we manually click the squash and merge button and manually cut-n-paste the desired comment msg into the PR ensuring it's got the right line width and no leading spaces etc. | 19:32 |
blackboxsw | ultimately we should instrument logic to do this for us as it's error-prone | 19:33 |
blackboxsw | currently no github actions that handle a custom label etc. | 19:33 |
rharper | yeah ... I miss just "push the button" / autoland ; and the Co-authored line? keep or remove ? | 19:33 |
blackboxsw | rharper: I drop the Co-authored line in almost all cases if it was inconsequential (mostly github adds it for anyone you clicked the "update branch" button) | 19:34 |
rharper | ok | 19:34 |
rharper | and you leave the (#XXX) in the subject ? | 19:34 |
rharper | doc: Replace remaining references to user-scripts as a config module (#866) | 19:35 |
blackboxsw | yep | 19:35 |
rharper | ok | 19:35 |
rharper | cool, thanks for the step-by-step | 19:35 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!