[15:13] <powersj> Reading the docs there is a "preserve_source_list" option to prevent sources.list from being modified. However, it looks like it does not have any effect.
[15:13] <powersj> Then I looked at a modified sources.list and at the top there is a comment about "apt_preserve_source_list", which is different thant he docs
[15:13] <powersj> 1) which "verb" should I use? 2) can I use it to prevent any changes? or only some?
[15:43] <jgrimm> powersj, i'm far from knowledgeable.. but the docs do say  "only  extra  source specifications will be written into /etc/apt/sources.list.d/* " even when preserve_source_list is true.    WRT you're question #2 (looks like only 'some' is the answer)
[15:45] <rharper> powersj: jgrimm:  recall that the base /etc/apt/sources.list is generated from a template
[15:45] <rharper> there's a default template built-in to cloud-init; or you have to provide your own template
[15:46] <rharper> combined with jgrimm comment, the preserve would only be for .d/ entries IIUC
[15:46] <rharper> cpaelzer might know more;  powersj also looking at curtin's vmtest for apt may be informative
[15:46] <jgrimm> rharper, good idea wrt curtin tests
[15:48] <jgrimm> powersj, i'm curious to know more by what you mean " there is a comment about "apt_preserve_source_list", which is different thant he docs"  .. sounds peculiar
[15:48] <powersj> rharper: jgrimm thanks, from the note in the docs made me think it would preserve sources.list not the .d/*
[15:49] <jgrimm> ah
[15:49] <jgrimm> if you think some clarification could improve things, please do submit fixup. :)
[15:49] <powersj> jgrimm: here is from a cloud-init generated source.list: https://paste.ubuntu.com/23324063/
[15:50] <powersj> so that mentions apt_preserve... but I don't see that mentioned anywhere else in the docs
[15:51] <magicalChicken> i think cloud-init should follow 'preserve_sources_list'. in cc_apt_configure.py at line 300 it checks if that's there before generating
[15:51] <rharper> powersj: that's if you snapshot and boot it in a different instance
[15:51] <rharper> then cloud-init won't re-create the templated file
[15:51] <rharper> powersj: are you seeing that sources.list is re-written when you switch instances ?
[15:52] <powersj> rharper: haven't tried that, just trying to understand all the options and which to actually use.
[15:53] <rharper> gotcha
[15:53] <magicalChicken> powersj: the apt_preserve_sources_list is rewritten to preserve_sources_list in cc_apt_configure.convert_v2_to_3_apt_format
[15:55] <powersj> magicalChicken: ah! so it is a legacy name so to speak
[15:55] <jgrimm> ah, apt_* version is old, so indeed maybe useful to fix up
[15:55] <jgrimm> yeah
[15:55] <powersj> "Convert old to new keys..."
[15:55] <magicalChicken> powersj: yeah, the new format is everything under a general apt: key
[15:56] <jgrimm> so looks like a minor cleanup bug for the emitted text
[16:22] <harlowja> smoser did u ever get my cla email ?
[16:24] <smoser> where did you send ? more contexst ?
[16:24] <harlowja> godaddy never got a copy of theres
[16:25] <harlowja> let me figure out where i sent it
[16:25] <harlowja> somewhere
[16:25] <harlowja> lol
[16:25] <harlowja> (canonical ccla for godaddy)
[16:26] <harlowja> Scott Moser <smoser@ubuntu.com>
[16:26] <harlowja> that one
[16:26] <harlowja> sent again :-P
[16:27] <smoser> k
[16:47] <harlowja> smoser so mdorman may have some resources for VMs :)
[16:47] <harlowja> if we can just figure out what we want :)
[16:47] <harlowja> we/canonical/us ...
[16:49] <mdorman> hello o/
[16:51] <harlowja> o/
[16:54] <harlowja> smoser is https://public.etherpad-mozilla.org/p/cloud-init-centos-unittest still the idea for rhel?
[16:55] <mdorman> so i am not totally up to speed with the whole discussion here, but we should be able to provide you N VMs to help with this.  but if you need an openstack api endpoint to hit as a way to dynamically spin up/down vms, that’s a little more complex and i’m not sure we can accommodate that (we don’t have an OS api exposed publically today)
[16:55] <harlowja> smoser is the one up to speed :-P
[16:56] <harlowja> haha
[16:57] <smoser> see the 'mostly now'
[16:57] <smoser> https://gist.github.com/smoser/19e65095b342e98fd4d6466e4d4fa1ac
[16:57] <smoser> that works last i checked
[16:58] <harlowja> so with that kind of stuff just VMs with LXC would be fine
[16:58] <harlowja> no need for API acccess
[16:58] <harlowja> just some VMs with ability to login (for certain keys)
[16:58] <harlowja> (perhaps?)
[16:58] <smoser> actually.. so using lxc i can npretty easily run unit tests as seen there.
[16:59] <smoser> and we can do that on existing canonical hardware.
[16:59] <harlowja> k
[16:59] <smoser> can/should
[16:59] <harlowja> canonical orange boxes?
[16:59] <mdorman> :)
[16:59] <harlowja> or whatever they were called :-P
[16:59] <smoser> so at the moment i guess i dont need anything
[16:59] <smoser> right, all our boxen are orange.
[16:59] <harlowja> just like the next president!
[16:59] <harlowja> lol
[16:59] <mdorman> ehe
[17:00] <mdorman> alright, cool.  well if that’s something we can assist with, feel free to reach out to me and/or via josh.  it’s pretty easy for us to stand up
[17:00] <smoser> ok.
[17:01] <harlowja> wfm