[02:42] I'm looking for some clarification of the expected behavior of various data sources (particularly ConfigDrive and CloudStack) ... [02:42] ... i see that data sources split into two sets: those that override the check_instance_id method and those that don't ... [02:44] ... ConfigDrive does (and hence honors the "immutable" first boot instance) and CloudStack does NOT (and hence regenerates the instance data on each boot) ... [02:46] ... site usage includes the notion of "VM_EDIT" (i.e. mutable user data) so I'd like to understand what it might take to modify the ConfigDrive data source to optionally support this ... [02:47] ... and understand whether this mode is really seen as "pure" by design [02:48] (By the way, cloud-init is the bees knees @smoser and crew) [12:49] generally speaking, i think at this point, cloud-init really expects instance-data to be "per-instance". [18:53] falcojr: I'm trying to run the OCI tests, and it's complaining I don't have `oci` installed (which is true): should our Python dependencies be pulling that executable in? [18:55] We could, but you'd likely just get a different error. All the clouds on pycloudlib have some pre-conditions that you have already set some things up [18:56] https://pycloudlib.readthedocs.io/en/latest/clouds/oci.html [18:59] Yeah, I already had a config file, and it's still failing after re-setup (and installing the CLI, ofc). [19:28] I am attempting to configure a user, and I am having some trouble. Even though I can use meta_data.json and network_data.json, it seems that user_data isn't getting processed as expected, and all the documentation for it seems to suggest that it should be yaml? I am getting an error: __init_.py[WARNING]: Unhandled non-multipart (text-x-not-multipart) userdata: 'b'{"users": ...