[09:53] <twouters> any idea what causes /var/lib/cloud/instance to become a directory instead of a symlink? i've seen this happening when /var filesystem is full (cloud-init fails with OSError: [Errno 28] No space left on device) and fails on subsequent reboots with IsADirectoryError: [Errno 21] Is a directory: '/var/lib/cloud/instance'
[09:53] <twouters> I understand that a full filesystem is likely to result in unexpected behaviour, but I'm trying to figure out when the path is created as a directory instead of symlink
[09:54] <twouters> I'm guessing a mkdir -p alike situation where /var/lib/cloud/instance doesn't exist as a symlink, but can't seem to spot it in the code
[14:23] <chillysurfer> thanks for the review smoser on https://github.com/canonical/cloud-init/pull/509! was there anything else you wanted to see changed or answered?
[15:02] <rharper> twouters: https://bugs.launchpad.net/cloud-init/+bug/1883903
[15:03] <twouters> oh, how could i miss that
[15:03] <rharper> it's recent fix
[15:05] <twouters> thanks, i know what to do now :)
[15:06] <rharper> twouters: cool!
[15:53] <chillysurfer> smoser: thanks for the comment on the pr! you ask about documentation, this seems like it's more helpful for the developer than the user. you think documentation in the form of comments in the get_public_ssh_keys function should suffice as long as it is verbose enough?
[15:56] <smoser> chillysurfer: some datasource
[15:56] <smoser> IBMCloud has a good docstring at the top of DAtaSourceIBMCloud
[15:56]  * smoser out
[15:58] <chillysurfer> smoser: "some datasource" -> not sure i understand? you mean document this behavior in the docstring at the top of DataSourceAzure?
[18:21] <Odd_Bloke> https://github.com/canonical/cloud-init/pull/515 <-- pytest have released 6.0.1 which fixes the pylint failures we saw with 6.0.0
[20:34] <smoser> chillysurfer: yes. that is a good place to document the general behavior of the IMDS/datasource