kube.conf doesnt exist when charm reports active/idle

[Alexander Balderson]

On vsphere, as soon as juju wait detects that all units are active/idle, we scp the config file from the master/leader unit. Recently we've had a handful of runs fail because the config file doesnt exist. This may exist on other silo's however we haven't had runs on them since seeing this on vsphere.

2020-11-12-02:53:01 root DEBUG [localhost]: juju scp -m foundation_vsphere:kubernetes kubernetes-master/0:config /home/ubuntu/project/generated/kubernetes/kube.conf
2020-11-12-02:53:03 root ERROR [localhost] Command failed: juju scp -m foundation_vsphere:kubernetes kubernetes-master/0:config /home/ubuntu/project/generated/kubernetes/kube.conf
2020-11-12-02:53:03 root ERROR [localhost] STDOUT follows:
ERROR exit status 1 (scp: config: No such file or directory)

2020-11-12-02:53:03 root ERROR [localhost] STDERR follows:
None

a link to runs with this error will follow.

[Alexander Balderson (asbalderson)]

Runs which this this bug can be found here:

https://solutions.qa.canonical.com/bugs/bugs/bug/1904028

[Alexander Balderson (asbalderson)]

After further investigation, i believe this is a bug with FCE