This means that travis jobs should now be nice and green in github, but we still test the poorly labeled test "latest-epel-centos7" (which tries to run fgci-ansible playbooks with master of all the roles in requirements.yml - and there are some issues when doing that causing it currently to fail).
Bonus fix: make the sshd role work in docker too. The ansible template module to configure sshd in that role has a validation. And that validation command fails because for some reason there are no sshd host keys generated at that stage. I assume this is a docker issue, but.
I didn't check if this is an issue for real hardware.
This means that travis jobs should now be nice and green in github, but we still test the poorly labeled test "latest-epel-centos7" (which tries to run fgci-ansible playbooks with master of all the roles in requirements.yml - and there are some issues when doing that causing it currently to fail).
Bonus fix: make the sshd role work in docker too. The ansible template module to configure sshd in that role has a validation. And that validation command fails because for some reason there are no sshd host keys generated at that stage. I assume this is a docker issue, but.
I didn't check if this is an issue for real hardware.