[sshd] Ensure that '/run/sshd' directory exists

There is an issue in certain LXC containers where after configuring
sshd, it fails to restart due to the '/run/sshd' directory missing. This
is caused by sshd using systemd socket activation in this environments,
and the directory being removed when the last session is closed.

This is a short them solution before the role is redesigned to take the
new configuration into account.

(cherry picked from commit 71e23a3)

CHANGELOG.rst

@@ -167,6 +167,12 @@ General
 
 - The password length can now be set using a number, not just a string.
 
+:ref:`debops.sshd` role
+'''''''''''''''''''''''
+
+- The role will make sure that the :file:`/run/sshd/` directory exists on all
+  hosts, to fix issues with :command:`sshd` daemon not restarting properly.
+
 :ref:`debops.sysctl` role
 '''''''''''''''''''''''''
 

ansible/roles/sshd/tasks/main.yml

@@ -89,7 +89,6 @@
     path: '/run/sshd'
     state: 'directory'
     mode: '0755'
-  when: ansible_service_mgr != 'systemd'
 
 - name: Ensure OpenSSH support is installed
   apt: