Playbook stalls when switching from SSH-over-Tor to SSH-over-LAN.

[zenmonkeykstop]

Description

During the 1.0.0-rc2 QA case:

- [ ] Now disable SSH over Tor, rerun `securedrop-admin install` and `./securedrop-admin tailsconfig`:

The install playbook stalls on the task "Force a reboot conditionally, when tor_over_ssh status changed". The reboot happens, but afterward the servers are still accessible over Tor and not via Direct IP.

Steps to Reproduce

• installed a v2 production system
• enabled v2+v3 via a second ./securedrop-admin install
• disabled ssh-over-tor via sdconfig and a third install

Expected Behavior

install completes successfully and servers are available via direct IP after tailsconfig has run.

Actual Behavior

• install stalls at task mentioned above.
• if admin runs install a second time, SSH-over-Tor is also disabled, locking them out of the servers (because the listen address for sshd changes and the ssh/sshv3 hidden services are not added in sshd_config and torrc respectively.)

[emkll]

I believe this is fixed in #4794, due the changes to the reboot task