Since there is no central storage for the restart actions and the tests
at the moment, they should have timestamped output to correlate events.
Using a standard format will make this easier to do programmatically,
too.
The output is sent both to stdout and a log file. Additionally, Ansible
output is suppressed now, so it doesn't pollute the console. Log times
are in UTC, to avoid timezone mismatches between nodes (observed in
multi-node AIOs).
The `configure_logging` function was copied into the tests/keystone.py
file instead of put into a common module in order to keep that file a
standalone unit for the time being.
Change-Id: I399d1827a8559896ca87e45c00241293ccd033d2
These tools help simulate rolling downtime by taking down service
containers in a controlled fashion. While this can be done manually,
using a script to inspect the inventory for targets is much easier.
Having the rolling downtime happen automatically is also a bit less
error prone.
Tests for things like response times will be placed in the tests
directory, allowing for scripts to be written against different services
to try different things (like adding resources while rolling downtime
happens).
There is no central orchestration for the two components currently, in
order to keep things simple. Restarting containers and running the tests
is therefore best accomplished in a pair of tmux sessions.
Change-Id: I2d8e3484bbb1a71d2cd0c4124f824d49f76e5c2c
