In functional tests of the ovsdb_monitor, sometimes during high load
of the host where tests are run, "error" message from the killed process
can be received by neutron's async_process after new process was
And that causes failure of the test as new ovsdb_monitor will not
receive events which are expected.
To workaround that issue this patch adds sleep for few seconds between
stopping old and starting new ovsdb_monitor process.
(cherry picked from commit