Comment 8 for bug 1794708

Reviewed: https://review.opendev.org/641907
Committed: https://git.openstack.org/cgit/openstack/oslo.service/commit/?id=e7dd2916893157854ca0fa5f4215d536151abbb3
Submitter: Zuul
Branch: master

commit e7dd2916893157854ca0fa5f4215d536151abbb3
Author: Mohammed Naser <email address hidden>
Date: Thu Mar 7 22:09:49 2019 -0500

    restart: don't stop process on sighup when mutating

    It seems that the code for handling SIGHUP currently calls stop()
    on the service, then calls reset(), then calls start() on it again.

    This is effectively a full service restart, which breaks the whole
    point behind using SIGHUP for hot and quick reloads. It also breaks
    our downstream projects in a few ways where they lose RPC on reload
    due to the fact that they don't expect to have stop() called on a
    reset().

    This patch removes the stop and start when the restart_method is
    set to 'mutate' because in that case we should just be signaling
    the service to check for changes in its mutable config options.
    It also changes the signal sent to children in that case to
    SIGHUP, since SIGTERM will cause unnecessary restarts of child
    processes.

    The previous behavior is maintained for the 'reload' restart_method
    since that does a complete reload of the service config, which is
    not safe to do without restarting the service completely.

    Change-Id: I86a34c22d41d87a9cce2d4ac6d95562d05823ecf
    Closes-Bug: #1794708
    Co-Authored-By: Ben Nemec <email address hidden>