Comment 6 for bug 1710328

Reviewed: https://review.openstack.org/503453
Committed: https://git.openstack.org/cgit/openstack/swift/commit/?id=6b4e6aeffaf1c6f7bc69998870796b7f9230ee91
Submitter: Jenkins
Branch: stable/newton

commit 6b4e6aeffaf1c6f7bc69998870796b7f9230ee91
Author: Samuel Merritt <email address hidden>
Date: Mon Aug 14 10:41:31 2017 -0700

    Fix deadlock when logging from a tpool thread.

    The object server runs certain IO-intensive methods outside the main
    pthread for performance. If one of those methods tries to log, this can
    cause a crash that eventually leads to an object server with hundreds
    or thousands of greenthreads, all deadlocked.

    The short version of the story is that logging.SysLogHandler has a
    mutex which Eventlet monkey-patches. However, the monkey-patched mutex
    sometimes breaks if used across different pthreads, and it breaks in
    such a way that it is still considered held. After that happens, any
    attempt to emit a log message blocks the calling greenthread forever.

    The fix is to use a mutex that works across different greenlets and
    across different pthreads. This patch introduces such a lock based on
    an anonymous pipe.

    Change-Id: I57decefaf5bbed57b97a62d0df8518b112917480
    Closes-Bug: 1710328
    (cherry picked from commit 6d160797fc3257942618a7914d526911ebbda328)