neutron

  1. Bug #1780139
  2. Comment #54

Comment 54 for bug 1780139

Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix merged to neutron (stable/stein)

Reviewed: https://review.opendev.org/708571
Committed: https://git.openstack.org/cgit/openstack/neutron/commit/?id=4916bbc5f3c3727a7f0ad822ec6c140b54e51271
Submitter: Zuul
Branch: stable/stein

commit 4916bbc5f3c3727a7f0ad822ec6c140b54e51271
Author: Bernard Cafarelli <email address hidden>
Date: Fri Feb 7 14:21:09 2020 +0100

    Re-use existing ProcessLauncher from wsgi in RPC workers

    If both are run under the same process, and api_workers >= 2, the server
    process will instantiate two oslo_service.ProcessLauncher instances

    This should be avoided [0], and indeed causes issues on subprocess and
    signal handling: killed RPC workers not respawning, SIGHUP on master
    process leading to unresponsive server, signal not properly sent to all
    child processes, ...

    To avoid this, use the wsgi ProcessLauncher instance if it exists

    [0] https://docs.openstack.org/oslo.service/latest/user/usage.html#launchers

    Change-Id: Ic821f8ca84add9c8137ef712031afb43e491591c
    Closes-Bug: #1780139
    (cherry picked from commit 13aa00026f768642f385259ba7ccddcfdad3c75f)