Comment 2 for bug 1691545

Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix merged to nova (master)

Reviewed: https://review.openstack.org/465042
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=47fa88d94754fcdad6bb132b45196b4d44c0f4cd
Submitter: Jenkins
Branch: master

commit 47fa88d94754fcdad6bb132b45196b4d44c0f4cd
Author: melanie witt <email address hidden>
Date: Tue May 16 10:25:42 2017 +0000

    Cache database and message queue connection objects

    Recently in the gate we have seen a trace on some work-in-progress
    patches:

      OperationalError: (pymysql.err.OperationalError)
                        (1040, u'Too many connections')

    and at least one operator has reported that the number of database
    connections increased significantly going from Mitaka to Newton.

    It was suspected that the increase was caused by creating new oslo.db
    transaction context managers on-the-fly when switching database
    connections for cells. Comparing the dstat --tcp output of runs of the
    gate-tempest-dsvm-neutron-full-ubuntu-xenial job with and without
    caching of the database connections showed a difference of 445 active
    TCP connections and 1495 active TCP connections, respectively [1].

    This adds caching of the oslo.db transaction context managers and the
    oslo.messaging transports to avoid creating a large number of objects
    that are not being garbage-collected as expected.

    Closes-Bug: #1691545

    [1] https://docs.google.com/spreadsheets/d/1DIfFfX3kaA_SRoCM-aO7BN4IBEShChXLztOBFeKryt4/edit?usp=sharing

    Change-Id: I17e0eb836dd87aac5859f506e7d771d42753d31a