Comment 2 for bug 1845176

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

Reviewed: https://review.opendev.org/687922
Committed: https://git.openstack.org/cgit/openstack/neutron/commit/?id=83d7eb961dc8a5d4e815bc3f110f8523a532caeb
Submitter: Zuul
Branch: master

commit 83d7eb961dc8a5d4e815bc3f110f8523a532caeb
Author: Rodolfo Alonso Hernandez <email address hidden>
Date: Tue Oct 15 15:51:40 2019 +0000

    Add OVS QoS driver cache for minimum bandwidth rules

    1. Add OVS QoS driver cache
    Added minimum bandwidth rules cache in the OVS QoS driver. If a new
    port is detected in the integration bridge, managed by the OVS agent
    driver, the QoS extension is applied (if configured).

    If this new port does not have a QoS rule, by default the QoS
    configuration is reset in this port. Because this port has no Queue
    related, the call "QosOVSAgentDriver.delete_minimum_bandwidth" should
    not execute the rule cleanup.

    This QoS rule cache implementation per porti is currently used in the
    OVS QoS driver for the DSCP and the bandwitdh limit rules.

    2. If a Queue cannot be deleted, log the QoS registers using it
    If a Queue is still used in a QoS register in the OVSDB, it cannot
    be deleted. With the current design, only one QoS rule is created in
    the database and several Queues are assigned to it.

    If something external to the OVS agent is handling the Queues and by
    mistake a Queue is assigned to other QoS registers, this error is now
    logged before raising the exception.

    Closes-Bug: #1845176
    Change-Id: Ia9077fc20e4ca360819a2e368c8c1f9250e5a6d8