[qos]set running vms ingress-bandwidth-limit, then delete running vm,but ovs queue is still residual

[Zachary Ma]

1.create vm:
[root@172e18e211e96 ~]# nova interface-list mzx4
+------------+--------------------------------------+--------------------------------------+--------------+-------------------+
| Port State | Port ID | Net ID | IP addresses | MAC Addr |
+------------+--------------------------------------+--------------------------------------+--------------+-------------------+
| ACTIVE | f4ae0545-ca86-40c5-b2f7-064e521f13db | 1464fb8c-3879-4e2a-9833-3aa0882285d5 | 5.5.5.11 | fa:16:3e:57:cb:ba |
+------------+--------------------------------------+--------------------------------------+--------------+-------------------+
2. Set the policy on the port.
[root@172e18e211e96 ~]# openstack port set --qos-policy bw-limiter f4ae0545-ca86-40c5-b2f7-064e521f13db

and check ovs by cmd：
[root@172e18e211e9 ~]# ovs-vsctl list queue
_uuid : 1e4f2c2d-4116-4484-8eb6-5b13d9de649f
dscp : []
external_ids : {id="tapf4ae0545-ca", queue_type="0"}
other_config : {burst="10000000", max-rate="100000000"}

3.then delete vm in running status,and check ovs again:
[root@172e18e211e9 ~]# ovs-vsctl list queue
_uuid : 1e4f2c2d-4116-4484-8eb6-5b13d9de649f
dscp : []
external_ids : {id="tapf4ae0545-ca", queue_type="0"}
other_config : {burst="10000000", max-rate="100000000"}

tapf4ae0545-ca has been removed in ovs, but ovs queue is still residual!!

[Slawek Kaplonski (slaweq)]

I can confirm that both qos and queue is not removed from ovs in such case

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/515055

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/515566

[OpenStack Infra (hudson-openstack)]

Change abandoned by Slawek Kaplonski (<email address hidden>) on branch: master
Review: https://review.openstack.org/515055
Reason: Better solution IMHO was proposed in https://review.openstack.org/#/c/515566/

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/pike
Review: https://review.openstack.org/525888

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/515566
Committed: https://git.openstack.org/cgit/openstack/neutron/commit/?id=ee423e1fa035146904eb8cdd78660cf1cafa64e8
Submitter: Zuul
Branch: master

commit ee423e1fa035146904eb8cdd78660cf1cafa64e8
Author: Zachary <[e-mail]>
Date: Fri Oct 27 10:13:58 2017 +0800

    [Qos] Fix residues of ovs in ingress bw limit

    When we delete vm port with attached QoS policy,
    it is just doing nothing if vif_port does not exist.

    This is fine for egress bandwidth limit as it is configured
    directly on vif_port in OVS.

    For ingress bw limit however it uses additional records in
    Openvswitch database: qos and queue. Those records are not
    cleaned up in such case.

    This patch also records port in self.ports in the case of
    bandwidth limit rules, just as in the case of dscp rules.
    Never execute port clear if vif_port not exists. Finally, ovs
    driver can clean such qos and queue records

    Change-Id: Iddeb49e1e6538a178ca468df0fdf9e0617ca4f1c
    Closes-Bug: #1726732

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/neutron 12.0.0.0b2 development milestone.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/525888
Committed: https://git.openstack.org/cgit/openstack/neutron/commit/?id=40a811bea47e02c403e20150cb31de1be952db1e
Submitter: Zuul
Branch: stable/pike

commit 40a811bea47e02c403e20150cb31de1be952db1e
Author: Zachary <[e-mail]>
Date: Fri Oct 27 10:13:58 2017 +0800

    [Qos] Fix residues of ovs in ingress bw limit

    When we delete vm port with attached QoS policy,
    it is just doing nothing if vif_port does not exist.

    This is fine for egress bandwidth limit as it is configured
    directly on vif_port in OVS.

    For ingress bw limit however it uses additional records in
    Openvswitch database: qos and queue. Those records are not
    cleaned up in such case.

    This patch also records port in self.ports in the case of
    bandwidth limit rules, just as in the case of dscp rules.
    Never execute port clear if vif_port not exists. Finally, ovs
    driver can clean such qos and queue records

    Change-Id: Iddeb49e1e6538a178ca468df0fdf9e0617ca4f1c
    Closes-Bug: #1726732
    (cherry picked from commit ee423e1fa035146904eb8cdd78660cf1cafa64e8)

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/neutron 11.0.3 release.