[9.2] Neutron QoS bandwith is lower than expected

Bug #1647386 reported by Yury Tregubov
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Fuel for OpenStack
Won't Fix
Medium
MOS QA Team
Mitaka
Confirmed
Medium
MOS QA Team

Bug Description

Since about 9.2 snapshot #541 the following scenario is failed on step #9
        """Check traffic restriction between vm for different max-burst
            parameter

            Scenario:
                1. Create net01, subnet
                2. Create router01, set gateway and add interface to net01
                3. Create new policy: neutron qos-policy-create policy_1
                4. Create new rule:
                    neutron qos-bandwidth-limit-rule-create rule-id policy_1 \
                    --max-kbps 4000 --max-burst-kbps 300
                5. Update net01 with --qos-policy parameter
                6. Boot ubuntu vm1 in net01 on compute-1
                7. Boot ubuntu vm2 in net01 on compute-1
                8. Start iperf between vm1 and vm2
                9. Check that egress traffic on vm1 must be eq
                    --max-kbps + --max-burst
                10. Update rule for vm port:
                    neutron qos-bandwidth-limit-rule-update rule-id bw-limiter \
                    --max-kbps 6000 --max-burst-kbps 500
                9. Check that egress traffic on vm1 must be eq
                    --max-kbps + --max-burst
            """

Expected amount of traffic is 4403200. But got only 1363148.

Test log example is: http://172.18.173.8:8080/jenkins/view/MOS-INT%209.X/job/9.x_VxLAN/172/testReport/junit/mos_tests.neutron.python_tests.test_qos/TestSingleCompute/test_traffic_restriction_with_max_burst__838298__/

Diagnostic snapshot is: https://drive.google.com/a/mirantis.com/file/d/0B1Crk-sAvGanYWp2czBTV1NuZ1k/view?usp=sharing

Changed in fuel:
milestone: none → 9.2
importance: Undecided → High
Changed in fuel:
importance: High → Medium
assignee: nobody → Fuel Sustaining (fuel-sustaining-team)
status: New → Confirmed
Revision history for this message
Yury Tregubov (ytregubov) wrote :
description: updated
Dmitry Pyzhov (dpyzhov)
Changed in fuel:
assignee: Fuel Sustaining (fuel-sustaining-team) → MOS Neutron (mos-neutron)
Revision history for this message
Alexander Ignatov (aignatov) wrote :

@Yury Tregubov, I've looked at sources of test "'838298'" here https://github.com/Mirantis/mos-integration-tests and may say that initial assumption in test that bandwidth is near to infinity is wrong.

Before any enablement of any value of any qos policy in test you have to check initial bandwidth of networks you are using in your test setup. It means that in scenario "'838298'" before step 1 you need to measure current network bandwidth otherwise this test will always fail if initial bandwidth was lower than the value (4000 in test case) in qos policy since initial bw never gets higher and it doesn't depend on qos values. It depends on hardware and openstack configurations like different MTU values, jumbo frames enables, type of segmentation (vlans, vxlans), other datapath acceleration techniques.

Changed in fuel:
assignee: MOS Neutron (mos-neutron) → MOS QA Team (mos-qa)
Revision history for this message
Alexander Ignatov (aignatov) wrote :

Moved this bug to mos qa since testcase is Invalid in its current logic.

Roman Vyalov (r0mikiam)
Changed in fuel:
status: Confirmed → Won't Fix
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.