Redundant DHCP agents assigned to networks after DHCP agents restart

[Eugene Nikanorov]

Affected versions: 6.1, 7.0

Scheduling accounts only active agents which leads to cases when more than configured number of agents get teh network.

Impact: waste of fixed ips, possible issues with namespaces consistency.

[Fuel Devops McRobotson (fuel-devops-robot)]

Fix proposed to branch: openstack-ci/fuel-6.1/2014.2
Change author: Eugene Nikanorov <email address hidden>
Review: https://review.fuel-infra.org/12847

[Fuel Devops McRobotson (fuel-devops-robot)]

Fix proposed to branch: openstack-ci/fuel-7.0/2015.1.0
Change author: watanabe isao <email address hidden>
Review: https://review.fuel-infra.org/12849

[Fuel Devops McRobotson (fuel-devops-robot)]

Reviewed: https://review.fuel-infra.org/12849
Submitter: mos-infra-ci <>
Branch: openstack-ci/fuel-7.0/2015.1.0

Commit: 330b5000add837a718efabed7c8dfda8d4586c44
Author: watanabe isao <email address hidden>
Date: Wed Oct 14 20:11:43 2015

Bug-Fix for unexpected DHCP agent redundant

This fix let DHCP scheduler not looks at only active agents, but all
available agents.
This helps db module to remove dead agent, when rebinding, properly.

Cherry-picked from: b095fa11a4df221a5d539d142bd3c0340935ad9d
Change-Id: I8174ddfae299724c05641183c2fe4c1c98c614e8
Closes-Bug: 1506198

[Kristina Berezovskaia (kkuznetsova)]

Verify on
VERSION:
  feature_groups:
    - mirantis
  production: "docker"
  release: "7.0"
  openstack_version: "2015.1.0-7.0"
  api: "1.0"
  build_number: "301"
  build_id: "301"
  nailgun_sha: "4162b0c15adb425b37608c787944d1983f543aa8"
  python-fuelclient_sha: "486bde57cda1badb68f915f66c61b544108606f3"
  fuel-agent_sha: "50e90af6e3d560e9085ff71d2950cfbcca91af67"
  fuel-nailgun-agent_sha: "d7027952870a35db8dc52f185bb1158cdd3d1ebd"
  astute_sha: "6c5b73f93e24cc781c809db9159927655ced5012"
  fuel-library_sha: "5d50055aeca1dd0dc53b43825dc4c8f7780be9dd"
  fuel-ostf_sha: "2cd967dccd66cfc3a0abd6af9f31e5b4d150a11c"
  fuelmain_sha: "a65d453215edb0284a2e4761be7a156bb5627677"
with updates.
Reproduced on the same iso without updates
neutron+vlan, 3 controller node, 2 compute node

Steps to reproduce:
1) Create many networks with script:
neutron quota-update --network 1000 --subnet 1000 --router 1000 --port 1000

for i in {9..59}
do
NET_ID=$(neutron net-create net$i | grep id | awk -F "|" '{print $3}' | head -1 | tr -d " ")
neutron subnet-create --name subnet$i net$i 10.0.$i.0/24
nova boot --flavor m1.micro --image TestVM --nic net-id=$NET_ID instance$i
sleep 10
nova delete instance$i
done

2) Check on what agents net9 is: neutron dhcp-agent-list-hosting-net net9
(should on 2 agents)
3) Disable all dhcp agents: on controller node
pcs resource disable p_neutron-dhcp-agent

4) Wait while all agents will not die with command 'neutron agent-list'
5) Enable agents: on controller node
pcs resource enable p_neutron-dhcp-agent
6) Wait some time while agents become alive
6) Repeat step 2

Expected result: net9 should be on 2 agents
Result without updates: net9 is on all 3 agents

On 7.0 with updates (update master-node, create new env, add in UI new repository and deploy env) rescheduling works correctly

[Fuel Devops McRobotson (fuel-devops-robot)]

Reviewed: https://review.fuel-infra.org/12847
Submitter: Vitaly Sedelnik <email address hidden>
Branch: openstack-ci/fuel-6.1/2014.2

Commit: ae2ac812a5d1e3068065547874e9de796a13d7b3
Author: Eugene Nikanorov <email address hidden>
Date: Thu Nov 26 17:04:12 2015

Count all agents when autoscheduling network

This will help to avoid cases when more agents are scheduled
for the network.

Change-Id: I206d1c7ded99e16834243649fa93569bf6a245ce
Closes-bug: #1506198

[Vadim Rovachev (vrovachev)]

Verifeid on 6.1 on Ubuntu.
Packages:
neutron packages
Version:
2014.2.2-1~u14.04+mos36

[Ekaterina Shutova (eshutova)]

Test case is automated: https://mirantis.testrail.com/index.php?/cases/view/843830