The instance's status is ACTIVE even though the TAP device is DOWN and it doesn't has an IP address

[Toni Freger]

During scale tests (80 instances) a few instances created and their status is ACTIVE (nova list/show) even though their TAP divices is DOWN and they don't have an IP address.

Version - Icehouse with RHEL7
GRE+ML2 - All-In-One+ Compute node

openstack-nova-cert-2014.1-7.el7ost.noarch
openstack-neutron-openvswitch-2014.1-35.el7ost.noarch
openstack-nova-compute-2014.1-7.el7ost.noarch
openstack-neutron-2014.1-35.el7ost.noarch
openstack-neutron-ml2-2014.1-35.el7ost.noarch

nova list
19d932b1-02fa-44bb-ade3-425d7d74baad | host1 | stress8-26 | private8=192.168.8.69 | ACTIVE

neutron port list
1d98786a-b4ce-4bb2-83d1-aa53c36fb047 | | fa:16:3e:90:c2:26 | {"subnet_id": "cff025fc-ac5c-4931-b019-927bcc9b0cb0", "ip_address": "192.168.8.69"}

ip a | grep 1d98786a-b4
569: qbr1d98786a-b4: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP
570: qvo1d98786a-b4: <BROADCAST,MULTICAST,PROMISC,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master ovs-system state UP qlen 1000
571: qvb1d98786a-b4: <BROADCAST,MULTICAST,PROMISC,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master qbr1d98786a-b4 state UP qlen 1000
575: tap1d98786a-b4: <BROADCAST,MULTICAST> mtu 1500 qdisc pfifo_fast master qbr1d98786a-b4 state DOWN qlen 500

Steps to reproduce:

1. ifconfig tap1d98786a-b4 down
2. nova show <instance id>

[haruka tanizawa (h-tanizawa)]

I have no idea about the relation between quantity of instances and tap device's status.
Moreover, if you down tap device intentionally after instance booting, it looks like something natural that instance remain active.

[Launchpad Janitor (janitor)]

[Expired for neutron because there has been no activity for 60 days.]