neutron

  1. Bug #1833653
  2. Activity log

Activity log for bug #1833653

Date Who What changed Old value New value Message
2019-06-21 03:57:57 Yang Li bug added bug
2019-06-21 07:28:41 Slawek Kaplonski tags l3-dvr-backlog
2019-06-21 10:11:47 YAMAMOTO Takashi tags l3-dvr-backlog l3-dvr-backlog l3-ha
2019-06-21 12:07:25 YAMAMOTO Takashi neutron: importance Undecided Medium
2019-06-21 12:07:25 YAMAMOTO Takashi neutron: status New Confirmed
2019-06-21 14:00:15 Brian Haley bug added subscriber Brian Haley
2019-06-24 11:59:15 OpenStack Infra neutron: status Confirmed In Progress
2019-06-24 11:59:15 OpenStack Infra neutron: assignee Yang Li (yang-li)
2019-07-02 05:25:48 Yang Li description If a router's keepalived is dead(kill -9 <pid>, or kill -HUP <pid> too many times), and the original role in this node is master, this will cause brain split. Then when we restart neutron-l3-agent, the ipv6 will be cleanup, but the ipv4 still exists, I think we should alse cleanup ipv4 before enable keepalived. The current state: original master: [root@node-1 ~]# ip netns exec qrouter-a8f9d4e4-622d-4ebf-b7a9-4818f63ef502 ip a 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN qlen 1 link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 inet 127.0.0.1/8 scope host lo valid_lft forever preferred_lft forever inet6 ::1/128 scope host valid_lft forever preferred_lft forever 28: ha-66734f93-5e: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UNKNOWN qlen 1000 link/ether fa:16:3e:fd:49:a1 brd ff:ff:ff:ff:ff:ff inet 169.254.192.1/18 brd 169.254.255.255 scope global ha-66734f93-5e valid_lft forever preferred_lft forever inet 169.254.0.1/24 scope global ha-66734f93-5e valid_lft forever preferred_lft forever inet6 fe80::f816:3eff:fefd:49a1/64 scope link valid_lft forever preferred_lft forever 29: qr-4f77a86c-c2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UNKNOWN qlen 1000 link/ether fa:16:3e:26:98:8f brd ff:ff:ff:ff:ff:ff inet 192.168.100.1/24 scope global qr-4f77a86c-c2 valid_lft forever preferred_lft forever 46: qg-e28d50e4-ba: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN qlen 1000 link/ether fa:16:3e:c4:3e:43 brd ff:ff:ff:ff:ff:ff inet 172.16.10.133/24 scope global qg-e28d50e4-ba valid_lft forever preferred_lft forever inet 172.16.10.134/32 scope global qg-e28d50e4-ba valid_lft forever preferred_lft forever currnet master: [root@node-2 ~]# ip netns exec qrouter-a8f9d4e4-622d-4ebf-b7a9-4818f63ef502 ip a 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN qlen 1 link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 inet 127.0.0.1/8 scope host lo valid_lft forever preferred_lft forever inet6 ::1/128 scope host valid_lft forever preferred_lft forever 32: ha-606f2f23-5f: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UNKNOWN qlen 1000 link/ether fa:16:3e:8e:dc:c4 brd ff:ff:ff:ff:ff:ff inet 169.254.192.2/18 brd 169.254.255.255 scope global ha-606f2f23-5f valid_lft forever preferred_lft forever inet 169.254.0.1/24 scope global ha-606f2f23-5f valid_lft forever preferred_lft forever inet6 fe80::f816:3eff:fe8e:dcc4/64 scope link valid_lft forever preferred_lft forever 33: qr-4f77a86c-c2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UNKNOWN qlen 1000 link/ether fa:16:3e:26:98:8f brd ff:ff:ff:ff:ff:ff inet 192.168.100.1/24 scope global qr-4f77a86c-c2 valid_lft forever preferred_lft forever inet6 fe80::f816:3eff:fe26:988f/64 scope link nodad valid_lft forever preferred_lft forever 50: qg-e28d50e4-ba: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN qlen 1000 link/ether fa:16:3e:c4:3e:43 brd ff:ff:ff:ff:ff:ff inet 172.16.10.133/24 scope global qg-e28d50e4-ba valid_lft forever preferred_lft forever inet 172.16.10.134/32 scope global qg-e28d50e4-ba valid_lft forever preferred_lft forever inet6 fe80::f816:3eff:fec4:3e43/64 scope link nodad valid_lft forever preferred_lft forever The command line output showed that the original master's ipv6 were cleanup. If a router's keepalived is dead(kill -9 <pid>, or kill -HUP <pid> too many times that cause process terminated), and the original role in this node is master, this will cause brain split. Then when we restart neutron-l3-agent, the ipv6 will be cleanup, but the ipv4 still exists, I think we should alse cleanup ipv4 before enable keepalived. The current state: original master: [root@node-1 ~]# ip netns exec qrouter-a8f9d4e4-622d-4ebf-b7a9-4818f63ef502 ip a 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN qlen 1     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00     inet 127.0.0.1/8 scope host lo        valid_lft forever preferred_lft forever     inet6 ::1/128 scope host        valid_lft forever preferred_lft forever 28: ha-66734f93-5e: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UNKNOWN qlen 1000     link/ether fa:16:3e:fd:49:a1 brd ff:ff:ff:ff:ff:ff     inet 169.254.192.1/18 brd 169.254.255.255 scope global ha-66734f93-5e        valid_lft forever preferred_lft forever     inet 169.254.0.1/24 scope global ha-66734f93-5e        valid_lft forever preferred_lft forever     inet6 fe80::f816:3eff:fefd:49a1/64 scope link        valid_lft forever preferred_lft forever 29: qr-4f77a86c-c2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UNKNOWN qlen 1000     link/ether fa:16:3e:26:98:8f brd ff:ff:ff:ff:ff:ff     inet 192.168.100.1/24 scope global qr-4f77a86c-c2        valid_lft forever preferred_lft forever 46: qg-e28d50e4-ba: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN qlen 1000     link/ether fa:16:3e:c4:3e:43 brd ff:ff:ff:ff:ff:ff     inet 172.16.10.133/24 scope global qg-e28d50e4-ba        valid_lft forever preferred_lft forever     inet 172.16.10.134/32 scope global qg-e28d50e4-ba        valid_lft forever preferred_lft forever currnet master: [root@node-2 ~]# ip netns exec qrouter-a8f9d4e4-622d-4ebf-b7a9-4818f63ef502 ip a 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN qlen 1     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00     inet 127.0.0.1/8 scope host lo        valid_lft forever preferred_lft forever     inet6 ::1/128 scope host        valid_lft forever preferred_lft forever 32: ha-606f2f23-5f: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UNKNOWN qlen 1000     link/ether fa:16:3e:8e:dc:c4 brd ff:ff:ff:ff:ff:ff     inet 169.254.192.2/18 brd 169.254.255.255 scope global ha-606f2f23-5f        valid_lft forever preferred_lft forever     inet 169.254.0.1/24 scope global ha-606f2f23-5f        valid_lft forever preferred_lft forever     inet6 fe80::f816:3eff:fe8e:dcc4/64 scope link        valid_lft forever preferred_lft forever 33: qr-4f77a86c-c2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UNKNOWN qlen 1000     link/ether fa:16:3e:26:98:8f brd ff:ff:ff:ff:ff:ff     inet 192.168.100.1/24 scope global qr-4f77a86c-c2        valid_lft forever preferred_lft forever     inet6 fe80::f816:3eff:fe26:988f/64 scope link nodad        valid_lft forever preferred_lft forever 50: qg-e28d50e4-ba: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN qlen 1000     link/ether fa:16:3e:c4:3e:43 brd ff:ff:ff:ff:ff:ff     inet 172.16.10.133/24 scope global qg-e28d50e4-ba        valid_lft forever preferred_lft forever     inet 172.16.10.134/32 scope global qg-e28d50e4-ba        valid_lft forever preferred_lft forever     inet6 fe80::f816:3eff:fec4:3e43/64 scope link nodad        valid_lft forever preferred_lft forever The command line output showed that the original master's ipv6 were cleanup.
2019-07-08 17:38:53 Swaminathan Vasudevan tags l3-dvr-backlog l3-ha l3-ha
2019-08-06 05:52:09 OpenStack Infra neutron: assignee Yang Li (yang-li) zhengyong (zhengy23)
2023-01-04 16:23:30 Rodolfo Alonso neutron: status In Progress Won't Fix