OpenStack网络知识片断(持续更新)

OVS创建虚拟机

创建虚拟机，例如网卡portid：1e2b09d7-e9d6-4aba-964e-79fdd8bcf3ba
ip link show dev qbr1e2b09d7-e9 #判断
brctl addbr qbr1e2b09d7-e9 #增加Linux网桥
ip link show dev qvo1e2b09d7-e9 #判断
ip link show dev qvb1e2b09d7-e9 #判断
ip link add qvb1e2b09d7-e9 type veth peer name qvo1e2b09d7-e9 #增加对等设备
ip link set qvb1e2b09d7-e9 up #激活设备
ip link set qvb1e2b09d7-e9 promisc on #混杂模式
ip link set qvo1e2b09d7-e9 up
ip link set qvo1e2b09d7-e9 promisc on
ip link set qbr1e2b09d7-e9 up #激活网桥
brctl addif qbr1e2b09d7-e9 qvb1e2b09d7-e9 #向Linux网桥添加设备
#下面的命令向OVS添加port
ovs-vsctl -- --may-exist add-port br-int qvo1e2b09d7-e9 -- set Interface qvo1e2b09d7-e9 external-ids:iface-id=1e2b09d7-e9d6-4aba-964e-79fdd8bcf3ba external-ids:iface-status=active external-ids:attached-mac=fa:16:3e:ea:ad:8d external-ids:vm-uuid=49b6d841-163f-4aab-b309-149727c227b4

OVSagent

初始化：
ovs-vsctl -- --if-exists del-port br-int patch-tun
ovs-ofctl del-flows br-int
ovs-ofctl add-flow br-int hard_timeout=0,idle_timeout=0,priority=1,actions=normal
循环主体：
1) ovs-vsctl list-ports br-int
输出：qvo1e2b09d7-e9\nqvo2d58d5dc-db\nqvo2e505b97-bb\nqvo5739b2dc-78\nqvo69121bea-6a\nqvod58fde4e-5f\nqvoe0a0b269-53\n
2) 循环调用:ovs-vsctl get Interface qvo1e2b09d7-e9 external_ids，获取iface-id（portid）
输出：
{attached-mac="fa:16:3e:ea:ad:8d", iface-id="1e2b09d7-e9d6-4aba-964e-79fdd8bcf3ba", iface-status=active, vm-uuid="49b6d841-163f-4aab-b309-149727c227b4"}\n
3) 根据portid循环：

a) 向Quantum查询信息，调用get_device_details接口
b) ovs-vsctl -- --columns=external_ids,name,ofport find Interface external_ids:iface-id="1e2b09d7-e9d6-4aba-964e-79fdd8bcf3ba"，输出：
external_ids : {attached-mac="fa:16:3e:ea:ad:8d", iface-id="1e2b09d7-e9d6-4aba-964e-79fdd8bcf3ba", iface-status=active, vm-uuid="49b6d841-163f-4aab-b309-149727c227b4"}
name : "qvo1e2b09d7-e9"
ofport : 6
c) 给port所属的network自动分配(如果已记录过该net，直接跳到下一步)local vlan id（1-4094），且对于network对应的physical net，节点上要有一个OVS网桥与之对应，如果是vlan模式，需要做如下操作：
# outbound，出口的vlan转换
br.add_flow(priority=4,
in_port=self.phys_ofports[physical_network], #与br-int连接的port的标号
dl_vlan=lvid, #自动分配的local vlan，从1开始
actions="mod_vlan_vid:%s,normal" % segmentation_id) #segmentation_id是plugin分配的vlan号
# inbound，入口的vlan转换
self.int_br.add_flow(priority=3,
in_port=self.int_ofports[physical_network], #与上面的br连接的port的标号
dl_vlan=segmentation_id,
actions="mod_vlan_vid:%s,normal" % lvid)
d) ovs-vsctl set Port qvo1e2b09d7-e9 tag=1 #这里的1是为network分配的local vlan id
e) ovs-ofctl del-flows br-int in_port=6 #这里的6指port的标号，表示不允许数据流入

LinuxBridge创建虚拟机

生成设备名：tap+portid
如果不是Xen，执行ip link show dev “tap+portid”看设备是否已存在，若存在返回；若不存在：
ip tuntap add “tap+portid” mode tap
ip link set “tap+portid” address <mac_address>
ip link set “tap+portid” up

LinuxBridge agent

使用pydev库获取本机所有以tap开头的设备
对每一个tap设备（port）循环：
向Quantum获取port的详细信息
ip lingk show dev <dev>
获取设备所在的网桥（对于新增设备应该是空）：在/sys/devices/virtual/net/目录下找到以brq开头的网桥名，对每一个网桥：
获取/sys/devices/virtual/net/<bridge>/brif/目录下所有设备
生成network对应的网桥(brq+networkid)
获取network对应phynet所对应的phyinterface(必须已存在)
ip link add link <phyinterface> name <phyinterface.vlanid> type vlan id <vlanid>
ip link set <phyinterface.vlanid> up
brctl addbr <brq+networkid>
brctl setfd <brq+networkid>
brctl stp <brq+networkid> off
ip link set <brq+networkid> up
brctl addif <brq+networkid> <phyinterface.vlanid>
brctl addif <brq+networkid> <dev>

dhcp agent

dhcp agent需要为不同的plugin配置不同的interface_driver
OVS：quantum.agent.linux.interface.OVSInterfaceDriver
LinuxBridge：quantum.agent.linux.interface.BridgeInterfaceDriver

dhcp agent执行的命令及输出:
#查看设备是否存在
ip netns exec qdhcp-c6e38a5a-2adf-42a5-8c6f-5eab99208869 ip -o link show tap9739ea30-d6
'17: tap9739ea30-d6: <BROADCAST,MULTICAST,PROMISC,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN \\ link/ether fa:16:3e:de:19:12 brd ff:ff:ff:ff:ff:ff\n'
#查看设备IP
ip netns exec qdhcp-c6e38a5a-2adf-42a5-8c6f-5eab99208869 ip addr show tap9739ea30-d6 permanent scope global
'17: tap9739ea30-d6: <BROADCAST,MULTICAST,PROMISC,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN \n link/ether fa:16:3e:de:19:12 brd ff:ff:ff:ff:ff:ff\n inet 10.10.11.2/24 brd 10.10.11.255 scope global tap9739ea30-d6\n'
#查看DHCP进程信息，其中的进程号是从文件中获取/var/lib/quantum/dhcp/{netid}/pid
cat /proc/13695/cmdline
'dnsmasq\x00--no-hosts\x00--no-resolv\x00--strict-order\x00--bind-interfaces\x00--interface=tap9739ea30-d6\x00--except-interface=lo\x00--domain=openstacklocal\x00--pid-file=/var/lib/quantum/dhcp/c6e38a5a-2adf-42a5-8c6f-5eab99208869/pid\x00--dhcp-hostsfile=/var/lib/quantum/dhcp/c6e38a5a-2adf-42a5-8c6f-5eab99208869/host\x00--dhcp-optsfile=/var/lib/quantum/dhcp/c6e38a5a-2adf-42a5-8c6f-5eab99208869/opts\x00--dhcp-script=/usr/bin/quantum-dhcp-agent-dnsmasq-lease-update\x00--leasefile-ro\x00--dhcp-range=set:tag0,10.10.11.0,static,120s\x00'
#停止进程
ip netns exec qdhcp-c6e38a5a-2adf-42a5-8c6f-5eab99208869 kill -9 13695
#启动进程
QUANTUM_RELAY_SOCKET_PATH=/var/lib/quantum/dhcp/lease_relay QUANTUM_NETWORK_ID=c6e38a5a-2adf-42a5-8c6f-5eab99208869 ip netns exec qdhcp-c6e38a5a-2adf-42a5-8c6f-5eab99208869 dnsmasq --no-hosts --no-resolv --strict-order --bind-interfaces --interface=tap9739ea30-d6 --except-interface=lo --domain=openstacklocal --pid-file=/var/lib/quantum/dhcp/c6e38a5a-2adf-42a5-8c6f-5eab99208869/pid --dhcp-hostsfile=/var/lib/quantum/dhcp/c6e38a5a-2adf-42a5-8c6f-5eab99208869/host --dhcp-optsfile=/var/lib/quantum/dhcp/c6e38a5a-2adf-42a5-8c6f-5eab99208869/opts --dhcp-script=/usr/bin/quantum-dhcp-agent-dnsmasq-lease-update --leasefile-ro --dhcp-range=set:tag0,10.10.11.0,static,120s

dhcp agent需要为不同的plugin配置不同的interface_driver

OVS：quantum.agent.linux.interface.OVSInterfaceDriver
LinuxBridge：quantum.agent.linux.interface.BridgeInterfaceDriver

dhcp agent执行的命令及输出:

ip netns exec qdhcp-c6e38a5a-2adf-42a5-8c6f-5eab99208869 ip -o link show tap9739ea30-d6

'17: tap9739ea30-d6: <BROADCAST,MULTICAST,PROMISC,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN \\ link/ether fa:16:3e:de:19:12 brd ff:ff:ff:ff:ff:ff\n'

ip netns exec qdhcp-c6e38a5a-2adf-42a5-8c6f-5eab99208869 ip addr show tap9739ea30-d6 permanent scope global

'17: tap9739ea30-d6: <BROADCAST,MULTICAST,PROMISC,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN \n link/ether fa:16:3e:de:19:12 brd ff:ff:ff:ff:ff:ff\n inet 10.10.11.2/24 brd 10.10.11.255 scope global tap9739ea30-d6\n'

cat /proc/13695/cmdline

'dnsmasq\x00--no-hosts\x00--no-resolv\x00--strict-order\x00--bind-interfaces\x00--interface=tap9739ea30-d6\x00--except-interface=lo\x00--domain=openstacklocal\x00--pid-file=/var/lib/quantum/dhcp/c6e38a5a-2adf-42a5-8c6f-5eab99208869/pid\x00--dhcp-hostsfile=/var/lib/quantum/dhcp/c6e38a5a-2adf-42a5-8c6f-5eab99208869/host\x00--dhcp-optsfile=/var/lib/quantum/dhcp/c6e38a5a-2adf-42a5-8c6f-5eab99208869/opts\x00--dhcp-script=/usr/bin/quantum-dhcp-agent-dnsmasq-lease-update\x00--leasefile-ro\x00--dhcp-range=set:tag0,10.10.11.0,static,120s\x00'

ip netns exec qdhcp-c6e38a5a-2adf-42a5-8c6f-5eab99208869 kill -9 13695

QUANTUM_RELAY_SOCKET_PATH=/var/lib/quantum/dhcp/lease_relay QUANTUM_NETWORK_ID=c6e38a5a-2adf-42a5-8c6f-5eab99208869 ip netns exec qdhcp-c6e38a5a-2adf-42a5-8c6f-5eab99208869 dnsmasq --no-hosts --no-resolv --strict-order --bind-interfaces --interface=tap9739ea30-d6 --except-interface=lo --domain=openstacklocal --pid-file=/var/lib/quantum/dhcp/c6e38a5a-2adf-42a5-8c6f-5eab99208869/pid --dhcp-hostsfile=/var/lib/quantum/dhcp/c6e38a5a-2adf-42a5-8c6f-5eab99208869/host --dhcp-optsfile=/var/lib/quantum/dhcp/c6e38a5a-2adf-42a5-8c6f-5eab99208869/opts --dhcp-script=/usr/bin/quantum-dhcp-agent-dnsmasq-lease-update --leasefile-ro --dhcp-range=set:tag0,10.10.11.0,static,120s

l3 agent

l3 agent需要配置与plugin对应的interface_driver:
OVS: quantum.agent.linux.interface.OVSInterfaceDriver，此时external_network_bridge为br-ex
LinuxBridge: quantum.agent.linux.interface.BridgeInterfaceDriver

初始化：

1. 加载interface_driver
2. ip netns list #列出以'qrouter-'开头的namespace
输出：
qdhcp-487f81ab-98d3-457a-b712-b29e71e89b52
qdhcp-084ae80a-b108-4f8a-90ca-f44aa1ca738a
qdhcp-7c25296d-bc81-45f6-bcc0-37fa44588b83
qdhcp-c6e38a5a-2adf-42a5-8c6f-5eab99208869
qrouter-0e38e30f-4fae-4f48-be4c-76d2fb803b23
qrouter-ccf5f323-2a41-41d1-8bb6-b772a8ae17fc
循环： ip netns exec qrouter-ccf5f323-2a41-41d1-8bb6-b772a8ae17fc ip -o link list
'9: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN \\ link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00\n25:qr-012c9d13-85: <BROADCAST,MULTICAST,PROMISC,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN \\ link/ether fa:16:3e:f2:a8:56 brd ff:ff:ff:ff:ff:ff\n26:qg-388798a1-55: <BROADCAST,MULTICAST,PROMISC,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN \\ link/ether fa:16:3e:f6:23:b9 brd ff:ff:ff:ff:ff:ff\n'
对命名空间内的设备名作循环（加粗字体）：
如果是qr开头：ovs-vsctl --timeout=2 -- --if-exists del-port br-int qr-012c9d13-85
如果是qg开头：ovs-vsctl --timeout=2 -- --if-exists del-port br-ex qg-388798a1-55
循环结束
循环结束

工作任务：

1. 保证br-ex存在
2. 向Quantum获取router:external的网络（一个l3 agent只处理一个external网络，默认是br-ex，可以在配置文件中配置external_network_bridge）
3.
循环获取Quantum中的router对象（只处理连接外网的router）：
增加qrouter-routerid命名空间
ip netns exec qrouter-0e38e30f-4fae-4f48-be4c-76d2fb803b23 sysctl -w net.ipv4.ip_forward=1
如果配置了metadata_ip，执行
rules.append(('INPUT', '-s 0.0.0.0/0 -d %s -p tcp -m tcp --dport %s -j ACCEPT' % (self.conf.metadata_ip, self.conf.metadata_port)))
rules.append(('PREROUTING', '-s 0.0.0.0/0 -d 169.254.169.254/32 -p tcp -m tcp --dport 80 -j DNAT --to-destination %s:%s' % (self.conf.metadata_ip, self.conf.metadata_port)))
应用iptable规则
ip netns exec qrouter-0e38e30f-4fae-4f48-be4c-76d2fb803b23 /sbin/iptables-save -t filter
ip netns exec qrouter-0e38e30f-4fae-4f48-be4c-76d2fb803b23 /sbin/iptables-restore
ip netns exec qrouter-0e38e30f-4fae-4f48-be4c-76d2fb803b23 /sbin/iptables-save -t nat
ip netns exec qrouter-0e38e30f-4fae-4f48-be4c-76d2fb803b23 /sbin/iptables-restore
对于router的新增内部port循环：
ip netns exec qrouter-ccf5f323-2a41-41d1-8bb6-b772a8ae17fc ip -o link showqr-012c9d13-85，如果设备不存在：
ovs-vsctl -- --may-exist add-port br-intqr-012c9d13-85
-- set Interface qr-012c9d13-85 type=internal
-- set Interface qr-012c9d13-85 external-ids:iface-id=012c9d13-8554-4b39-96b8-e4bd2e787559
-- set Interface qr-012c9d13-85 external-ids:iface-status=active
-- set Interface qr-012c9d13-85 external-ids:attached-mac=fa:16:3e:f2:a8:56
ip link set qr-012c9d13-85 address fa:16:3e:f2:a8:56
ip link set qr-012c9d13-85 netns qrouter-ccf5f323-2a41-41d1-8bb6-b772a8ae17fc
ip netns exec qrouter-ccf5f323-2a41-41d1-8bb6-b772a8ae17fc ip link set qr-012c9d13-85 up
ip netns exec qrouter-ccf5f323-2a41-41d1-8bb6-b772a8ae17fc ip addr show qr-012c9d13-85 permanent scope global
ip netns exec qrouter-ccf5f323-2a41-41d1-8bb6-b772a8ae17fc ip -4 addr add 10.10.10.1/24 brd 10.10.10.255 scope global devqr-012c9d13-85
如果router连接到外部网络，增加snat规则(将内部的IP转换为外网的IP)，应用iptable规则
循环结束

对于router上删除的port循环：
在OVS上删除port
删除nat规则
循环结束

初始化gw_port：ip netns exec qrouter-ccf5f323-2a41-41d1-8bb6-b772a8ae17fc ip -o link showqg-388798a1-55，如果设备不存在，同上（操作br-ex）
ip netns exec qrouter-ccf5f323-2a41-41d1-8bb6-b772a8ae17fc route add default <gw_port的网关IP>
增加snat规则

处理router上的floatingIP，对br-ex上的port(qg-388798a1-55)配置外网地址，配置snat/dnat规则
循环结束