跳到主要内容

Bonding

提示
  • 无论如何配置, 单个网络链接都不会超过单个物理链路的速度
  • 802.3ad 需要交换机支持 LACP 组, 性能最好 - 基于链接信息 Hash
# /etc/network/if-post-down.d/bonding
# /etc/network/if-pre-up.d/bonding
# /etc/network/if-up.d/bonding
apk add bonding
# 手动加载驱动
modprobe bonding
echo bonding | sudo tee /etc/modules-load.d/bonding.conf

# 查看模式
cat /sys/class/net/bond0/bonding/mode
# 查看状态
# 能看到组成网卡的实际 mac 地址
cat /proc/net/bonding/bond0

ethtool eth0

# jumbo frames
ifconfig bond0 mtu 9000 up

# 添加 slave
ifenslave bond0 eth0 eth1 eth2
# 移除
ifenslave -d bond0 eth1
# 连接
ifenslave -c bond0 eth1

cat /sys/class/net/bonding_masters
# 修改配置
echo balance-rr > /sys/class/net/bond0/bonding/mode

模式​

  • 默认 balance-rr
  • balance-rr - 0 - 轮询负载
    • 唯一单 TCP/IP 流能利用多个网口的模式
    • 代价是碎片化,无序,需要 TCP/IP 拥挤协议控制
    • net.ipv4.tcp_reordering 控制拥挤程度
    • 使用对顺序无要求的协议, 例如 UDP, 基本可以做到线性性能放大
    • 需要交换机配置 etherchannel 或 trunking
  • active-backup - 1 - 主备
    • 保持相同的发送网口
  • balance-xor - 2 - XOR
    • 基于 HASH 算法进行负载均衡
    • 需要交换机配置 etherchannel 或 trunking
  • broadcast - 3 - 广播
    • 所有绑定的网卡都收到相同的数据, 用于特殊需求, 例如两个互相没连接的交换机发送相同的数据
  • 802.3ad - 4 - IEEE 802.3ad - LACP
    • 要求交换机支持 IEEE 802.3ad, 网卡带宽理论上可以翻倍
    • 实现方式也是基于 hash
      • 通常包括 src ip/mac/port/protocol, dst ip/mac/port/protocol
    • 因此较多连接的时候才会有明显的效果
  • balance-tlb - Adaptive transmit load balancing - 5 - 适配器传输负载均衡
    • 输出的数据会通过所有被绑定的网卡输出, 接收则只选择其中一个
  • balance-alb - Adaptive load balancing - 6 - 适配器输入/输出负载模式
    • balance-tlb + receive load balancing (rlb)
    • 在 5 的基础上, 接收数据也实现负载均衡
  • balance-rr, active-backup, balance-tlb 和 balance-alb 不需要交换机支持
  • balance-alb 和 balance-tlb 不一定所有交换机都能使用
    • 可能会有 arp 问题, 例如有些机器不能相互连接
    • 需要对 miimon, updelay 进行调试
  • balance-xor 可能会需要交换机配置
    • You need to set up an interface group (not LACP) on HP and Cisco switches, but apparently it's not necessary on D-Link, Netgear and Fujitsu switches.
  • 选项
    • lacp_rate - lacp pdu - 110bytes
      • slow - 30s
      • fast - 1s
    • downdelay

配置案例​

  • 需要安装 bonding 才能使用 bond-slaves 这样的指令

bonding if hook​

auto eth0
iface eth0 inet manual

auto wlan0
iface wlan0 inet manual
  # 可直接设置 master
  bond-master eth0
auto bond0
iface bond0 inet dhcp
  # 所有选项位于
  # /sys/class/net/bond0/bonding
  # 无
  bond-slaves none
  # 所有 eth
  bond-slaves all
  # 列表
  bond-slaves eth0 wlan0
  # 也可以直接使用 slaves
  slaves eth0 wlan0

  # 以下选项均可选
  # 默认为 slaves 的第一个
  bond-primary eth0
  # 热 slave
  bond-active-slave wlan0

  # 默认
  bond-mode balance-rr
  bond-miimon 0
  bond-use-carrier 1
  bond-updelay 0
  bond-downdelay 0
  bond-arp-validate none 0
  bond-fail-over_mac none 0
  bond-xmit-hash-policy layer2 0
  bond-lacp-rate slow 0
  bond-arp-ip-target ""

命令控制​

auto bond0
iface bond0 inet dhcp
  down 'servicenk set $IFACE down
  post-down rmmod bonding
  pre-up modprobe bonding mode=4 miimon=200
  up 'servicenk set $IFACE up mtu 9000
  up udevadm trigger

allow-hotplug eth0
iface eth0 inet manual
  up ifenslave bond0 $IFACE
  down ifenslave -d bond0 $IFACE 2> /dev/null

allow-hotplug eth1
iface eth1 inet manual
  up ifenslave bond0 $IFACE
  down ifenslave -d bond0 $IFACE 2> /dev/null

最简单的配置​

auto bond0
iface bond0 inet static
	address 192.168.0.2
	netmask 255.255.255.0
	gateway 192.168.0.1
	# specify the ethernet interfaces that should be bonded
	bond-slaves eth0 eth1 eth2 eth3

四网口聚合​

allow-hotplug eth0
iface eth0 inet manual

allow-hotplug eth1
iface eth1 inet manual

allow-hotplug eth2
iface eth2 inet manual

allow-hotplug eth3
iface eth3 inet manual

auto bond0
iface bond0 inet static
    address 192.168.1.5
    netmask 255.255.252.0
    gateway 192.168.1.1
    bond-slaves eth0 eth1 eth2 eth3
    bond-mode active-backup
    bond-miimon 100
    bond-downdelay 200
    bond-updelay 200

无线和有线聚合​

allow-hotplug wlan0
iface wlan0 inet manual
    wpa-ssid "*censored*"
    wpa-key-mgmt WPA-PSK
    wpa-group TKIP CCMP
    wpa-psk *censored*
    wpa-bridge bond0 # fixes mac address of outgoing packets so that they are consistent
    bond-master bond0
    bond-mode active-backup
    bond-miimon 100 # checks link status every 100 msec
    bond-give-a-chance 10 # when wlan comes up wait up to 10 seconds for it to

allow-hotplug bond0
iface bond0 inet static
    address 192.168.178.130
    netmask 255.255.255.0
    gateway 192.168.178.1
    bond-slaves eth0 # automatically brings up eth0 and slaves it to this bond
    bond-mode active-backup # uses primary if available, otherwise fallback to other
    bond-primary eth0 # priority to use eth0 when available
    bond-miimon 100

FAQ

bond stp​

  • 如果有多 lan，部分组了 bond 部分没有，那么需要开启 stp
  • 不开启 stp 会导致网络内有多条路径达到相同地方，这是有问题的
auto bond0
iface bond0 inet manual
  bond-slaves eth0 eth1
  bond-mode 802.3ad
  bond-xmit-hash-policy layer2+3

auto br0
iface br0 inet dhcp
  bridge-ports bond0
  bridge-stp 0

10gbe & 1gbe​

  • 不支持混合 lacp bonding, 可以 active-backup
  • 建议将 10gbe 设置为默认即可
# 1gbe
iface eth0 inet manual
#Onboard  #1

# 10gbe
iface eth1 inet manual

auto bond0
iface bond0 inet manual
  bond-slaves eth0 eth1
  bond-primary eth1
  bond-mode active-backup

auto br0
iface br0 inet static
  address  192.168.1.3
  netmask  255.255.255.0
  gateway  192.168.1.1
  bridge-ports bond0
  bridge-stp off

write error: Directory not empty​

bond 为 down 且无 slave 时才能修改

bond0: option mode: unable to set because the bond device has slaves

the permanent HWaddr of slave - < mac > - is still in use by bond - set the HWaddr of slave to a different address to avoid conflicts​

ifconfig | grep HWaddr

bond 802.3ad​

auto bond0
iface bond0 inet static
  bond-slaves eth0 eth1 eth2 eth3
  bond-mode 802.3ad
  bond-xmit-hash-policy layer2+3
  address 192.168.1.100
  netmask 255.255.255.0
  gateway 192.168.1.1

LAG tagged vs untagged​

  • VLAN tagged/untagged

/proc/net/bonding/bond0​

Ethernet Channel Bonding Driver: v5.15.16-0-lts

Bonding Mode: IEEE 802.3ad Dynamic link aggregation
Transmit Hash Policy: layer2 (0)
MII Status: up
MII Polling Interval (ms): 100
Up Delay (ms): 0
Down Delay (ms): 0
Peer Notification Delay (ms): 0

802.3ad info
LACP active: on
LACP rate: slow
Min links: 0
Aggregator selection policy (ad_select): stable

Slave Interface: eth0
MII Status: up
Speed: 1000 Mbps
Duplex: full
Link Failure Count: 0
Permanent HW addr: 74:00:11:22:33:40
Slave queue ID: 0
Aggregator ID: 1
Actor Churn State: none
Partner Churn State: none
Actor Churned Count: 0
Partner Churned Count: 0

Slave Interface: eth1
MII Status: up
Speed: 1000 Mbps
Duplex: full
Link Failure Count: 0
Permanent HW addr: 74:00:11:22:33:41
Slave queue ID: 0
Aggregator ID: 1
Actor Churn State: none
Partner Churn State: none
Actor Churned Count: 0
Partner Churned Count: 0

Slave Interface: eth2
MII Status: up
Speed: 1000 Mbps
Duplex: full
Link Failure Count: 0
Permanent HW addr: 74:00:11:22:33:42
Slave queue ID: 0
Aggregator ID: 1
Actor Churn State: none
Partner Churn State: none
Actor Churned Count: 0
Partner Churned Count: 0

Slave Interface: eth3
MII Status: up
Speed: 1000 Mbps
Duplex: full
Link Failure Count: 0
Permanent HW addr: 74:00:11:22:33:43
Slave queue ID: 0
Aggregator ID: 1
Actor Churn State: none
Partner Churn State: none
Actor Churned Count: 0
Partner Churned Count: 0

tree /sys/class/net/bond0​

  • 所有信息
grep . /sys/class/net/bond0/bonding/
/sys/class/net/bond0
├── addr_assign_type
├── addr_len
├── address
├── bonding
│   ├── active_slave
│   ├── ad_actor_key
│   ├── ad_actor_sys_prio
│   ├── ad_actor_system
│   ├── ad_aggregator
│   ├── ad_num_ports
│   ├── ad_partner_key
│   ├── ad_partner_mac
│   ├── ad_select
│   ├── ad_user_port_key
│   ├── all_slaves_active
│   ├── arp_all_targets
│   ├── arp_interval
│   ├── arp_ip_target
│   ├── arp_validate
│   ├── downdelay
│   ├── fail_over_mac
│   ├── lacp_rate
│   ├── lp_interval
│   ├── mii_status
│   ├── miimon
│   ├── min_links
│   ├── mode
│   ├── num_grat_arp
│   ├── num_unsol_na
│   ├── packets_per_slave
│   ├── primary
│   ├── primary_reselect
│   ├── queue_id
│   ├── resend_igmp
│   ├── slaves
│   ├── tlb_dynamic_lb
│   ├── updelay
│   ├── use_carrier
│   └── xmit_hash_policy
├── broadcast
├── carrier
├── carrier_changes
├── dev_id
├── dev_port
├── dormant
├── duplex
├── flags
├── gro_flush_timeout
├── ifalias
├── ifindex
├── iflink
├── link_mode
├── lower_eth1 -> ../../../pci0000:00/0000:00:1c.4/0000:02:00.1/net/eth1
├── lower_eth2 -> ../../../pci0000:00/0000:00:1c.4/0000:02:00.2/net/eth2
├── lower_eth3 -> ../../../pci0000:00/0000:00:1c.4/0000:02:00.3/net/eth3
├── mtu
├── name_assign_type
├── netdev_group
├── operstate
├── phys_port_id
├── phys_port_name
├── phys_switch_id
├── power
│   ├── autosuspend_delay_ms
│   ├── control
│   ├── runtime_active_time
│   ├── runtime_status
│   └── runtime_suspended_time
├── proto_down
├── queues
│   ├── rx-0
│   │   ├── rps_cpus
│   │   └── rps_flow_cnt
│   ├── tx-0
│   │   ├── byte_queue_limits
│   │   │   ├── hold_time
│   │   │   ├── inflight
│   │   │   ├── limit
│   │   │   ├── limit_max
│   │   │   └── limit_min
│   │   ├── tx_maxrate
│   │   ├── tx_timeout
│   │   └── xps_cpus
├── speed
├── statistics
│   ├── collisions
│   ├── multicast
│   ├── rx_bytes
│   ├── rx_compressed
│   ├── rx_crc_errors
│   ├── rx_dropped
│   ├── rx_errors
│   ├── rx_fifo_errors
│   ├── rx_frame_errors
│   ├── rx_length_errors
│   ├── rx_missed_errors
│   ├── rx_nohandler
│   ├── rx_over_errors
│   ├── rx_packets
│   ├── tx_aborted_errors
│   ├── tx_bytes
│   ├── tx_carrier_errors
│   ├── tx_compressed
│   ├── tx_dropped
│   ├── tx_errors
│   ├── tx_fifo_errors
│   ├── tx_heartbeat_errors
│   ├── tx_packets
│   └── tx_window_errors
├── subsystem -> ../../../../class/net
├── tx_queue_len
├── type
└── uevent