Различия

Показаны различия между двумя версиями страницы.

Ссылка на это сравнение

Следующая версия 		Предыдущая версия
wiki:iproute2:ip_link [19/06/2018 00:06]
Boris V. Vuks создано 		wiki:iproute2:ip_link [19/06/2018 01:06] (текущий)
Boris V. Vuks [Создание]
Строка 44: Строка 44:
  
 Другие расширения типа: vepa, veb и hairpin режимов могут быть найдены на соответвствующей странице команды bridge. Другие расширения типа: vepa, veb и hairpin режимов могут быть найдены на соответвствующей странице команды bridge.
 +
 +====== Vlan 802.1q манипуляции ======
 +
 +==== Создание ==== 
 +  ip link add link eth0 name eth0.2 type vlan id 2
 +  ip link set eth0.2 up
 +  
 +==== Удаление ==== 
 +  ip link set eth0.2 down
 +  ip link del eth0.2
 +  
 +====== bonding манипуляции ======  
 +[[https://www.kernel.org/doc/Documentation/networking/bonding.txt| Подробности читаем тут]].
 +  modprobe bonding
 +
 +==== Создание ====   
 +  ip link add bond0 type bond
 +  echo 4 > /sys/class/net/bond0/bonding/mode
 +  ip link set eth0 master bond0
 +  ip link set eth1 master bond0
 +  ip link set bond0 up
 +
 +
 +<spoiler | режимы в /sys/class/net/bond0/bonding/mode >
 +
 + Specifies one of the bonding policies. The default is
 + balance-rr (round robin).  Possible values are:
 +<code>
 + balance-rr or 0
 + Round-robin policy: Transmit packets in sequential
 + order from the first available slave through the
 + last.  This mode provides load balancing and fault
 + tolerance.
 +</code>
 +<code>
 + active-backup or 1
 + Active-backup policy: Only one slave in the bond is
 + active.  A different slave becomes active if, and only
 + if, the active slave fails.  The bond's MAC address is
 + externally visible on only one port (network adapter)
 + to avoid confusing the switch.
 +
 + In bonding version 2.6.2 or later, when a failover
 + occurs in active-backup mode, bonding will issue one
 + or more gratuitous ARPs on the newly active slave.
 + One gratuitous ARP is issued for the bonding master
 + interface and each VLAN interfaces configured above
 + it, provided that the interface has at least one IP
 + address configured.  Gratuitous ARPs issued for VLAN
 + interfaces are tagged with the appropriate VLAN id.
 +
 + This mode provides fault tolerance.  The primary
 + option, documented below, affects the behavior of this
 + mode.
 +</code>
 +<code>
 + balance-xor or 2
 + XOR policy: Transmit based on the selected transmit
 + hash policy.  The default policy is a simple [(source
 + MAC address XOR'd with destination MAC address XOR
 + packet type ID) modulo slave count].  Alternate transmit
 + policies may be selected via the xmit_hash_policy option,
 + described below.
 + This mode provides load balancing and fault tolerance.
 +</code>
 +<code>
 + broadcast or 3
 + Broadcast policy: transmits everything on all slave
 + interfaces.  This mode provides fault tolerance.
 +</code>
 +<code>
 + 802.3ad or 4
 + IEEE 802.3ad Dynamic link aggregation.  Creates
 + aggregation groups that share the same speed and
 + duplex settings.  Utilizes all slaves in the active
 + aggregator according to the 802.3ad specification.
 +
 + Slave selection for outgoing traffic is done according
 + to the transmit hash policy, which may be changed from
 + the default simple XOR policy via the xmit_hash_policy
 + option, documented below.  Note that not all transmit
 + policies may be 802.3ad compliant, particularly in
 + regards to the packet mis-ordering requirements of
 + section 43.2.4 of the 802.3ad standard.  Differing
 + peer implementations will have varying tolerances for
 + noncompliance.
 + Prerequisites:
 + 1. Ethtool support in the base drivers for retrieving
 + the speed and duplex of each slave.
 + 2. A switch that supports IEEE 802.3ad Dynamic link
 + aggregation.
 + Most switches will require some type of configuration
 + to enable 802.3ad mode.
 +</code>
 +<code>
 + balance-tlb or 5
 + Adaptive transmit load balancing: channel bonding that
 + does not require any special switch support.
 +
 + In tlb_dynamic_lb=1 mode; the outgoing traffic is
 + distributed according to the current load (computed
 + relative to the speed) on each slave.
 +
 + In tlb_dynamic_lb=0 mode; the load balancing based on
 + current load is disabled and the load is distributed
 + only using the hash distribution.
 +
 + Incoming traffic is received by the current slave.
 + If the receiving slave fails, another slave takes over
 + the MAC address of the failed receiving slave.
 +
 + Prerequisite:
 +
 + Ethtool support in the base drivers for retrieving the
 + speed of each slave.
 +</code>
 +<code>
 + balance-alb or 6
 + Adaptive load balancing: includes balance-tlb plus
 + receive load balancing (rlb) for IPV4 traffic, and
 + does not require any special switch support.  The
 + receive load balancing is achieved by ARP negotiation.
 + The bonding driver intercepts the ARP Replies sent by
 + the local system on their way out and overwrites the
 + source hardware address with the unique hardware
 + address of one of the slaves in the bond such that
 + different peers use different hardware addresses for
 + the server.
 +
 + Receive traffic from connections created by the server
 + is also balanced.  When the local system sends an ARP
 + Request the bonding driver copies and saves the peer's
 + IP information from the ARP packet.  When the ARP
 + Reply arrives from the peer, its hardware address is
 + retrieved and the bonding driver initiates an ARP
 + reply to this peer assigning it to one of the slaves
 + in the bond.  A problematic outcome of using ARP
 + negotiation for balancing is that each time that an
 + ARP request is broadcast it uses the hardware address
 + of the bond.  Hence, peers learn the hardware address
 + of the bond and the balancing of receive traffic
 + collapses to the current slave.  This is handled by
 + sending updates (ARP Replies) to all the peers with
 + their individually assigned hardware address such that
 + the traffic is redistributed.  Receive traffic is also
 + redistributed when a new slave is added to the bond
 + and when an inactive slave is re-activated.  The
 + receive load is distributed sequentially (round robin)
 + among the group of highest speed slaves in the bond.
 +
 + When a link is reconnected or a new slave joins the
 + bond the receive traffic is redistributed among all
 + active slaves in the bond by initiating ARP Replies
 + with the selected MAC address to each of the
 + clients. The updelay parameter (detailed below) must
 + be set to a value equal or greater than the switch's
 + forwarding delay so that the ARP Replies sent to the
 + peers will not be blocked by the switch.
 +
 + Prerequisites:
 + 1. Ethtool support in the base drivers for retrieving
 + the speed of each slave.
 + 2. Base driver support for setting the hardware
 + address of a device while it is open.  This is
 + required so that there will always be one slave in the
 + team using the bond hardware address (the
 + curr_active_slave) while having a unique hardware
 + address for each slave in the bond.  If the
 + curr_active_slave fails its hardware address is
 + swapped with the new curr_active_slave that was
 + chosen.
 +</code>
 +</spoiler>
 +
 +==== Удаление ==== 
 +  ip link set bond0 down
 +  ip link del bond0
  
  

Вы посетили:

QR Code
QR Code wiki:iproute2:ip_link (generated for current page)