2 Linux Ethernet Bonding Driver HOWTO
4 Latest update: 12 November 2007
6 Initial release : Thomas Davis <tadavis at lbl.gov>
7 Corrections, HA extensions : 2000/10/03-15 :
8 - Willy Tarreau <willy at meta-x.org>
9 - Constantine Gavrilov <const-g at xpert.com>
10 - Chad N. Tindel <ctindel at ieee dot org>
11 - Janice Girouard <girouard at us dot ibm dot com>
12 - Jay Vosburgh <fubar at us dot ibm dot com>
14 Reorganized and updated Feb 2005 by Jay Vosburgh
15 Added Sysfs information: 2006/04/24
16 - Mitch Williams <mitch.a.williams at intel.com>
21 The Linux bonding driver provides a method for aggregating
22 multiple network interfaces into a single logical "bonded" interface.
23 The behavior of the bonded interfaces depends upon the mode; generally
24 speaking, modes provide either hot standby or load balancing services.
25 Additionally, link integrity monitoring may be performed.
27 The bonding driver originally came from Donald Becker's
28 beowulf patches for kernel 2.0. It has changed quite a bit since, and
29 the original tools from extreme-linux and beowulf sites will not work
30 with this version of the driver.
32 For new versions of the driver, updated userspace tools, and
33 who to ask for help, please follow the links at the end of this file.
38 1. Bonding Driver Installation
40 2. Bonding Driver Options
42 3. Configuring Bonding Devices
43 3.1 Configuration with Sysconfig Support
44 3.1.1 Using DHCP with Sysconfig
45 3.1.2 Configuring Multiple Bonds with Sysconfig
46 3.2 Configuration with Initscripts Support
47 3.2.1 Using DHCP with Initscripts
48 3.2.2 Configuring Multiple Bonds with Initscripts
49 3.3 Configuring Bonding Manually with Ifenslave
50 3.3.1 Configuring Multiple Bonds Manually
51 3.4 Configuring Bonding Manually via Sysfs
53 4. Querying Bonding Configuration
54 4.1 Bonding Configuration
55 4.2 Network Configuration
57 5. Switch Configuration
59 6. 802.1q VLAN Support
62 7.1 ARP Monitor Operation
63 7.2 Configuring Multiple ARP Targets
64 7.3 MII Monitor Operation
66 8. Potential Trouble Sources
67 8.1 Adventures in Routing
68 8.2 Ethernet Device Renaming
69 8.3 Painfully Slow Or No Failed Link Detection By Miimon
75 11. Configuring Bonding for High Availability
76 11.1 High Availability in a Single Switch Topology
77 11.2 High Availability in a Multiple Switch Topology
78 11.2.1 HA Bonding Mode Selection for Multiple Switch Topology
79 11.2.2 HA Link Monitoring for Multiple Switch Topology
81 12. Configuring Bonding for Maximum Throughput
82 12.1 Maximum Throughput in a Single Switch Topology
83 12.1.1 MT Bonding Mode Selection for Single Switch Topology
84 12.1.2 MT Link Monitoring for Single Switch Topology
85 12.2 Maximum Throughput in a Multiple Switch Topology
86 12.2.1 MT Bonding Mode Selection for Multiple Switch Topology
87 12.2.2 MT Link Monitoring for Multiple Switch Topology
89 13. Switch Behavior Issues
90 13.1 Link Establishment and Failover Delays
91 13.2 Duplicated Incoming Packets
93 14. Hardware Specific Considerations
96 15. Frequently Asked Questions
98 16. Resources and Links
101 1. Bonding Driver Installation
102 ==============================
104 Most popular distro kernels ship with the bonding driver
105 already available as a module and the ifenslave user level control
106 program installed and ready for use. If your distro does not, or you
107 have need to compile bonding from source (e.g., configuring and
108 installing a mainline kernel from kernel.org), you'll need to perform
111 1.1 Configure and build the kernel with bonding
112 -----------------------------------------------
114 The current version of the bonding driver is available in the
115 drivers/net/bonding subdirectory of the most recent kernel source
116 (which is available on http://kernel.org). Most users "rolling their
117 own" will want to use the most recent kernel from kernel.org.
119 Configure kernel with "make menuconfig" (or "make xconfig" or
120 "make config"), then select "Bonding driver support" in the "Network
121 device support" section. It is recommended that you configure the
122 driver as module since it is currently the only way to pass parameters
123 to the driver or configure more than one bonding device.
125 Build and install the new kernel and modules, then continue
126 below to install ifenslave.
128 1.2 Install ifenslave Control Utility
129 -------------------------------------
131 The ifenslave user level control program is included in the
132 kernel source tree, in the file Documentation/networking/ifenslave.c.
133 It is generally recommended that you use the ifenslave that
134 corresponds to the kernel that you are using (either from the same
135 source tree or supplied with the distro), however, ifenslave
136 executables from older kernels should function (but features newer
137 than the ifenslave release are not supported). Running an ifenslave
138 that is newer than the kernel is not supported, and may or may not
141 To install ifenslave, do the following:
143 # gcc -Wall -O -I/usr/src/linux/include ifenslave.c -o ifenslave
144 # cp ifenslave /sbin/ifenslave
146 If your kernel source is not in "/usr/src/linux," then replace
147 "/usr/src/linux/include" in the above with the location of your kernel
148 source include directory.
150 You may wish to back up any existing /sbin/ifenslave, or, for
151 testing or informal use, tag the ifenslave to the kernel version
152 (e.g., name the ifenslave executable /sbin/ifenslave-2.6.10).
156 If you omit the "-I" or specify an incorrect directory, you
157 may end up with an ifenslave that is incompatible with the kernel
158 you're trying to build it for. Some distros (e.g., Red Hat from 7.1
159 onwards) do not have /usr/include/linux symbolically linked to the
160 default kernel source include directory.
162 SECOND IMPORTANT NOTE:
163 If you plan to configure bonding using sysfs, you do not need
166 2. Bonding Driver Options
167 =========================
169 Options for the bonding driver are supplied as parameters to the
170 bonding module at load time, or are specified via sysfs.
172 Module options may be given as command line arguments to the
173 insmod or modprobe command, but are usually specified in either the
174 /etc/modules.conf or /etc/modprobe.conf configuration file, or in a
175 distro-specific configuration file (some of which are detailed in the next
178 Details on bonding support for sysfs is provided in the
179 "Configuring Bonding Manually via Sysfs" section, below.
181 The available bonding driver parameters are listed below. If a
182 parameter is not specified the default value is used. When initially
183 configuring a bond, it is recommended "tail -f /var/log/messages" be
184 run in a separate window to watch for bonding driver error messages.
186 It is critical that either the miimon or arp_interval and
187 arp_ip_target parameters be specified, otherwise serious network
188 degradation will occur during link failures. Very few devices do not
189 support at least miimon, so there is really no reason not to use it.
191 Options with textual values will accept either the text name
192 or, for backwards compatibility, the option value. E.g.,
193 "mode=802.3ad" and "mode=4" set the same mode.
195 The parameters are as follows:
199 Specifies the ARP link monitoring frequency in milliseconds.
201 The ARP monitor works by periodically checking the slave
202 devices to determine whether they have sent or received
203 traffic recently (the precise criteria depends upon the
204 bonding mode, and the state of the slave). Regular traffic is
205 generated via ARP probes issued for the addresses specified by
206 the arp_ip_target option.
208 This behavior can be modified by the arp_validate option,
211 If ARP monitoring is used in an etherchannel compatible mode
212 (modes 0 and 2), the switch should be configured in a mode
213 that evenly distributes packets across all links. If the
214 switch is configured to distribute the packets in an XOR
215 fashion, all replies from the ARP targets will be received on
216 the same link which could cause the other team members to
217 fail. ARP monitoring should not be used in conjunction with
218 miimon. A value of 0 disables ARP monitoring. The default
223 Specifies the IP addresses to use as ARP monitoring peers when
224 arp_interval is > 0. These are the targets of the ARP request
225 sent to determine the health of the link to the targets.
226 Specify these values in ddd.ddd.ddd.ddd format. Multiple IP
227 addresses must be separated by a comma. At least one IP
228 address must be given for ARP monitoring to function. The
229 maximum number of targets that can be specified is 16. The
230 default value is no IP addresses.
234 Specifies whether or not ARP probes and replies should be
235 validated in the active-backup mode. This causes the ARP
236 monitor to examine the incoming ARP requests and replies, and
237 only consider a slave to be up if it is receiving the
238 appropriate ARP traffic.
244 No validation is performed. This is the default.
248 Validation is performed only for the active slave.
252 Validation is performed only for backup slaves.
256 Validation is performed for all slaves.
258 For the active slave, the validation checks ARP replies to
259 confirm that they were generated by an arp_ip_target. Since
260 backup slaves do not typically receive these replies, the
261 validation performed for backup slaves is on the ARP request
262 sent out via the active slave. It is possible that some
263 switch or network configurations may result in situations
264 wherein the backup slaves do not receive the ARP requests; in
265 such a situation, validation of backup slaves must be
268 This option is useful in network configurations in which
269 multiple bonding hosts are concurrently issuing ARPs to one or
270 more targets beyond a common switch. Should the link between
271 the switch and target fail (but not the switch itself), the
272 probe traffic generated by the multiple bonding instances will
273 fool the standard ARP monitor into considering the links as
274 still up. Use of the arp_validate option can resolve this, as
275 the ARP monitor will only consider ARP requests and replies
276 associated with its own instance of bonding.
278 This option was added in bonding version 3.1.0.
282 Specifies the time, in milliseconds, to wait before disabling
283 a slave after a link failure has been detected. This option
284 is only valid for the miimon link monitor. The downdelay
285 value should be a multiple of the miimon value; if not, it
286 will be rounded down to the nearest multiple. The default
291 Specifies whether active-backup mode should set all slaves to
292 the same MAC address at enslavement (the traditional
293 behavior), or, when enabled, perform special handling of the
294 bond's MAC address in accordance with the selected policy.
300 This setting disables fail_over_mac, and causes
301 bonding to set all slaves of an active-backup bond to
302 the same MAC address at enslavement time. This is the
307 The "active" fail_over_mac policy indicates that the
308 MAC address of the bond should always be the MAC
309 address of the currently active slave. The MAC
310 address of the slaves is not changed; instead, the MAC
311 address of the bond changes during a failover.
313 This policy is useful for devices that cannot ever
314 alter their MAC address, or for devices that refuse
315 incoming broadcasts with their own source MAC (which
316 interferes with the ARP monitor).
318 The down side of this policy is that every device on
319 the network must be updated via gratuitous ARP,
320 vs. just updating a switch or set of switches (which
321 often takes place for any traffic, not just ARP
322 traffic, if the switch snoops incoming traffic to
323 update its tables) for the traditional method. If the
324 gratuitous ARP is lost, communication may be
327 When this policy is used in conjuction with the mii
328 monitor, devices which assert link up prior to being
329 able to actually transmit and receive are particularly
330 susecptible to loss of the gratuitous ARP, and an
331 appropriate updelay setting may be required.
335 The "follow" fail_over_mac policy causes the MAC
336 address of the bond to be selected normally (normally
337 the MAC address of the first slave added to the bond).
338 However, the second and subsequent slaves are not set
339 to this MAC address while they are in a backup role; a
340 slave is programmed with the bond's MAC address at
341 failover time (and the formerly active slave receives
342 the newly active slave's MAC address).
344 This policy is useful for multiport devices that
345 either become confused or incur a performance penalty
346 when multiple ports are programmed with the same MAC
350 The default policy is none, unless the first slave cannot
351 change its MAC address, in which case the active policy is
354 This option may be modified via sysfs only when no slaves are
357 This option was added in bonding version 3.2.0. The "follow"
358 policy was added in bonding version 3.3.0.
362 Option specifying the rate in which we'll ask our link partner
363 to transmit LACPDU packets in 802.3ad mode. Possible values
367 Request partner to transmit LACPDUs every 30 seconds
370 Request partner to transmit LACPDUs every 1 second
376 Specifies the number of bonding devices to create for this
377 instance of the bonding driver. E.g., if max_bonds is 3, and
378 the bonding driver is not already loaded, then bond0, bond1
379 and bond2 will be created. The default value is 1.
383 Specifies the MII link monitoring frequency in milliseconds.
384 This determines how often the link state of each slave is
385 inspected for link failures. A value of zero disables MII
386 link monitoring. A value of 100 is a good starting point.
387 The use_carrier option, below, affects how the link state is
388 determined. See the High Availability section for additional
389 information. The default value is 0.
393 Specifies one of the bonding policies. The default is
394 balance-rr (round robin). Possible values are:
398 Round-robin policy: Transmit packets in sequential
399 order from the first available slave through the
400 last. This mode provides load balancing and fault
405 Active-backup policy: Only one slave in the bond is
406 active. A different slave becomes active if, and only
407 if, the active slave fails. The bond's MAC address is
408 externally visible on only one port (network adapter)
409 to avoid confusing the switch.
411 In bonding version 2.6.2 or later, when a failover
412 occurs in active-backup mode, bonding will issue one
413 or more gratuitous ARPs on the newly active slave.
414 One gratuitous ARP is issued for the bonding master
415 interface and each VLAN interfaces configured above
416 it, provided that the interface has at least one IP
417 address configured. Gratuitous ARPs issued for VLAN
418 interfaces are tagged with the appropriate VLAN id.
420 This mode provides fault tolerance. The primary
421 option, documented below, affects the behavior of this
426 XOR policy: Transmit based on the selected transmit
427 hash policy. The default policy is a simple [(source
428 MAC address XOR'd with destination MAC address) modulo
429 slave count]. Alternate transmit policies may be
430 selected via the xmit_hash_policy option, described
433 This mode provides load balancing and fault tolerance.
437 Broadcast policy: transmits everything on all slave
438 interfaces. This mode provides fault tolerance.
442 IEEE 802.3ad Dynamic link aggregation. Creates
443 aggregation groups that share the same speed and
444 duplex settings. Utilizes all slaves in the active
445 aggregator according to the 802.3ad specification.
447 Slave selection for outgoing traffic is done according
448 to the transmit hash policy, which may be changed from
449 the default simple XOR policy via the xmit_hash_policy
450 option, documented below. Note that not all transmit
451 policies may be 802.3ad compliant, particularly in
452 regards to the packet mis-ordering requirements of
453 section 43.2.4 of the 802.3ad standard. Differing
454 peer implementations will have varying tolerances for
459 1. Ethtool support in the base drivers for retrieving
460 the speed and duplex of each slave.
462 2. A switch that supports IEEE 802.3ad Dynamic link
465 Most switches will require some type of configuration
466 to enable 802.3ad mode.
470 Adaptive transmit load balancing: channel bonding that
471 does not require any special switch support. The
472 outgoing traffic is distributed according to the
473 current load (computed relative to the speed) on each
474 slave. Incoming traffic is received by the current
475 slave. If the receiving slave fails, another slave
476 takes over the MAC address of the failed receiving
481 Ethtool support in the base drivers for retrieving the
486 Adaptive load balancing: includes balance-tlb plus
487 receive load balancing (rlb) for IPV4 traffic, and
488 does not require any special switch support. The
489 receive load balancing is achieved by ARP negotiation.
490 The bonding driver intercepts the ARP Replies sent by
491 the local system on their way out and overwrites the
492 source hardware address with the unique hardware
493 address of one of the slaves in the bond such that
494 different peers use different hardware addresses for
497 Receive traffic from connections created by the server
498 is also balanced. When the local system sends an ARP
499 Request the bonding driver copies and saves the peer's
500 IP information from the ARP packet. When the ARP
501 Reply arrives from the peer, its hardware address is
502 retrieved and the bonding driver initiates an ARP
503 reply to this peer assigning it to one of the slaves
504 in the bond. A problematic outcome of using ARP
505 negotiation for balancing is that each time that an
506 ARP request is broadcast it uses the hardware address
507 of the bond. Hence, peers learn the hardware address
508 of the bond and the balancing of receive traffic
509 collapses to the current slave. This is handled by
510 sending updates (ARP Replies) to all the peers with
511 their individually assigned hardware address such that
512 the traffic is redistributed. Receive traffic is also
513 redistributed when a new slave is added to the bond
514 and when an inactive slave is re-activated. The
515 receive load is distributed sequentially (round robin)
516 among the group of highest speed slaves in the bond.
518 When a link is reconnected or a new slave joins the
519 bond the receive traffic is redistributed among all
520 active slaves in the bond by initiating ARP Replies
521 with the selected MAC address to each of the
522 clients. The updelay parameter (detailed below) must
523 be set to a value equal or greater than the switch's
524 forwarding delay so that the ARP Replies sent to the
525 peers will not be blocked by the switch.
529 1. Ethtool support in the base drivers for retrieving
530 the speed of each slave.
532 2. Base driver support for setting the hardware
533 address of a device while it is open. This is
534 required so that there will always be one slave in the
535 team using the bond hardware address (the
536 curr_active_slave) while having a unique hardware
537 address for each slave in the bond. If the
538 curr_active_slave fails its hardware address is
539 swapped with the new curr_active_slave that was
544 Specifies the number of gratuitous ARPs to be issued after a
545 failover event. One gratuitous ARP is issued immediately after
546 the failover, subsequent ARPs are sent at a rate of one per link
547 monitor interval (arp_interval or miimon, whichever is active).
549 The valid range is 0 - 255; the default value is 1. This option
550 affects only the active-backup mode. This option was added for
551 bonding version 3.3.0.
555 A string (eth0, eth2, etc) specifying which slave is the
556 primary device. The specified device will always be the
557 active slave while it is available. Only when the primary is
558 off-line will alternate devices be used. This is useful when
559 one slave is preferred over another, e.g., when one slave has
560 higher throughput than another.
562 The primary option is only valid for active-backup mode.
566 Specifies the time, in milliseconds, to wait before enabling a
567 slave after a link recovery has been detected. This option is
568 only valid for the miimon link monitor. The updelay value
569 should be a multiple of the miimon value; if not, it will be
570 rounded down to the nearest multiple. The default value is 0.
574 Specifies whether or not miimon should use MII or ETHTOOL
575 ioctls vs. netif_carrier_ok() to determine the link
576 status. The MII or ETHTOOL ioctls are less efficient and
577 utilize a deprecated calling sequence within the kernel. The
578 netif_carrier_ok() relies on the device driver to maintain its
579 state with netif_carrier_on/off; at this writing, most, but
580 not all, device drivers support this facility.
582 If bonding insists that the link is up when it should not be,
583 it may be that your network device driver does not support
584 netif_carrier_on/off. The default state for netif_carrier is
585 "carrier on," so if a driver does not support netif_carrier,
586 it will appear as if the link is always up. In this case,
587 setting use_carrier to 0 will cause bonding to revert to the
588 MII / ETHTOOL ioctl method to determine the link state.
590 A value of 1 enables the use of netif_carrier_ok(), a value of
591 0 will use the deprecated MII / ETHTOOL ioctls. The default
596 Selects the transmit hash policy to use for slave selection in
597 balance-xor and 802.3ad modes. Possible values are:
601 Uses XOR of hardware MAC addresses to generate the
604 (source MAC XOR destination MAC) modulo slave count
606 This algorithm will place all traffic to a particular
607 network peer on the same slave.
609 This algorithm is 802.3ad compliant.
613 This policy uses a combination of layer2 and layer3
614 protocol information to generate the hash.
616 Uses XOR of hardware MAC addresses and IP addresses to
617 generate the hash. The formula is
619 (((source IP XOR dest IP) AND 0xffff) XOR
620 ( source MAC XOR destination MAC ))
623 This algorithm will place all traffic to a particular
624 network peer on the same slave. For non-IP traffic,
625 the formula is the same as for the layer2 transmit
628 This policy is intended to provide a more balanced
629 distribution of traffic than layer2 alone, especially
630 in environments where a layer3 gateway device is
631 required to reach most destinations.
633 This algorithm is 802.3ad complient.
637 This policy uses upper layer protocol information,
638 when available, to generate the hash. This allows for
639 traffic to a particular network peer to span multiple
640 slaves, although a single connection will not span
643 The formula for unfragmented TCP and UDP packets is
645 ((source port XOR dest port) XOR
646 ((source IP XOR dest IP) AND 0xffff)
649 For fragmented TCP or UDP packets and all other IP
650 protocol traffic, the source and destination port
651 information is omitted. For non-IP traffic, the
652 formula is the same as for the layer2 transmit hash
655 This policy is intended to mimic the behavior of
656 certain switches, notably Cisco switches with PFC2 as
657 well as some Foundry and IBM products.
659 This algorithm is not fully 802.3ad compliant. A
660 single TCP or UDP conversation containing both
661 fragmented and unfragmented packets will see packets
662 striped across two interfaces. This may result in out
663 of order delivery. Most traffic types will not meet
664 this criteria, as TCP rarely fragments traffic, and
665 most UDP traffic is not involved in extended
666 conversations. Other implementations of 802.3ad may
667 or may not tolerate this noncompliance.
669 The default value is layer2. This option was added in bonding
670 version 2.6.3. In earlier versions of bonding, this parameter
671 does not exist, and the layer2 policy is the only policy. The
672 layer2+3 value was added for bonding version 3.2.2.
675 3. Configuring Bonding Devices
676 ==============================
678 You can configure bonding using either your distro's network
679 initialization scripts, or manually using either ifenslave or the
680 sysfs interface. Distros generally use one of two packages for the
681 network initialization scripts: initscripts or sysconfig. Recent
682 versions of these packages have support for bonding, while older
685 We will first describe the options for configuring bonding for
686 distros using versions of initscripts and sysconfig with full or
687 partial support for bonding, then provide information on enabling
688 bonding without support from the network initialization scripts (i.e.,
689 older versions of initscripts or sysconfig).
691 If you're unsure whether your distro uses sysconfig or
692 initscripts, or don't know if it's new enough, have no fear.
693 Determining this is fairly straightforward.
695 First, issue the command:
699 It will respond with a line of text starting with either
700 "initscripts" or "sysconfig," followed by some numbers. This is the
701 package that provides your network initialization scripts.
703 Next, to determine if your installation supports bonding,
706 $ grep ifenslave /sbin/ifup
708 If this returns any matches, then your initscripts or
709 sysconfig has support for bonding.
711 3.1 Configuration with Sysconfig Support
712 ----------------------------------------
714 This section applies to distros using a version of sysconfig
715 with bonding support, for example, SuSE Linux Enterprise Server 9.
717 SuSE SLES 9's networking configuration system does support
718 bonding, however, at this writing, the YaST system configuration
719 front end does not provide any means to work with bonding devices.
720 Bonding devices can be managed by hand, however, as follows.
722 First, if they have not already been configured, configure the
723 slave devices. On SLES 9, this is most easily done by running the
724 yast2 sysconfig configuration utility. The goal is for to create an
725 ifcfg-id file for each slave device. The simplest way to accomplish
726 this is to configure the devices for DHCP (this is only to get the
727 file ifcfg-id file created; see below for some issues with DHCP). The
728 name of the configuration file for each device will be of the form:
730 ifcfg-id-xx:xx:xx:xx:xx:xx
732 Where the "xx" portion will be replaced with the digits from
733 the device's permanent MAC address.
735 Once the set of ifcfg-id-xx:xx:xx:xx:xx:xx files has been
736 created, it is necessary to edit the configuration files for the slave
737 devices (the MAC addresses correspond to those of the slave devices).
738 Before editing, the file will contain multiple lines, and will look
744 UNIQUE='XNzu.WeZGOGF+4wE'
745 _nm_name='bus-pci-0001:61:01.0'
747 Change the BOOTPROTO and STARTMODE lines to the following:
752 Do not alter the UNIQUE or _nm_name lines. Remove any other
753 lines (USERCTL, etc).
755 Once the ifcfg-id-xx:xx:xx:xx:xx:xx files have been modified,
756 it's time to create the configuration file for the bonding device
757 itself. This file is named ifcfg-bondX, where X is the number of the
758 bonding device to create, starting at 0. The first such file is
759 ifcfg-bond0, the second is ifcfg-bond1, and so on. The sysconfig
760 network configuration system will correctly start multiple instances
763 The contents of the ifcfg-bondX file is as follows:
766 BROADCAST="10.0.2.255"
768 NETMASK="255.255.0.0"
773 BONDING_MODULE_OPTS="mode=active-backup miimon=100"
774 BONDING_SLAVE0="eth0"
775 BONDING_SLAVE1="bus-pci-0000:06:08.1"
777 Replace the sample BROADCAST, IPADDR, NETMASK and NETWORK
778 values with the appropriate values for your network.
780 The STARTMODE specifies when the device is brought online.
781 The possible values are:
783 onboot: The device is started at boot time. If you're not
784 sure, this is probably what you want.
786 manual: The device is started only when ifup is called
787 manually. Bonding devices may be configured this
788 way if you do not wish them to start automatically
789 at boot for some reason.
791 hotplug: The device is started by a hotplug event. This is not
792 a valid choice for a bonding device.
794 off or ignore: The device configuration is ignored.
796 The line BONDING_MASTER='yes' indicates that the device is a
797 bonding master device. The only useful value is "yes."
799 The contents of BONDING_MODULE_OPTS are supplied to the
800 instance of the bonding module for this device. Specify the options
801 for the bonding mode, link monitoring, and so on here. Do not include
802 the max_bonds bonding parameter; this will confuse the configuration
803 system if you have multiple bonding devices.
805 Finally, supply one BONDING_SLAVEn="slave device" for each
806 slave. where "n" is an increasing value, one for each slave. The
807 "slave device" is either an interface name, e.g., "eth0", or a device
808 specifier for the network device. The interface name is easier to
809 find, but the ethN names are subject to change at boot time if, e.g.,
810 a device early in the sequence has failed. The device specifiers
811 (bus-pci-0000:06:08.1 in the example above) specify the physical
812 network device, and will not change unless the device's bus location
813 changes (for example, it is moved from one PCI slot to another). The
814 example above uses one of each type for demonstration purposes; most
815 configurations will choose one or the other for all slave devices.
817 When all configuration files have been modified or created,
818 networking must be restarted for the configuration changes to take
819 effect. This can be accomplished via the following:
821 # /etc/init.d/network restart
823 Note that the network control script (/sbin/ifdown) will
824 remove the bonding module as part of the network shutdown processing,
825 so it is not necessary to remove the module by hand if, e.g., the
826 module parameters have changed.
828 Also, at this writing, YaST/YaST2 will not manage bonding
829 devices (they do not show bonding interfaces on its list of network
830 devices). It is necessary to edit the configuration file by hand to
831 change the bonding configuration.
833 Additional general options and details of the ifcfg file
834 format can be found in an example ifcfg template file:
836 /etc/sysconfig/network/ifcfg.template
838 Note that the template does not document the various BONDING_
839 settings described above, but does describe many of the other options.
841 3.1.1 Using DHCP with Sysconfig
842 -------------------------------
844 Under sysconfig, configuring a device with BOOTPROTO='dhcp'
845 will cause it to query DHCP for its IP address information. At this
846 writing, this does not function for bonding devices; the scripts
847 attempt to obtain the device address from DHCP prior to adding any of
848 the slave devices. Without active slaves, the DHCP requests are not
851 3.1.2 Configuring Multiple Bonds with Sysconfig
852 -----------------------------------------------
854 The sysconfig network initialization system is capable of
855 handling multiple bonding devices. All that is necessary is for each
856 bonding instance to have an appropriately configured ifcfg-bondX file
857 (as described above). Do not specify the "max_bonds" parameter to any
858 instance of bonding, as this will confuse sysconfig. If you require
859 multiple bonding devices with identical parameters, create multiple
862 Because the sysconfig scripts supply the bonding module
863 options in the ifcfg-bondX file, it is not necessary to add them to
864 the system /etc/modules.conf or /etc/modprobe.conf configuration file.
866 3.2 Configuration with Initscripts Support
867 ------------------------------------------
869 This section applies to distros using a recent version of
870 initscripts with bonding support, for example, Red Hat Enterprise Linux
871 version 3 or later, Fedora, etc. On these systems, the network
872 initialization scripts have knowledge of bonding, and can be configured to
873 control bonding devices. Note that older versions of the initscripts
874 package have lower levels of support for bonding; this will be noted where
877 These distros will not automatically load the network adapter
878 driver unless the ethX device is configured with an IP address.
879 Because of this constraint, users must manually configure a
880 network-script file for all physical adapters that will be members of
881 a bondX link. Network script files are located in the directory:
883 /etc/sysconfig/network-scripts
885 The file name must be prefixed with "ifcfg-eth" and suffixed
886 with the adapter's physical adapter number. For example, the script
887 for eth0 would be named /etc/sysconfig/network-scripts/ifcfg-eth0.
888 Place the following text in the file:
897 The DEVICE= line will be different for every ethX device and
898 must correspond with the name of the file, i.e., ifcfg-eth1 must have
899 a device line of DEVICE=eth1. The setting of the MASTER= line will
900 also depend on the final bonding interface name chosen for your bond.
901 As with other network devices, these typically start at 0, and go up
902 one for each device, i.e., the first bonding instance is bond0, the
903 second is bond1, and so on.
905 Next, create a bond network script. The file name for this
906 script will be /etc/sysconfig/network-scripts/ifcfg-bondX where X is
907 the number of the bond. For bond0 the file is named "ifcfg-bond0",
908 for bond1 it is named "ifcfg-bond1", and so on. Within that file,
909 place the following text:
913 NETMASK=255.255.255.0
915 BROADCAST=192.168.1.255
920 Be sure to change the networking specific lines (IPADDR,
921 NETMASK, NETWORK and BROADCAST) to match your network configuration.
923 For later versions of initscripts, such as that found with Fedora
924 7 and Red Hat Enterprise Linux version 5 (or later), it is possible, and,
925 indeed, preferable, to specify the bonding options in the ifcfg-bond0
926 file, e.g. a line of the format:
928 BONDING_OPTS="mode=active-backup arp_interval=60 arp_ip_target=+192.168.1.254"
930 will configure the bond with the specified options. The options
931 specified in BONDING_OPTS are identical to the bonding module parameters
932 except for the arp_ip_target field. Each target should be included as a
933 separate option and should be preceded by a '+' to indicate it should be
934 added to the list of queried targets, e.g.,
936 arp_ip_target=+192.168.1.1 arp_ip_target=+192.168.1.2
938 is the proper syntax to specify multiple targets. When specifying
939 options via BONDING_OPTS, it is not necessary to edit /etc/modules.conf or
942 For older versions of initscripts that do not support
943 BONDING_OPTS, it is necessary to edit /etc/modules.conf (or
944 /etc/modprobe.conf, depending upon your distro) to load the bonding module
945 with your desired options when the bond0 interface is brought up. The
946 following lines in /etc/modules.conf (or modprobe.conf) will load the
947 bonding module, and select its options:
950 options bond0 mode=balance-alb miimon=100
952 Replace the sample parameters with the appropriate set of
953 options for your configuration.
955 Finally run "/etc/rc.d/init.d/network restart" as root. This
956 will restart the networking subsystem and your bond link should be now
959 3.2.1 Using DHCP with Initscripts
960 ---------------------------------
962 Recent versions of initscripts (the versions supplied with Fedora
963 Core 3 and Red Hat Enterprise Linux 4, or later versions, are reported to
964 work) have support for assigning IP information to bonding devices via
967 To configure bonding for DHCP, configure it as described
968 above, except replace the line "BOOTPROTO=none" with "BOOTPROTO=dhcp"
969 and add a line consisting of "TYPE=Bonding". Note that the TYPE value
972 3.2.2 Configuring Multiple Bonds with Initscripts
973 -------------------------------------------------
975 Initscripts packages that are included with Fedora 7 and Red Hat
976 Enterprise Linux 5 support multiple bonding interfaces by simply
977 specifying the appropriate BONDING_OPTS= in ifcfg-bondX where X is the
978 number of the bond. This support requires sysfs support in the kernel,
979 and a bonding driver of version 3.0.0 or later. Other configurations may
980 not support this method for specifying multiple bonding interfaces; for
981 those instances, see the "Configuring Multiple Bonds Manually" section,
984 3.3 Configuring Bonding Manually with Ifenslave
985 -----------------------------------------------
987 This section applies to distros whose network initialization
988 scripts (the sysconfig or initscripts package) do not have specific
989 knowledge of bonding. One such distro is SuSE Linux Enterprise Server
992 The general method for these systems is to place the bonding
993 module parameters into /etc/modules.conf or /etc/modprobe.conf (as
994 appropriate for the installed distro), then add modprobe and/or
995 ifenslave commands to the system's global init script. The name of
996 the global init script differs; for sysconfig, it is
997 /etc/init.d/boot.local and for initscripts it is /etc/rc.d/rc.local.
999 For example, if you wanted to make a simple bond of two e100
1000 devices (presumed to be eth0 and eth1), and have it persist across
1001 reboots, edit the appropriate file (/etc/init.d/boot.local or
1002 /etc/rc.d/rc.local), and add the following:
1004 modprobe bonding mode=balance-alb miimon=100
1006 ifconfig bond0 192.168.1.1 netmask 255.255.255.0 up
1007 ifenslave bond0 eth0
1008 ifenslave bond0 eth1
1010 Replace the example bonding module parameters and bond0
1011 network configuration (IP address, netmask, etc) with the appropriate
1012 values for your configuration.
1014 Unfortunately, this method will not provide support for the
1015 ifup and ifdown scripts on the bond devices. To reload the bonding
1016 configuration, it is necessary to run the initialization script, e.g.,
1018 # /etc/init.d/boot.local
1022 # /etc/rc.d/rc.local
1024 It may be desirable in such a case to create a separate script
1025 which only initializes the bonding configuration, then call that
1026 separate script from within boot.local. This allows for bonding to be
1027 enabled without re-running the entire global init script.
1029 To shut down the bonding devices, it is necessary to first
1030 mark the bonding device itself as being down, then remove the
1031 appropriate device driver modules. For our example above, you can do
1034 # ifconfig bond0 down
1038 Again, for convenience, it may be desirable to create a script
1039 with these commands.
1042 3.3.1 Configuring Multiple Bonds Manually
1043 -----------------------------------------
1045 This section contains information on configuring multiple
1046 bonding devices with differing options for those systems whose network
1047 initialization scripts lack support for configuring multiple bonds.
1049 If you require multiple bonding devices, but all with the same
1050 options, you may wish to use the "max_bonds" module parameter,
1053 To create multiple bonding devices with differing options, it is
1054 preferrable to use bonding parameters exported by sysfs, documented in the
1057 For versions of bonding without sysfs support, the only means to
1058 provide multiple instances of bonding with differing options is to load
1059 the bonding driver multiple times. Note that current versions of the
1060 sysconfig network initialization scripts handle this automatically; if
1061 your distro uses these scripts, no special action is needed. See the
1062 section Configuring Bonding Devices, above, if you're not sure about your
1063 network initialization scripts.
1065 To load multiple instances of the module, it is necessary to
1066 specify a different name for each instance (the module loading system
1067 requires that every loaded module, even multiple instances of the same
1068 module, have a unique name). This is accomplished by supplying multiple
1069 sets of bonding options in /etc/modprobe.conf, for example:
1072 options bond0 -o bond0 mode=balance-rr miimon=100
1075 options bond1 -o bond1 mode=balance-alb miimon=50
1077 will load the bonding module two times. The first instance is
1078 named "bond0" and creates the bond0 device in balance-rr mode with an
1079 miimon of 100. The second instance is named "bond1" and creates the
1080 bond1 device in balance-alb mode with an miimon of 50.
1082 In some circumstances (typically with older distributions),
1083 the above does not work, and the second bonding instance never sees
1084 its options. In that case, the second options line can be substituted
1087 install bond1 /sbin/modprobe --ignore-install bonding -o bond1 \
1088 mode=balance-alb miimon=50
1090 This may be repeated any number of times, specifying a new and
1091 unique name in place of bond1 for each subsequent instance.
1093 It has been observed that some Red Hat supplied kernels are unable
1094 to rename modules at load time (the "-o bond1" part). Attempts to pass
1095 that option to modprobe will produce an "Operation not permitted" error.
1096 This has been reported on some Fedora Core kernels, and has been seen on
1097 RHEL 4 as well. On kernels exhibiting this problem, it will be impossible
1098 to configure multiple bonds with differing parameters (as they are older
1099 kernels, and also lack sysfs support).
1101 3.4 Configuring Bonding Manually via Sysfs
1102 ------------------------------------------
1104 Starting with version 3.0.0, Channel Bonding may be configured
1105 via the sysfs interface. This interface allows dynamic configuration
1106 of all bonds in the system without unloading the module. It also
1107 allows for adding and removing bonds at runtime. Ifenslave is no
1108 longer required, though it is still supported.
1110 Use of the sysfs interface allows you to use multiple bonds
1111 with different configurations without having to reload the module.
1112 It also allows you to use multiple, differently configured bonds when
1113 bonding is compiled into the kernel.
1115 You must have the sysfs filesystem mounted to configure
1116 bonding this way. The examples in this document assume that you
1117 are using the standard mount point for sysfs, e.g. /sys. If your
1118 sysfs filesystem is mounted elsewhere, you will need to adjust the
1119 example paths accordingly.
1121 Creating and Destroying Bonds
1122 -----------------------------
1123 To add a new bond foo:
1124 # echo +foo > /sys/class/net/bonding_masters
1126 To remove an existing bond bar:
1127 # echo -bar > /sys/class/net/bonding_masters
1129 To show all existing bonds:
1130 # cat /sys/class/net/bonding_masters
1132 NOTE: due to 4K size limitation of sysfs files, this list may be
1133 truncated if you have more than a few hundred bonds. This is unlikely
1134 to occur under normal operating conditions.
1136 Adding and Removing Slaves
1137 --------------------------
1138 Interfaces may be enslaved to a bond using the file
1139 /sys/class/net/<bond>/bonding/slaves. The semantics for this file
1140 are the same as for the bonding_masters file.
1142 To enslave interface eth0 to bond bond0:
1144 # echo +eth0 > /sys/class/net/bond0/bonding/slaves
1146 To free slave eth0 from bond bond0:
1147 # echo -eth0 > /sys/class/net/bond0/bonding/slaves
1149 When an interface is enslaved to a bond, symlinks between the
1150 two are created in the sysfs filesystem. In this case, you would get
1151 /sys/class/net/bond0/slave_eth0 pointing to /sys/class/net/eth0, and
1152 /sys/class/net/eth0/master pointing to /sys/class/net/bond0.
1154 This means that you can tell quickly whether or not an
1155 interface is enslaved by looking for the master symlink. Thus:
1156 # echo -eth0 > /sys/class/net/eth0/master/bonding/slaves
1157 will free eth0 from whatever bond it is enslaved to, regardless of
1158 the name of the bond interface.
1160 Changing a Bond's Configuration
1161 -------------------------------
1162 Each bond may be configured individually by manipulating the
1163 files located in /sys/class/net/<bond name>/bonding
1165 The names of these files correspond directly with the command-
1166 line parameters described elsewhere in this file, and, with the
1167 exception of arp_ip_target, they accept the same values. To see the
1168 current setting, simply cat the appropriate file.
1170 A few examples will be given here; for specific usage
1171 guidelines for each parameter, see the appropriate section in this
1174 To configure bond0 for balance-alb mode:
1175 # ifconfig bond0 down
1176 # echo 6 > /sys/class/net/bond0/bonding/mode
1178 # echo balance-alb > /sys/class/net/bond0/bonding/mode
1179 NOTE: The bond interface must be down before the mode can be
1182 To enable MII monitoring on bond0 with a 1 second interval:
1183 # echo 1000 > /sys/class/net/bond0/bonding/miimon
1184 NOTE: If ARP monitoring is enabled, it will disabled when MII
1185 monitoring is enabled, and vice-versa.
1188 # echo +192.168.0.100 > /sys/class/net/bond0/bonding/arp_ip_target
1189 # echo +192.168.0.101 > /sys/class/net/bond0/bonding/arp_ip_target
1190 NOTE: up to 10 target addresses may be specified.
1192 To remove an ARP target:
1193 # echo -192.168.0.100 > /sys/class/net/bond0/bonding/arp_ip_target
1195 Example Configuration
1196 ---------------------
1197 We begin with the same example that is shown in section 3.3,
1198 executed with sysfs, and without using ifenslave.
1200 To make a simple bond of two e100 devices (presumed to be eth0
1201 and eth1), and have it persist across reboots, edit the appropriate
1202 file (/etc/init.d/boot.local or /etc/rc.d/rc.local), and add the
1207 echo balance-alb > /sys/class/net/bond0/bonding/mode
1208 ifconfig bond0 192.168.1.1 netmask 255.255.255.0 up
1209 echo 100 > /sys/class/net/bond0/bonding/miimon
1210 echo +eth0 > /sys/class/net/bond0/bonding/slaves
1211 echo +eth1 > /sys/class/net/bond0/bonding/slaves
1213 To add a second bond, with two e1000 interfaces in
1214 active-backup mode, using ARP monitoring, add the following lines to
1218 echo +bond1 > /sys/class/net/bonding_masters
1219 echo active-backup > /sys/class/net/bond1/bonding/mode
1220 ifconfig bond1 192.168.2.1 netmask 255.255.255.0 up
1221 echo +192.168.2.100 /sys/class/net/bond1/bonding/arp_ip_target
1222 echo 2000 > /sys/class/net/bond1/bonding/arp_interval
1223 echo +eth2 > /sys/class/net/bond1/bonding/slaves
1224 echo +eth3 > /sys/class/net/bond1/bonding/slaves
1227 4. Querying Bonding Configuration
1228 =================================
1230 4.1 Bonding Configuration
1231 -------------------------
1233 Each bonding device has a read-only file residing in the
1234 /proc/net/bonding directory. The file contents include information
1235 about the bonding configuration, options and state of each slave.
1237 For example, the contents of /proc/net/bonding/bond0 after the
1238 driver is loaded with parameters of mode=0 and miimon=1000 is
1239 generally as follows:
1241 Ethernet Channel Bonding Driver: 2.6.1 (October 29, 2004)
1242 Bonding Mode: load balancing (round-robin)
1243 Currently Active Slave: eth0
1245 MII Polling Interval (ms): 1000
1249 Slave Interface: eth1
1251 Link Failure Count: 1
1253 Slave Interface: eth0
1255 Link Failure Count: 1
1257 The precise format and contents will change depending upon the
1258 bonding configuration, state, and version of the bonding driver.
1260 4.2 Network configuration
1261 -------------------------
1263 The network configuration can be inspected using the ifconfig
1264 command. Bonding devices will have the MASTER flag set; Bonding slave
1265 devices will have the SLAVE flag set. The ifconfig output does not
1266 contain information on which slaves are associated with which masters.
1268 In the example below, the bond0 interface is the master
1269 (MASTER) while eth0 and eth1 are slaves (SLAVE). Notice all slaves of
1270 bond0 have the same MAC address (HWaddr) as bond0 for all modes except
1271 TLB and ALB that require a unique MAC address for each slave.
1274 bond0 Link encap:Ethernet HWaddr 00:C0:F0:1F:37:B4
1275 inet addr:XXX.XXX.XXX.YYY Bcast:XXX.XXX.XXX.255 Mask:255.255.252.0
1276 UP BROADCAST RUNNING MASTER MULTICAST MTU:1500 Metric:1
1277 RX packets:7224794 errors:0 dropped:0 overruns:0 frame:0
1278 TX packets:3286647 errors:1 dropped:0 overruns:1 carrier:0
1279 collisions:0 txqueuelen:0
1281 eth0 Link encap:Ethernet HWaddr 00:C0:F0:1F:37:B4
1282 UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
1283 RX packets:3573025 errors:0 dropped:0 overruns:0 frame:0
1284 TX packets:1643167 errors:1 dropped:0 overruns:1 carrier:0
1285 collisions:0 txqueuelen:100
1286 Interrupt:10 Base address:0x1080
1288 eth1 Link encap:Ethernet HWaddr 00:C0:F0:1F:37:B4
1289 UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
1290 RX packets:3651769 errors:0 dropped:0 overruns:0 frame:0
1291 TX packets:1643480 errors:0 dropped:0 overruns:0 carrier:0
1292 collisions:0 txqueuelen:100
1293 Interrupt:9 Base address:0x1400
1295 5. Switch Configuration
1296 =======================
1298 For this section, "switch" refers to whatever system the
1299 bonded devices are directly connected to (i.e., where the other end of
1300 the cable plugs into). This may be an actual dedicated switch device,
1301 or it may be another regular system (e.g., another computer running
1304 The active-backup, balance-tlb and balance-alb modes do not
1305 require any specific configuration of the switch.
1307 The 802.3ad mode requires that the switch have the appropriate
1308 ports configured as an 802.3ad aggregation. The precise method used
1309 to configure this varies from switch to switch, but, for example, a
1310 Cisco 3550 series switch requires that the appropriate ports first be
1311 grouped together in a single etherchannel instance, then that
1312 etherchannel is set to mode "lacp" to enable 802.3ad (instead of
1313 standard EtherChannel).
1315 The balance-rr, balance-xor and broadcast modes generally
1316 require that the switch have the appropriate ports grouped together.
1317 The nomenclature for such a group differs between switches, it may be
1318 called an "etherchannel" (as in the Cisco example, above), a "trunk
1319 group" or some other similar variation. For these modes, each switch
1320 will also have its own configuration options for the switch's transmit
1321 policy to the bond. Typical choices include XOR of either the MAC or
1322 IP addresses. The transmit policy of the two peers does not need to
1323 match. For these three modes, the bonding mode really selects a
1324 transmit policy for an EtherChannel group; all three will interoperate
1325 with another EtherChannel group.
1328 6. 802.1q VLAN Support
1329 ======================
1331 It is possible to configure VLAN devices over a bond interface
1332 using the 8021q driver. However, only packets coming from the 8021q
1333 driver and passing through bonding will be tagged by default. Self
1334 generated packets, for example, bonding's learning packets or ARP
1335 packets generated by either ALB mode or the ARP monitor mechanism, are
1336 tagged internally by bonding itself. As a result, bonding must
1337 "learn" the VLAN IDs configured above it, and use those IDs to tag
1338 self generated packets.
1340 For reasons of simplicity, and to support the use of adapters
1341 that can do VLAN hardware acceleration offloading, the bonding
1342 interface declares itself as fully hardware offloading capable, it gets
1343 the add_vid/kill_vid notifications to gather the necessary
1344 information, and it propagates those actions to the slaves. In case
1345 of mixed adapter types, hardware accelerated tagged packets that
1346 should go through an adapter that is not offloading capable are
1347 "un-accelerated" by the bonding driver so the VLAN tag sits in the
1350 VLAN interfaces *must* be added on top of a bonding interface
1351 only after enslaving at least one slave. The bonding interface has a
1352 hardware address of 00:00:00:00:00:00 until the first slave is added.
1353 If the VLAN interface is created prior to the first enslavement, it
1354 would pick up the all-zeroes hardware address. Once the first slave
1355 is attached to the bond, the bond device itself will pick up the
1356 slave's hardware address, which is then available for the VLAN device.
1358 Also, be aware that a similar problem can occur if all slaves
1359 are released from a bond that still has one or more VLAN interfaces on
1360 top of it. When a new slave is added, the bonding interface will
1361 obtain its hardware address from the first slave, which might not
1362 match the hardware address of the VLAN interfaces (which was
1363 ultimately copied from an earlier slave).
1365 There are two methods to insure that the VLAN device operates
1366 with the correct hardware address if all slaves are removed from a
1369 1. Remove all VLAN interfaces then recreate them
1371 2. Set the bonding interface's hardware address so that it
1372 matches the hardware address of the VLAN interfaces.
1374 Note that changing a VLAN interface's HW address would set the
1375 underlying device -- i.e. the bonding interface -- to promiscuous
1376 mode, which might not be what you want.
1382 The bonding driver at present supports two schemes for
1383 monitoring a slave device's link state: the ARP monitor and the MII
1386 At the present time, due to implementation restrictions in the
1387 bonding driver itself, it is not possible to enable both ARP and MII
1388 monitoring simultaneously.
1390 7.1 ARP Monitor Operation
1391 -------------------------
1393 The ARP monitor operates as its name suggests: it sends ARP
1394 queries to one or more designated peer systems on the network, and
1395 uses the response as an indication that the link is operating. This
1396 gives some assurance that traffic is actually flowing to and from one
1397 or more peers on the local network.
1399 The ARP monitor relies on the device driver itself to verify
1400 that traffic is flowing. In particular, the driver must keep up to
1401 date the last receive time, dev->last_rx, and transmit start time,
1402 dev->trans_start. If these are not updated by the driver, then the
1403 ARP monitor will immediately fail any slaves using that driver, and
1404 those slaves will stay down. If networking monitoring (tcpdump, etc)
1405 shows the ARP requests and replies on the network, then it may be that
1406 your device driver is not updating last_rx and trans_start.
1408 7.2 Configuring Multiple ARP Targets
1409 ------------------------------------
1411 While ARP monitoring can be done with just one target, it can
1412 be useful in a High Availability setup to have several targets to
1413 monitor. In the case of just one target, the target itself may go
1414 down or have a problem making it unresponsive to ARP requests. Having
1415 an additional target (or several) increases the reliability of the ARP
1418 Multiple ARP targets must be separated by commas as follows:
1420 # example options for ARP monitoring with three targets
1422 options bond0 arp_interval=60 arp_ip_target=192.168.0.1,192.168.0.3,192.168.0.9
1424 For just a single target the options would resemble:
1426 # example options for ARP monitoring with one target
1428 options bond0 arp_interval=60 arp_ip_target=192.168.0.100
1431 7.3 MII Monitor Operation
1432 -------------------------
1434 The MII monitor monitors only the carrier state of the local
1435 network interface. It accomplishes this in one of three ways: by
1436 depending upon the device driver to maintain its carrier state, by
1437 querying the device's MII registers, or by making an ethtool query to
1440 If the use_carrier module parameter is 1 (the default value),
1441 then the MII monitor will rely on the driver for carrier state
1442 information (via the netif_carrier subsystem). As explained in the
1443 use_carrier parameter information, above, if the MII monitor fails to
1444 detect carrier loss on the device (e.g., when the cable is physically
1445 disconnected), it may be that the driver does not support
1448 If use_carrier is 0, then the MII monitor will first query the
1449 device's (via ioctl) MII registers and check the link state. If that
1450 request fails (not just that it returns carrier down), then the MII
1451 monitor will make an ethtool ETHOOL_GLINK request to attempt to obtain
1452 the same information. If both methods fail (i.e., the driver either
1453 does not support or had some error in processing both the MII register
1454 and ethtool requests), then the MII monitor will assume the link is
1457 8. Potential Sources of Trouble
1458 ===============================
1460 8.1 Adventures in Routing
1461 -------------------------
1463 When bonding is configured, it is important that the slave
1464 devices not have routes that supersede routes of the master (or,
1465 generally, not have routes at all). For example, suppose the bonding
1466 device bond0 has two slaves, eth0 and eth1, and the routing table is
1469 Kernel IP routing table
1470 Destination Gateway Genmask Flags MSS Window irtt Iface
1471 10.0.0.0 0.0.0.0 255.255.0.0 U 40 0 0 eth0
1472 10.0.0.0 0.0.0.0 255.255.0.0 U 40 0 0 eth1
1473 10.0.0.0 0.0.0.0 255.255.0.0 U 40 0 0 bond0
1474 127.0.0.0 0.0.0.0 255.0.0.0 U 40 0 0 lo
1476 This routing configuration will likely still update the
1477 receive/transmit times in the driver (needed by the ARP monitor), but
1478 may bypass the bonding driver (because outgoing traffic to, in this
1479 case, another host on network 10 would use eth0 or eth1 before bond0).
1481 The ARP monitor (and ARP itself) may become confused by this
1482 configuration, because ARP requests (generated by the ARP monitor)
1483 will be sent on one interface (bond0), but the corresponding reply
1484 will arrive on a different interface (eth0). This reply looks to ARP
1485 as an unsolicited ARP reply (because ARP matches replies on an
1486 interface basis), and is discarded. The MII monitor is not affected
1487 by the state of the routing table.
1489 The solution here is simply to insure that slaves do not have
1490 routes of their own, and if for some reason they must, those routes do
1491 not supersede routes of their master. This should generally be the
1492 case, but unusual configurations or errant manual or automatic static
1493 route additions may cause trouble.
1495 8.2 Ethernet Device Renaming
1496 ----------------------------
1498 On systems with network configuration scripts that do not
1499 associate physical devices directly with network interface names (so
1500 that the same physical device always has the same "ethX" name), it may
1501 be necessary to add some special logic to either /etc/modules.conf or
1502 /etc/modprobe.conf (depending upon which is installed on the system).
1504 For example, given a modules.conf containing the following:
1507 options bond0 mode=some-mode miimon=50
1513 If neither eth0 and eth1 are slaves to bond0, then when the
1514 bond0 interface comes up, the devices may end up reordered. This
1515 happens because bonding is loaded first, then its slave device's
1516 drivers are loaded next. Since no other drivers have been loaded,
1517 when the e1000 driver loads, it will receive eth0 and eth1 for its
1518 devices, but the bonding configuration tries to enslave eth2 and eth3
1519 (which may later be assigned to the tg3 devices).
1521 Adding the following:
1523 add above bonding e1000 tg3
1525 causes modprobe to load e1000 then tg3, in that order, when
1526 bonding is loaded. This command is fully documented in the
1527 modules.conf manual page.
1529 On systems utilizing modprobe.conf (or modprobe.conf.local),
1530 an equivalent problem can occur. In this case, the following can be
1531 added to modprobe.conf (or modprobe.conf.local, as appropriate), as
1532 follows (all on one line; it has been split here for clarity):
1534 install bonding /sbin/modprobe tg3; /sbin/modprobe e1000;
1535 /sbin/modprobe --ignore-install bonding
1537 This will, when loading the bonding module, rather than
1538 performing the normal action, instead execute the provided command.
1539 This command loads the device drivers in the order needed, then calls
1540 modprobe with --ignore-install to cause the normal action to then take
1541 place. Full documentation on this can be found in the modprobe.conf
1542 and modprobe manual pages.
1544 8.3. Painfully Slow Or No Failed Link Detection By Miimon
1545 ---------------------------------------------------------
1547 By default, bonding enables the use_carrier option, which
1548 instructs bonding to trust the driver to maintain carrier state.
1550 As discussed in the options section, above, some drivers do
1551 not support the netif_carrier_on/_off link state tracking system.
1552 With use_carrier enabled, bonding will always see these links as up,
1553 regardless of their actual state.
1555 Additionally, other drivers do support netif_carrier, but do
1556 not maintain it in real time, e.g., only polling the link state at
1557 some fixed interval. In this case, miimon will detect failures, but
1558 only after some long period of time has expired. If it appears that
1559 miimon is very slow in detecting link failures, try specifying
1560 use_carrier=0 to see if that improves the failure detection time. If
1561 it does, then it may be that the driver checks the carrier state at a
1562 fixed interval, but does not cache the MII register values (so the
1563 use_carrier=0 method of querying the registers directly works). If
1564 use_carrier=0 does not improve the failover, then the driver may cache
1565 the registers, or the problem may be elsewhere.
1567 Also, remember that miimon only checks for the device's
1568 carrier state. It has no way to determine the state of devices on or
1569 beyond other ports of a switch, or if a switch is refusing to pass
1570 traffic while still maintaining carrier on.
1575 If running SNMP agents, the bonding driver should be loaded
1576 before any network drivers participating in a bond. This requirement
1577 is due to the interface index (ipAdEntIfIndex) being associated to
1578 the first interface found with a given IP address. That is, there is
1579 only one ipAdEntIfIndex for each IP address. For example, if eth0 and
1580 eth1 are slaves of bond0 and the driver for eth0 is loaded before the
1581 bonding driver, the interface for the IP address will be associated
1582 with the eth0 interface. This configuration is shown below, the IP
1583 address 192.168.1.1 has an interface index of 2 which indexes to eth0
1584 in the ifDescr table (ifDescr.2).
1586 interfaces.ifTable.ifEntry.ifDescr.1 = lo
1587 interfaces.ifTable.ifEntry.ifDescr.2 = eth0
1588 interfaces.ifTable.ifEntry.ifDescr.3 = eth1
1589 interfaces.ifTable.ifEntry.ifDescr.4 = eth2
1590 interfaces.ifTable.ifEntry.ifDescr.5 = eth3
1591 interfaces.ifTable.ifEntry.ifDescr.6 = bond0
1592 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.10.10.10 = 5
1593 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.192.168.1.1 = 2
1594 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.74.20.94 = 4
1595 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.127.0.0.1 = 1
1597 This problem is avoided by loading the bonding driver before
1598 any network drivers participating in a bond. Below is an example of
1599 loading the bonding driver first, the IP address 192.168.1.1 is
1600 correctly associated with ifDescr.2.
1602 interfaces.ifTable.ifEntry.ifDescr.1 = lo
1603 interfaces.ifTable.ifEntry.ifDescr.2 = bond0
1604 interfaces.ifTable.ifEntry.ifDescr.3 = eth0
1605 interfaces.ifTable.ifEntry.ifDescr.4 = eth1
1606 interfaces.ifTable.ifEntry.ifDescr.5 = eth2
1607 interfaces.ifTable.ifEntry.ifDescr.6 = eth3
1608 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.10.10.10 = 6
1609 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.192.168.1.1 = 2
1610 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.74.20.94 = 5
1611 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.127.0.0.1 = 1
1613 While some distributions may not report the interface name in
1614 ifDescr, the association between the IP address and IfIndex remains
1615 and SNMP functions such as Interface_Scan_Next will report that
1618 10. Promiscuous mode
1619 ====================
1621 When running network monitoring tools, e.g., tcpdump, it is
1622 common to enable promiscuous mode on the device, so that all traffic
1623 is seen (instead of seeing only traffic destined for the local host).
1624 The bonding driver handles promiscuous mode changes to the bonding
1625 master device (e.g., bond0), and propagates the setting to the slave
1628 For the balance-rr, balance-xor, broadcast, and 802.3ad modes,
1629 the promiscuous mode setting is propagated to all slaves.
1631 For the active-backup, balance-tlb and balance-alb modes, the
1632 promiscuous mode setting is propagated only to the active slave.
1634 For balance-tlb mode, the active slave is the slave currently
1635 receiving inbound traffic.
1637 For balance-alb mode, the active slave is the slave used as a
1638 "primary." This slave is used for mode-specific control traffic, for
1639 sending to peers that are unassigned or if the load is unbalanced.
1641 For the active-backup, balance-tlb and balance-alb modes, when
1642 the active slave changes (e.g., due to a link failure), the
1643 promiscuous setting will be propagated to the new active slave.
1645 11. Configuring Bonding for High Availability
1646 =============================================
1648 High Availability refers to configurations that provide
1649 maximum network availability by having redundant or backup devices,
1650 links or switches between the host and the rest of the world. The
1651 goal is to provide the maximum availability of network connectivity
1652 (i.e., the network always works), even though other configurations
1653 could provide higher throughput.
1655 11.1 High Availability in a Single Switch Topology
1656 --------------------------------------------------
1658 If two hosts (or a host and a single switch) are directly
1659 connected via multiple physical links, then there is no availability
1660 penalty to optimizing for maximum bandwidth. In this case, there is
1661 only one switch (or peer), so if it fails, there is no alternative
1662 access to fail over to. Additionally, the bonding load balance modes
1663 support link monitoring of their members, so if individual links fail,
1664 the load will be rebalanced across the remaining devices.
1666 See Section 13, "Configuring Bonding for Maximum Throughput"
1667 for information on configuring bonding with one peer device.
1669 11.2 High Availability in a Multiple Switch Topology
1670 ----------------------------------------------------
1672 With multiple switches, the configuration of bonding and the
1673 network changes dramatically. In multiple switch topologies, there is
1674 a trade off between network availability and usable bandwidth.
1676 Below is a sample network, configured to maximize the
1677 availability of the network:
1681 +-----+----+ +-----+----+
1682 | |port2 ISL port2| |
1683 | switch A +--------------------------+ switch B |
1685 +-----+----+ +-----++---+
1688 +-------------+ host1 +---------------+
1691 In this configuration, there is a link between the two
1692 switches (ISL, or inter switch link), and multiple ports connecting to
1693 the outside world ("port3" on each switch). There is no technical
1694 reason that this could not be extended to a third switch.
1696 11.2.1 HA Bonding Mode Selection for Multiple Switch Topology
1697 -------------------------------------------------------------
1699 In a topology such as the example above, the active-backup and
1700 broadcast modes are the only useful bonding modes when optimizing for
1701 availability; the other modes require all links to terminate on the
1702 same peer for them to behave rationally.
1704 active-backup: This is generally the preferred mode, particularly if
1705 the switches have an ISL and play together well. If the
1706 network configuration is such that one switch is specifically
1707 a backup switch (e.g., has lower capacity, higher cost, etc),
1708 then the primary option can be used to insure that the
1709 preferred link is always used when it is available.
1711 broadcast: This mode is really a special purpose mode, and is suitable
1712 only for very specific needs. For example, if the two
1713 switches are not connected (no ISL), and the networks beyond
1714 them are totally independent. In this case, if it is
1715 necessary for some specific one-way traffic to reach both
1716 independent networks, then the broadcast mode may be suitable.
1718 11.2.2 HA Link Monitoring Selection for Multiple Switch Topology
1719 ----------------------------------------------------------------
1721 The choice of link monitoring ultimately depends upon your
1722 switch. If the switch can reliably fail ports in response to other
1723 failures, then either the MII or ARP monitors should work. For
1724 example, in the above example, if the "port3" link fails at the remote
1725 end, the MII monitor has no direct means to detect this. The ARP
1726 monitor could be configured with a target at the remote end of port3,
1727 thus detecting that failure without switch support.
1729 In general, however, in a multiple switch topology, the ARP
1730 monitor can provide a higher level of reliability in detecting end to
1731 end connectivity failures (which may be caused by the failure of any
1732 individual component to pass traffic for any reason). Additionally,
1733 the ARP monitor should be configured with multiple targets (at least
1734 one for each switch in the network). This will insure that,
1735 regardless of which switch is active, the ARP monitor has a suitable
1738 Note, also, that of late many switches now support a functionality
1739 generally referred to as "trunk failover." This is a feature of the
1740 switch that causes the link state of a particular switch port to be set
1741 down (or up) when the state of another switch port goes down (or up).
1742 It's purpose is to propogate link failures from logically "exterior" ports
1743 to the logically "interior" ports that bonding is able to monitor via
1744 miimon. Availability and configuration for trunk failover varies by
1745 switch, but this can be a viable alternative to the ARP monitor when using
1748 12. Configuring Bonding for Maximum Throughput
1749 ==============================================
1751 12.1 Maximizing Throughput in a Single Switch Topology
1752 ------------------------------------------------------
1754 In a single switch configuration, the best method to maximize
1755 throughput depends upon the application and network environment. The
1756 various load balancing modes each have strengths and weaknesses in
1757 different environments, as detailed below.
1759 For this discussion, we will break down the topologies into
1760 two categories. Depending upon the destination of most traffic, we
1761 categorize them into either "gatewayed" or "local" configurations.
1763 In a gatewayed configuration, the "switch" is acting primarily
1764 as a router, and the majority of traffic passes through this router to
1765 other networks. An example would be the following:
1768 +----------+ +----------+
1769 | |eth0 port1| | to other networks
1770 | Host A +---------------------+ router +------------------->
1771 | +---------------------+ | Hosts B and C are out
1772 | |eth1 port2| | here somewhere
1773 +----------+ +----------+
1775 The router may be a dedicated router device, or another host
1776 acting as a gateway. For our discussion, the important point is that
1777 the majority of traffic from Host A will pass through the router to
1778 some other network before reaching its final destination.
1780 In a gatewayed network configuration, although Host A may
1781 communicate with many other systems, all of its traffic will be sent
1782 and received via one other peer on the local network, the router.
1784 Note that the case of two systems connected directly via
1785 multiple physical links is, for purposes of configuring bonding, the
1786 same as a gatewayed configuration. In that case, it happens that all
1787 traffic is destined for the "gateway" itself, not some other network
1790 In a local configuration, the "switch" is acting primarily as
1791 a switch, and the majority of traffic passes through this switch to
1792 reach other stations on the same network. An example would be the
1795 +----------+ +----------+ +--------+
1796 | |eth0 port1| +-------+ Host B |
1797 | Host A +------------+ switch |port3 +--------+
1798 | +------------+ | +--------+
1799 | |eth1 port2| +------------------+ Host C |
1800 +----------+ +----------+port4 +--------+
1803 Again, the switch may be a dedicated switch device, or another
1804 host acting as a gateway. For our discussion, the important point is
1805 that the majority of traffic from Host A is destined for other hosts
1806 on the same local network (Hosts B and C in the above example).
1808 In summary, in a gatewayed configuration, traffic to and from
1809 the bonded device will be to the same MAC level peer on the network
1810 (the gateway itself, i.e., the router), regardless of its final
1811 destination. In a local configuration, traffic flows directly to and
1812 from the final destinations, thus, each destination (Host B, Host C)
1813 will be addressed directly by their individual MAC addresses.
1815 This distinction between a gatewayed and a local network
1816 configuration is important because many of the load balancing modes
1817 available use the MAC addresses of the local network source and
1818 destination to make load balancing decisions. The behavior of each
1819 mode is described below.
1822 12.1.1 MT Bonding Mode Selection for Single Switch Topology
1823 -----------------------------------------------------------
1825 This configuration is the easiest to set up and to understand,
1826 although you will have to decide which bonding mode best suits your
1827 needs. The trade offs for each mode are detailed below:
1829 balance-rr: This mode is the only mode that will permit a single
1830 TCP/IP connection to stripe traffic across multiple
1831 interfaces. It is therefore the only mode that will allow a
1832 single TCP/IP stream to utilize more than one interface's
1833 worth of throughput. This comes at a cost, however: the
1834 striping generally results in peer systems receiving packets out
1835 of order, causing TCP/IP's congestion control system to kick
1836 in, often by retransmitting segments.
1838 It is possible to adjust TCP/IP's congestion limits by
1839 altering the net.ipv4.tcp_reordering sysctl parameter. The
1840 usual default value is 3, and the maximum useful value is 127.
1841 For a four interface balance-rr bond, expect that a single
1842 TCP/IP stream will utilize no more than approximately 2.3
1843 interface's worth of throughput, even after adjusting
1846 Note that the fraction of packets that will be delivered out of
1847 order is highly variable, and is unlikely to be zero. The level
1848 of reordering depends upon a variety of factors, including the
1849 networking interfaces, the switch, and the topology of the
1850 configuration. Speaking in general terms, higher speed network
1851 cards produce more reordering (due to factors such as packet
1852 coalescing), and a "many to many" topology will reorder at a
1853 higher rate than a "many slow to one fast" configuration.
1855 Many switches do not support any modes that stripe traffic
1856 (instead choosing a port based upon IP or MAC level addresses);
1857 for those devices, traffic for a particular connection flowing
1858 through the switch to a balance-rr bond will not utilize greater
1859 than one interface's worth of bandwidth.
1861 If you are utilizing protocols other than TCP/IP, UDP for
1862 example, and your application can tolerate out of order
1863 delivery, then this mode can allow for single stream datagram
1864 performance that scales near linearly as interfaces are added
1867 This mode requires the switch to have the appropriate ports
1868 configured for "etherchannel" or "trunking."
1870 active-backup: There is not much advantage in this network topology to
1871 the active-backup mode, as the inactive backup devices are all
1872 connected to the same peer as the primary. In this case, a
1873 load balancing mode (with link monitoring) will provide the
1874 same level of network availability, but with increased
1875 available bandwidth. On the plus side, active-backup mode
1876 does not require any configuration of the switch, so it may
1877 have value if the hardware available does not support any of
1878 the load balance modes.
1880 balance-xor: This mode will limit traffic such that packets destined
1881 for specific peers will always be sent over the same
1882 interface. Since the destination is determined by the MAC
1883 addresses involved, this mode works best in a "local" network
1884 configuration (as described above), with destinations all on
1885 the same local network. This mode is likely to be suboptimal
1886 if all your traffic is passed through a single router (i.e., a
1887 "gatewayed" network configuration, as described above).
1889 As with balance-rr, the switch ports need to be configured for
1890 "etherchannel" or "trunking."
1892 broadcast: Like active-backup, there is not much advantage to this
1893 mode in this type of network topology.
1895 802.3ad: This mode can be a good choice for this type of network
1896 topology. The 802.3ad mode is an IEEE standard, so all peers
1897 that implement 802.3ad should interoperate well. The 802.3ad
1898 protocol includes automatic configuration of the aggregates,
1899 so minimal manual configuration of the switch is needed
1900 (typically only to designate that some set of devices is
1901 available for 802.3ad). The 802.3ad standard also mandates
1902 that frames be delivered in order (within certain limits), so
1903 in general single connections will not see misordering of
1904 packets. The 802.3ad mode does have some drawbacks: the
1905 standard mandates that all devices in the aggregate operate at
1906 the same speed and duplex. Also, as with all bonding load
1907 balance modes other than balance-rr, no single connection will
1908 be able to utilize more than a single interface's worth of
1911 Additionally, the linux bonding 802.3ad implementation
1912 distributes traffic by peer (using an XOR of MAC addresses),
1913 so in a "gatewayed" configuration, all outgoing traffic will
1914 generally use the same device. Incoming traffic may also end
1915 up on a single device, but that is dependent upon the
1916 balancing policy of the peer's 8023.ad implementation. In a
1917 "local" configuration, traffic will be distributed across the
1918 devices in the bond.
1920 Finally, the 802.3ad mode mandates the use of the MII monitor,
1921 therefore, the ARP monitor is not available in this mode.
1923 balance-tlb: The balance-tlb mode balances outgoing traffic by peer.
1924 Since the balancing is done according to MAC address, in a
1925 "gatewayed" configuration (as described above), this mode will
1926 send all traffic across a single device. However, in a
1927 "local" network configuration, this mode balances multiple
1928 local network peers across devices in a vaguely intelligent
1929 manner (not a simple XOR as in balance-xor or 802.3ad mode),
1930 so that mathematically unlucky MAC addresses (i.e., ones that
1931 XOR to the same value) will not all "bunch up" on a single
1934 Unlike 802.3ad, interfaces may be of differing speeds, and no
1935 special switch configuration is required. On the down side,
1936 in this mode all incoming traffic arrives over a single
1937 interface, this mode requires certain ethtool support in the
1938 network device driver of the slave interfaces, and the ARP
1939 monitor is not available.
1941 balance-alb: This mode is everything that balance-tlb is, and more.
1942 It has all of the features (and restrictions) of balance-tlb,
1943 and will also balance incoming traffic from local network
1944 peers (as described in the Bonding Module Options section,
1947 The only additional down side to this mode is that the network
1948 device driver must support changing the hardware address while
1951 12.1.2 MT Link Monitoring for Single Switch Topology
1952 ----------------------------------------------------
1954 The choice of link monitoring may largely depend upon which
1955 mode you choose to use. The more advanced load balancing modes do not
1956 support the use of the ARP monitor, and are thus restricted to using
1957 the MII monitor (which does not provide as high a level of end to end
1958 assurance as the ARP monitor).
1960 12.2 Maximum Throughput in a Multiple Switch Topology
1961 -----------------------------------------------------
1963 Multiple switches may be utilized to optimize for throughput
1964 when they are configured in parallel as part of an isolated network
1965 between two or more systems, for example:
1971 +--------+ | +---------+
1973 +------+---+ +-----+----+ +-----+----+
1974 | Switch A | | Switch B | | Switch C |
1975 +------+---+ +-----+----+ +-----+----+
1977 +--------+ | +---------+
1983 In this configuration, the switches are isolated from one
1984 another. One reason to employ a topology such as this is for an
1985 isolated network with many hosts (a cluster configured for high
1986 performance, for example), using multiple smaller switches can be more
1987 cost effective than a single larger switch, e.g., on a network with 24
1988 hosts, three 24 port switches can be significantly less expensive than
1989 a single 72 port switch.
1991 If access beyond the network is required, an individual host
1992 can be equipped with an additional network device connected to an
1993 external network; this host then additionally acts as a gateway.
1995 12.2.1 MT Bonding Mode Selection for Multiple Switch Topology
1996 -------------------------------------------------------------
1998 In actual practice, the bonding mode typically employed in
1999 configurations of this type is balance-rr. Historically, in this
2000 network configuration, the usual caveats about out of order packet
2001 delivery are mitigated by the use of network adapters that do not do
2002 any kind of packet coalescing (via the use of NAPI, or because the
2003 device itself does not generate interrupts until some number of
2004 packets has arrived). When employed in this fashion, the balance-rr
2005 mode allows individual connections between two hosts to effectively
2006 utilize greater than one interface's bandwidth.
2008 12.2.2 MT Link Monitoring for Multiple Switch Topology
2009 ------------------------------------------------------
2011 Again, in actual practice, the MII monitor is most often used
2012 in this configuration, as performance is given preference over
2013 availability. The ARP monitor will function in this topology, but its
2014 advantages over the MII monitor are mitigated by the volume of probes
2015 needed as the number of systems involved grows (remember that each
2016 host in the network is configured with bonding).
2018 13. Switch Behavior Issues
2019 ==========================
2021 13.1 Link Establishment and Failover Delays
2022 -------------------------------------------
2024 Some switches exhibit undesirable behavior with regard to the
2025 timing of link up and down reporting by the switch.
2027 First, when a link comes up, some switches may indicate that
2028 the link is up (carrier available), but not pass traffic over the
2029 interface for some period of time. This delay is typically due to
2030 some type of autonegotiation or routing protocol, but may also occur
2031 during switch initialization (e.g., during recovery after a switch
2032 failure). If you find this to be a problem, specify an appropriate
2033 value to the updelay bonding module option to delay the use of the
2034 relevant interface(s).
2036 Second, some switches may "bounce" the link state one or more
2037 times while a link is changing state. This occurs most commonly while
2038 the switch is initializing. Again, an appropriate updelay value may
2041 Note that when a bonding interface has no active links, the
2042 driver will immediately reuse the first link that goes up, even if the
2043 updelay parameter has been specified (the updelay is ignored in this
2044 case). If there are slave interfaces waiting for the updelay timeout
2045 to expire, the interface that first went into that state will be
2046 immediately reused. This reduces down time of the network if the
2047 value of updelay has been overestimated, and since this occurs only in
2048 cases with no connectivity, there is no additional penalty for
2049 ignoring the updelay.
2051 In addition to the concerns about switch timings, if your
2052 switches take a long time to go into backup mode, it may be desirable
2053 to not activate a backup interface immediately after a link goes down.
2054 Failover may be delayed via the downdelay bonding module option.
2056 13.2 Duplicated Incoming Packets
2057 --------------------------------
2059 NOTE: Starting with version 3.0.2, the bonding driver has logic to
2060 suppress duplicate packets, which should largely eliminate this problem.
2061 The following description is kept for reference.
2063 It is not uncommon to observe a short burst of duplicated
2064 traffic when the bonding device is first used, or after it has been
2065 idle for some period of time. This is most easily observed by issuing
2066 a "ping" to some other host on the network, and noticing that the
2067 output from ping flags duplicates (typically one per slave).
2069 For example, on a bond in active-backup mode with five slaves
2070 all connected to one switch, the output may appear as follows:
2073 PING 10.0.4.2 (10.0.4.2) from 10.0.3.10 : 56(84) bytes of data.
2074 64 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.7 ms
2075 64 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.8 ms (DUP!)
2076 64 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.8 ms (DUP!)
2077 64 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.8 ms (DUP!)
2078 64 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.8 ms (DUP!)
2079 64 bytes from 10.0.4.2: icmp_seq=2 ttl=64 time=0.216 ms
2080 64 bytes from 10.0.4.2: icmp_seq=3 ttl=64 time=0.267 ms
2081 64 bytes from 10.0.4.2: icmp_seq=4 ttl=64 time=0.222 ms
2083 This is not due to an error in the bonding driver, rather, it
2084 is a side effect of how many switches update their MAC forwarding
2085 tables. Initially, the switch does not associate the MAC address in
2086 the packet with a particular switch port, and so it may send the
2087 traffic to all ports until its MAC forwarding table is updated. Since
2088 the interfaces attached to the bond may occupy multiple ports on a
2089 single switch, when the switch (temporarily) floods the traffic to all
2090 ports, the bond device receives multiple copies of the same packet
2091 (one per slave device).
2093 The duplicated packet behavior is switch dependent, some
2094 switches exhibit this, and some do not. On switches that display this
2095 behavior, it can be induced by clearing the MAC forwarding table (on
2096 most Cisco switches, the privileged command "clear mac address-table
2097 dynamic" will accomplish this).
2099 14. Hardware Specific Considerations
2100 ====================================
2102 This section contains additional information for configuring
2103 bonding on specific hardware platforms, or for interfacing bonding
2104 with particular switches or other devices.
2106 14.1 IBM BladeCenter
2107 --------------------
2109 This applies to the JS20 and similar systems.
2111 On the JS20 blades, the bonding driver supports only
2112 balance-rr, active-backup, balance-tlb and balance-alb modes. This is
2113 largely due to the network topology inside the BladeCenter, detailed
2116 JS20 network adapter information
2117 --------------------------------
2119 All JS20s come with two Broadcom Gigabit Ethernet ports
2120 integrated on the planar (that's "motherboard" in IBM-speak). In the
2121 BladeCenter chassis, the eth0 port of all JS20 blades is hard wired to
2122 I/O Module #1; similarly, all eth1 ports are wired to I/O Module #2.
2123 An add-on Broadcom daughter card can be installed on a JS20 to provide
2124 two more Gigabit Ethernet ports. These ports, eth2 and eth3, are
2125 wired to I/O Modules 3 and 4, respectively.
2127 Each I/O Module may contain either a switch or a passthrough
2128 module (which allows ports to be directly connected to an external
2129 switch). Some bonding modes require a specific BladeCenter internal
2130 network topology in order to function; these are detailed below.
2132 Additional BladeCenter-specific networking information can be
2133 found in two IBM Redbooks (www.ibm.com/redbooks):
2135 "IBM eServer BladeCenter Networking Options"
2136 "IBM eServer BladeCenter Layer 2-7 Network Switching"
2138 BladeCenter networking configuration
2139 ------------------------------------
2141 Because a BladeCenter can be configured in a very large number
2142 of ways, this discussion will be confined to describing basic
2145 Normally, Ethernet Switch Modules (ESMs) are used in I/O
2146 modules 1 and 2. In this configuration, the eth0 and eth1 ports of a
2147 JS20 will be connected to different internal switches (in the
2148 respective I/O modules).
2150 A passthrough module (OPM or CPM, optical or copper,
2151 passthrough module) connects the I/O module directly to an external
2152 switch. By using PMs in I/O module #1 and #2, the eth0 and eth1
2153 interfaces of a JS20 can be redirected to the outside world and
2154 connected to a common external switch.
2156 Depending upon the mix of ESMs and PMs, the network will
2157 appear to bonding as either a single switch topology (all PMs) or as a
2158 multiple switch topology (one or more ESMs, zero or more PMs). It is
2159 also possible to connect ESMs together, resulting in a configuration
2160 much like the example in "High Availability in a Multiple Switch
2163 Requirements for specific modes
2164 -------------------------------
2166 The balance-rr mode requires the use of passthrough modules
2167 for devices in the bond, all connected to an common external switch.
2168 That switch must be configured for "etherchannel" or "trunking" on the
2169 appropriate ports, as is usual for balance-rr.
2171 The balance-alb and balance-tlb modes will function with
2172 either switch modules or passthrough modules (or a mix). The only
2173 specific requirement for these modes is that all network interfaces
2174 must be able to reach all destinations for traffic sent over the
2175 bonding device (i.e., the network must converge at some point outside
2178 The active-backup mode has no additional requirements.
2180 Link monitoring issues
2181 ----------------------
2183 When an Ethernet Switch Module is in place, only the ARP
2184 monitor will reliably detect link loss to an external switch. This is
2185 nothing unusual, but examination of the BladeCenter cabinet would
2186 suggest that the "external" network ports are the ethernet ports for
2187 the system, when it fact there is a switch between these "external"
2188 ports and the devices on the JS20 system itself. The MII monitor is
2189 only able to detect link failures between the ESM and the JS20 system.
2191 When a passthrough module is in place, the MII monitor does
2192 detect failures to the "external" port, which is then directly
2193 connected to the JS20 system.
2198 The Serial Over LAN (SoL) link is established over the primary
2199 ethernet (eth0) only, therefore, any loss of link to eth0 will result
2200 in losing your SoL connection. It will not fail over with other
2201 network traffic, as the SoL system is beyond the control of the
2204 It may be desirable to disable spanning tree on the switch
2205 (either the internal Ethernet Switch Module, or an external switch) to
2206 avoid fail-over delay issues when using bonding.
2209 15. Frequently Asked Questions
2210 ==============================
2214 Yes. The old 2.0.xx channel bonding patch was not SMP safe.
2215 The new driver was designed to be SMP safe from the start.
2217 2. What type of cards will work with it?
2219 Any Ethernet type cards (you can even mix cards - a Intel
2220 EtherExpress PRO/100 and a 3com 3c905b, for example). For most modes,
2221 devices need not be of the same speed.
2223 Starting with version 3.2.1, bonding also supports Infiniband
2224 slaves in active-backup mode.
2226 3. How many bonding devices can I have?
2230 4. How many slaves can a bonding device have?
2232 This is limited only by the number of network interfaces Linux
2233 supports and/or the number of network cards you can place in your
2236 5. What happens when a slave link dies?
2238 If link monitoring is enabled, then the failing device will be
2239 disabled. The active-backup mode will fail over to a backup link, and
2240 other modes will ignore the failed link. The link will continue to be
2241 monitored, and should it recover, it will rejoin the bond (in whatever
2242 manner is appropriate for the mode). See the sections on High
2243 Availability and the documentation for each mode for additional
2246 Link monitoring can be enabled via either the miimon or
2247 arp_interval parameters (described in the module parameters section,
2248 above). In general, miimon monitors the carrier state as sensed by
2249 the underlying network device, and the arp monitor (arp_interval)
2250 monitors connectivity to another host on the local network.
2252 If no link monitoring is configured, the bonding driver will
2253 be unable to detect link failures, and will assume that all links are
2254 always available. This will likely result in lost packets, and a
2255 resulting degradation of performance. The precise performance loss
2256 depends upon the bonding mode and network configuration.
2258 6. Can bonding be used for High Availability?
2260 Yes. See the section on High Availability for details.
2262 7. Which switches/systems does it work with?
2264 The full answer to this depends upon the desired mode.
2266 In the basic balance modes (balance-rr and balance-xor), it
2267 works with any system that supports etherchannel (also called
2268 trunking). Most managed switches currently available have such
2269 support, and many unmanaged switches as well.
2271 The advanced balance modes (balance-tlb and balance-alb) do
2272 not have special switch requirements, but do need device drivers that
2273 support specific features (described in the appropriate section under
2274 module parameters, above).
2276 In 802.3ad mode, it works with systems that support IEEE
2277 802.3ad Dynamic Link Aggregation. Most managed and many unmanaged
2278 switches currently available support 802.3ad.
2280 The active-backup mode should work with any Layer-II switch.
2282 8. Where does a bonding device get its MAC address from?
2284 When using slave devices that have fixed MAC addresses, or when
2285 the fail_over_mac option is enabled, the bonding device's MAC address is
2286 the MAC address of the active slave.
2288 For other configurations, if not explicitly configured (with
2289 ifconfig or ip link), the MAC address of the bonding device is taken from
2290 its first slave device. This MAC address is then passed to all following
2291 slaves and remains persistent (even if the first slave is removed) until
2292 the bonding device is brought down or reconfigured.
2294 If you wish to change the MAC address, you can set it with
2295 ifconfig or ip link:
2297 # ifconfig bond0 hw ether 00:11:22:33:44:55
2299 # ip link set bond0 address 66:77:88:99:aa:bb
2301 The MAC address can be also changed by bringing down/up the
2302 device and then changing its slaves (or their order):
2304 # ifconfig bond0 down ; modprobe -r bonding
2305 # ifconfig bond0 .... up
2306 # ifenslave bond0 eth...
2308 This method will automatically take the address from the next
2309 slave that is added.
2311 To restore your slaves' MAC addresses, you need to detach them
2312 from the bond (`ifenslave -d bond0 eth0'). The bonding driver will
2313 then restore the MAC addresses that the slaves had before they were
2316 16. Resources and Links
2317 =======================
2319 The latest version of the bonding driver can be found in the latest
2320 version of the linux kernel, found on http://kernel.org
2322 The latest version of this document can be found in either the latest
2323 kernel source (named Documentation/networking/bonding.txt), or on the
2324 bonding sourceforge site:
2326 http://www.sourceforge.net/projects/bonding
2328 Discussions regarding the bonding driver take place primarily on the
2329 bonding-devel mailing list, hosted at sourceforge.net. If you have
2330 questions or problems, post them to the list. The list address is:
2332 bonding-devel@lists.sourceforge.net
2334 The administrative interface (to subscribe or unsubscribe) can
2337 https://lists.sourceforge.net/lists/listinfo/bonding-devel
2339 Donald Becker's Ethernet Drivers and diag programs may be found at :
2340 - http://www.scyld.com/network/
2342 You will also find a lot of information regarding Ethernet, NWay, MII,
2343 etc. at www.scyld.com.