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. Specifying
380 a value of 0 will load bonding, but will not create any devices.
384 Specifies the MII link monitoring frequency in milliseconds.
385 This determines how often the link state of each slave is
386 inspected for link failures. A value of zero disables MII
387 link monitoring. A value of 100 is a good starting point.
388 The use_carrier option, below, affects how the link state is
389 determined. See the High Availability section for additional
390 information. The default value is 0.
394 Specifies one of the bonding policies. The default is
395 balance-rr (round robin). Possible values are:
399 Round-robin policy: Transmit packets in sequential
400 order from the first available slave through the
401 last. This mode provides load balancing and fault
406 Active-backup policy: Only one slave in the bond is
407 active. A different slave becomes active if, and only
408 if, the active slave fails. The bond's MAC address is
409 externally visible on only one port (network adapter)
410 to avoid confusing the switch.
412 In bonding version 2.6.2 or later, when a failover
413 occurs in active-backup mode, bonding will issue one
414 or more gratuitous ARPs on the newly active slave.
415 One gratuitous ARP is issued for the bonding master
416 interface and each VLAN interfaces configured above
417 it, provided that the interface has at least one IP
418 address configured. Gratuitous ARPs issued for VLAN
419 interfaces are tagged with the appropriate VLAN id.
421 This mode provides fault tolerance. The primary
422 option, documented below, affects the behavior of this
427 XOR policy: Transmit based on the selected transmit
428 hash policy. The default policy is a simple [(source
429 MAC address XOR'd with destination MAC address) modulo
430 slave count]. Alternate transmit policies may be
431 selected via the xmit_hash_policy option, described
434 This mode provides load balancing and fault tolerance.
438 Broadcast policy: transmits everything on all slave
439 interfaces. This mode provides fault tolerance.
443 IEEE 802.3ad Dynamic link aggregation. Creates
444 aggregation groups that share the same speed and
445 duplex settings. Utilizes all slaves in the active
446 aggregator according to the 802.3ad specification.
448 Slave selection for outgoing traffic is done according
449 to the transmit hash policy, which may be changed from
450 the default simple XOR policy via the xmit_hash_policy
451 option, documented below. Note that not all transmit
452 policies may be 802.3ad compliant, particularly in
453 regards to the packet mis-ordering requirements of
454 section 43.2.4 of the 802.3ad standard. Differing
455 peer implementations will have varying tolerances for
460 1. Ethtool support in the base drivers for retrieving
461 the speed and duplex of each slave.
463 2. A switch that supports IEEE 802.3ad Dynamic link
466 Most switches will require some type of configuration
467 to enable 802.3ad mode.
471 Adaptive transmit load balancing: channel bonding that
472 does not require any special switch support. The
473 outgoing traffic is distributed according to the
474 current load (computed relative to the speed) on each
475 slave. Incoming traffic is received by the current
476 slave. If the receiving slave fails, another slave
477 takes over the MAC address of the failed receiving
482 Ethtool support in the base drivers for retrieving the
487 Adaptive load balancing: includes balance-tlb plus
488 receive load balancing (rlb) for IPV4 traffic, and
489 does not require any special switch support. The
490 receive load balancing is achieved by ARP negotiation.
491 The bonding driver intercepts the ARP Replies sent by
492 the local system on their way out and overwrites the
493 source hardware address with the unique hardware
494 address of one of the slaves in the bond such that
495 different peers use different hardware addresses for
498 Receive traffic from connections created by the server
499 is also balanced. When the local system sends an ARP
500 Request the bonding driver copies and saves the peer's
501 IP information from the ARP packet. When the ARP
502 Reply arrives from the peer, its hardware address is
503 retrieved and the bonding driver initiates an ARP
504 reply to this peer assigning it to one of the slaves
505 in the bond. A problematic outcome of using ARP
506 negotiation for balancing is that each time that an
507 ARP request is broadcast it uses the hardware address
508 of the bond. Hence, peers learn the hardware address
509 of the bond and the balancing of receive traffic
510 collapses to the current slave. This is handled by
511 sending updates (ARP Replies) to all the peers with
512 their individually assigned hardware address such that
513 the traffic is redistributed. Receive traffic is also
514 redistributed when a new slave is added to the bond
515 and when an inactive slave is re-activated. The
516 receive load is distributed sequentially (round robin)
517 among the group of highest speed slaves in the bond.
519 When a link is reconnected or a new slave joins the
520 bond the receive traffic is redistributed among all
521 active slaves in the bond by initiating ARP Replies
522 with the selected MAC address to each of the
523 clients. The updelay parameter (detailed below) must
524 be set to a value equal or greater than the switch's
525 forwarding delay so that the ARP Replies sent to the
526 peers will not be blocked by the switch.
530 1. Ethtool support in the base drivers for retrieving
531 the speed of each slave.
533 2. Base driver support for setting the hardware
534 address of a device while it is open. This is
535 required so that there will always be one slave in the
536 team using the bond hardware address (the
537 curr_active_slave) while having a unique hardware
538 address for each slave in the bond. If the
539 curr_active_slave fails its hardware address is
540 swapped with the new curr_active_slave that was
545 Specifies the number of gratuitous ARPs to be issued after a
546 failover event. One gratuitous ARP is issued immediately after
547 the failover, subsequent ARPs are sent at a rate of one per link
548 monitor interval (arp_interval or miimon, whichever is active).
550 The valid range is 0 - 255; the default value is 1. This option
551 affects only the active-backup mode. This option was added for
552 bonding version 3.3.0.
556 A string (eth0, eth2, etc) specifying which slave is the
557 primary device. The specified device will always be the
558 active slave while it is available. Only when the primary is
559 off-line will alternate devices be used. This is useful when
560 one slave is preferred over another, e.g., when one slave has
561 higher throughput than another.
563 The primary option is only valid for active-backup mode.
567 Specifies the time, in milliseconds, to wait before enabling a
568 slave after a link recovery has been detected. This option is
569 only valid for the miimon link monitor. The updelay value
570 should be a multiple of the miimon value; if not, it will be
571 rounded down to the nearest multiple. The default value is 0.
575 Specifies whether or not miimon should use MII or ETHTOOL
576 ioctls vs. netif_carrier_ok() to determine the link
577 status. The MII or ETHTOOL ioctls are less efficient and
578 utilize a deprecated calling sequence within the kernel. The
579 netif_carrier_ok() relies on the device driver to maintain its
580 state with netif_carrier_on/off; at this writing, most, but
581 not all, device drivers support this facility.
583 If bonding insists that the link is up when it should not be,
584 it may be that your network device driver does not support
585 netif_carrier_on/off. The default state for netif_carrier is
586 "carrier on," so if a driver does not support netif_carrier,
587 it will appear as if the link is always up. In this case,
588 setting use_carrier to 0 will cause bonding to revert to the
589 MII / ETHTOOL ioctl method to determine the link state.
591 A value of 1 enables the use of netif_carrier_ok(), a value of
592 0 will use the deprecated MII / ETHTOOL ioctls. The default
597 Selects the transmit hash policy to use for slave selection in
598 balance-xor and 802.3ad modes. Possible values are:
602 Uses XOR of hardware MAC addresses to generate the
605 (source MAC XOR destination MAC) modulo slave count
607 This algorithm will place all traffic to a particular
608 network peer on the same slave.
610 This algorithm is 802.3ad compliant.
614 This policy uses a combination of layer2 and layer3
615 protocol information to generate the hash.
617 Uses XOR of hardware MAC addresses and IP addresses to
618 generate the hash. The formula is
620 (((source IP XOR dest IP) AND 0xffff) XOR
621 ( source MAC XOR destination MAC ))
624 This algorithm will place all traffic to a particular
625 network peer on the same slave. For non-IP traffic,
626 the formula is the same as for the layer2 transmit
629 This policy is intended to provide a more balanced
630 distribution of traffic than layer2 alone, especially
631 in environments where a layer3 gateway device is
632 required to reach most destinations.
634 This algorithm is 802.3ad compliant.
638 This policy uses upper layer protocol information,
639 when available, to generate the hash. This allows for
640 traffic to a particular network peer to span multiple
641 slaves, although a single connection will not span
644 The formula for unfragmented TCP and UDP packets is
646 ((source port XOR dest port) XOR
647 ((source IP XOR dest IP) AND 0xffff)
650 For fragmented TCP or UDP packets and all other IP
651 protocol traffic, the source and destination port
652 information is omitted. For non-IP traffic, the
653 formula is the same as for the layer2 transmit hash
656 This policy is intended to mimic the behavior of
657 certain switches, notably Cisco switches with PFC2 as
658 well as some Foundry and IBM products.
660 This algorithm is not fully 802.3ad compliant. A
661 single TCP or UDP conversation containing both
662 fragmented and unfragmented packets will see packets
663 striped across two interfaces. This may result in out
664 of order delivery. Most traffic types will not meet
665 this criteria, as TCP rarely fragments traffic, and
666 most UDP traffic is not involved in extended
667 conversations. Other implementations of 802.3ad may
668 or may not tolerate this noncompliance.
670 The default value is layer2. This option was added in bonding
671 version 2.6.3. In earlier versions of bonding, this parameter
672 does not exist, and the layer2 policy is the only policy. The
673 layer2+3 value was added for bonding version 3.2.2.
676 3. Configuring Bonding Devices
677 ==============================
679 You can configure bonding using either your distro's network
680 initialization scripts, or manually using either ifenslave or the
681 sysfs interface. Distros generally use one of two packages for the
682 network initialization scripts: initscripts or sysconfig. Recent
683 versions of these packages have support for bonding, while older
686 We will first describe the options for configuring bonding for
687 distros using versions of initscripts and sysconfig with full or
688 partial support for bonding, then provide information on enabling
689 bonding without support from the network initialization scripts (i.e.,
690 older versions of initscripts or sysconfig).
692 If you're unsure whether your distro uses sysconfig or
693 initscripts, or don't know if it's new enough, have no fear.
694 Determining this is fairly straightforward.
696 First, issue the command:
700 It will respond with a line of text starting with either
701 "initscripts" or "sysconfig," followed by some numbers. This is the
702 package that provides your network initialization scripts.
704 Next, to determine if your installation supports bonding,
707 $ grep ifenslave /sbin/ifup
709 If this returns any matches, then your initscripts or
710 sysconfig has support for bonding.
712 3.1 Configuration with Sysconfig Support
713 ----------------------------------------
715 This section applies to distros using a version of sysconfig
716 with bonding support, for example, SuSE Linux Enterprise Server 9.
718 SuSE SLES 9's networking configuration system does support
719 bonding, however, at this writing, the YaST system configuration
720 front end does not provide any means to work with bonding devices.
721 Bonding devices can be managed by hand, however, as follows.
723 First, if they have not already been configured, configure the
724 slave devices. On SLES 9, this is most easily done by running the
725 yast2 sysconfig configuration utility. The goal is for to create an
726 ifcfg-id file for each slave device. The simplest way to accomplish
727 this is to configure the devices for DHCP (this is only to get the
728 file ifcfg-id file created; see below for some issues with DHCP). The
729 name of the configuration file for each device will be of the form:
731 ifcfg-id-xx:xx:xx:xx:xx:xx
733 Where the "xx" portion will be replaced with the digits from
734 the device's permanent MAC address.
736 Once the set of ifcfg-id-xx:xx:xx:xx:xx:xx files has been
737 created, it is necessary to edit the configuration files for the slave
738 devices (the MAC addresses correspond to those of the slave devices).
739 Before editing, the file will contain multiple lines, and will look
745 UNIQUE='XNzu.WeZGOGF+4wE'
746 _nm_name='bus-pci-0001:61:01.0'
748 Change the BOOTPROTO and STARTMODE lines to the following:
753 Do not alter the UNIQUE or _nm_name lines. Remove any other
754 lines (USERCTL, etc).
756 Once the ifcfg-id-xx:xx:xx:xx:xx:xx files have been modified,
757 it's time to create the configuration file for the bonding device
758 itself. This file is named ifcfg-bondX, where X is the number of the
759 bonding device to create, starting at 0. The first such file is
760 ifcfg-bond0, the second is ifcfg-bond1, and so on. The sysconfig
761 network configuration system will correctly start multiple instances
764 The contents of the ifcfg-bondX file is as follows:
767 BROADCAST="10.0.2.255"
769 NETMASK="255.255.0.0"
774 BONDING_MODULE_OPTS="mode=active-backup miimon=100"
775 BONDING_SLAVE0="eth0"
776 BONDING_SLAVE1="bus-pci-0000:06:08.1"
778 Replace the sample BROADCAST, IPADDR, NETMASK and NETWORK
779 values with the appropriate values for your network.
781 The STARTMODE specifies when the device is brought online.
782 The possible values are:
784 onboot: The device is started at boot time. If you're not
785 sure, this is probably what you want.
787 manual: The device is started only when ifup is called
788 manually. Bonding devices may be configured this
789 way if you do not wish them to start automatically
790 at boot for some reason.
792 hotplug: The device is started by a hotplug event. This is not
793 a valid choice for a bonding device.
795 off or ignore: The device configuration is ignored.
797 The line BONDING_MASTER='yes' indicates that the device is a
798 bonding master device. The only useful value is "yes."
800 The contents of BONDING_MODULE_OPTS are supplied to the
801 instance of the bonding module for this device. Specify the options
802 for the bonding mode, link monitoring, and so on here. Do not include
803 the max_bonds bonding parameter; this will confuse the configuration
804 system if you have multiple bonding devices.
806 Finally, supply one BONDING_SLAVEn="slave device" for each
807 slave. where "n" is an increasing value, one for each slave. The
808 "slave device" is either an interface name, e.g., "eth0", or a device
809 specifier for the network device. The interface name is easier to
810 find, but the ethN names are subject to change at boot time if, e.g.,
811 a device early in the sequence has failed. The device specifiers
812 (bus-pci-0000:06:08.1 in the example above) specify the physical
813 network device, and will not change unless the device's bus location
814 changes (for example, it is moved from one PCI slot to another). The
815 example above uses one of each type for demonstration purposes; most
816 configurations will choose one or the other for all slave devices.
818 When all configuration files have been modified or created,
819 networking must be restarted for the configuration changes to take
820 effect. This can be accomplished via the following:
822 # /etc/init.d/network restart
824 Note that the network control script (/sbin/ifdown) will
825 remove the bonding module as part of the network shutdown processing,
826 so it is not necessary to remove the module by hand if, e.g., the
827 module parameters have changed.
829 Also, at this writing, YaST/YaST2 will not manage bonding
830 devices (they do not show bonding interfaces on its list of network
831 devices). It is necessary to edit the configuration file by hand to
832 change the bonding configuration.
834 Additional general options and details of the ifcfg file
835 format can be found in an example ifcfg template file:
837 /etc/sysconfig/network/ifcfg.template
839 Note that the template does not document the various BONDING_
840 settings described above, but does describe many of the other options.
842 3.1.1 Using DHCP with Sysconfig
843 -------------------------------
845 Under sysconfig, configuring a device with BOOTPROTO='dhcp'
846 will cause it to query DHCP for its IP address information. At this
847 writing, this does not function for bonding devices; the scripts
848 attempt to obtain the device address from DHCP prior to adding any of
849 the slave devices. Without active slaves, the DHCP requests are not
852 3.1.2 Configuring Multiple Bonds with Sysconfig
853 -----------------------------------------------
855 The sysconfig network initialization system is capable of
856 handling multiple bonding devices. All that is necessary is for each
857 bonding instance to have an appropriately configured ifcfg-bondX file
858 (as described above). Do not specify the "max_bonds" parameter to any
859 instance of bonding, as this will confuse sysconfig. If you require
860 multiple bonding devices with identical parameters, create multiple
863 Because the sysconfig scripts supply the bonding module
864 options in the ifcfg-bondX file, it is not necessary to add them to
865 the system /etc/modules.conf or /etc/modprobe.conf configuration file.
867 3.2 Configuration with Initscripts Support
868 ------------------------------------------
870 This section applies to distros using a recent version of
871 initscripts with bonding support, for example, Red Hat Enterprise Linux
872 version 3 or later, Fedora, etc. On these systems, the network
873 initialization scripts have knowledge of bonding, and can be configured to
874 control bonding devices. Note that older versions of the initscripts
875 package have lower levels of support for bonding; this will be noted where
878 These distros will not automatically load the network adapter
879 driver unless the ethX device is configured with an IP address.
880 Because of this constraint, users must manually configure a
881 network-script file for all physical adapters that will be members of
882 a bondX link. Network script files are located in the directory:
884 /etc/sysconfig/network-scripts
886 The file name must be prefixed with "ifcfg-eth" and suffixed
887 with the adapter's physical adapter number. For example, the script
888 for eth0 would be named /etc/sysconfig/network-scripts/ifcfg-eth0.
889 Place the following text in the file:
898 The DEVICE= line will be different for every ethX device and
899 must correspond with the name of the file, i.e., ifcfg-eth1 must have
900 a device line of DEVICE=eth1. The setting of the MASTER= line will
901 also depend on the final bonding interface name chosen for your bond.
902 As with other network devices, these typically start at 0, and go up
903 one for each device, i.e., the first bonding instance is bond0, the
904 second is bond1, and so on.
906 Next, create a bond network script. The file name for this
907 script will be /etc/sysconfig/network-scripts/ifcfg-bondX where X is
908 the number of the bond. For bond0 the file is named "ifcfg-bond0",
909 for bond1 it is named "ifcfg-bond1", and so on. Within that file,
910 place the following text:
914 NETMASK=255.255.255.0
916 BROADCAST=192.168.1.255
921 Be sure to change the networking specific lines (IPADDR,
922 NETMASK, NETWORK and BROADCAST) to match your network configuration.
924 For later versions of initscripts, such as that found with Fedora
925 7 (or later) and Red Hat Enterprise Linux version 5 (or later), it is possible,
926 and, indeed, preferable, to specify the bonding options in the ifcfg-bond0
927 file, e.g. a line of the format:
929 BONDING_OPTS="mode=active-backup arp_interval=60 arp_ip_target=192.168.1.254"
931 will configure the bond with the specified options. The options
932 specified in BONDING_OPTS are identical to the bonding module parameters
933 except for the arp_ip_target field when using versions of initscripts older
934 than and 8.57 (Fedora 8) and 8.45.19 (Red Hat Enterprise Linux 5.2). When
935 using older versions each target should be included as a separate option and
936 should be preceded by a '+' to indicate it should be added to the list of
937 queried targets, e.g.,
939 arp_ip_target=+192.168.1.1 arp_ip_target=+192.168.1.2
941 is the proper syntax to specify multiple targets. When specifying
942 options via BONDING_OPTS, it is not necessary to edit /etc/modules.conf or
945 For even older versions of initscripts that do not support
946 BONDING_OPTS, it is necessary to edit /etc/modules.conf (or
947 /etc/modprobe.conf, depending upon your distro) to load the bonding module
948 with your desired options when the bond0 interface is brought up. The
949 following lines in /etc/modules.conf (or modprobe.conf) will load the
950 bonding module, and select its options:
953 options bond0 mode=balance-alb miimon=100
955 Replace the sample parameters with the appropriate set of
956 options for your configuration.
958 Finally run "/etc/rc.d/init.d/network restart" as root. This
959 will restart the networking subsystem and your bond link should be now
962 3.2.1 Using DHCP with Initscripts
963 ---------------------------------
965 Recent versions of initscripts (the versions supplied with Fedora
966 Core 3 and Red Hat Enterprise Linux 4, or later versions, are reported to
967 work) have support for assigning IP information to bonding devices via
970 To configure bonding for DHCP, configure it as described
971 above, except replace the line "BOOTPROTO=none" with "BOOTPROTO=dhcp"
972 and add a line consisting of "TYPE=Bonding". Note that the TYPE value
975 3.2.2 Configuring Multiple Bonds with Initscripts
976 -------------------------------------------------
978 Initscripts packages that are included with Fedora 7 and Red Hat
979 Enterprise Linux 5 support multiple bonding interfaces by simply
980 specifying the appropriate BONDING_OPTS= in ifcfg-bondX where X is the
981 number of the bond. This support requires sysfs support in the kernel,
982 and a bonding driver of version 3.0.0 or later. Other configurations may
983 not support this method for specifying multiple bonding interfaces; for
984 those instances, see the "Configuring Multiple Bonds Manually" section,
987 3.3 Configuring Bonding Manually with Ifenslave
988 -----------------------------------------------
990 This section applies to distros whose network initialization
991 scripts (the sysconfig or initscripts package) do not have specific
992 knowledge of bonding. One such distro is SuSE Linux Enterprise Server
995 The general method for these systems is to place the bonding
996 module parameters into /etc/modules.conf or /etc/modprobe.conf (as
997 appropriate for the installed distro), then add modprobe and/or
998 ifenslave commands to the system's global init script. The name of
999 the global init script differs; for sysconfig, it is
1000 /etc/init.d/boot.local and for initscripts it is /etc/rc.d/rc.local.
1002 For example, if you wanted to make a simple bond of two e100
1003 devices (presumed to be eth0 and eth1), and have it persist across
1004 reboots, edit the appropriate file (/etc/init.d/boot.local or
1005 /etc/rc.d/rc.local), and add the following:
1007 modprobe bonding mode=balance-alb miimon=100
1009 ifconfig bond0 192.168.1.1 netmask 255.255.255.0 up
1010 ifenslave bond0 eth0
1011 ifenslave bond0 eth1
1013 Replace the example bonding module parameters and bond0
1014 network configuration (IP address, netmask, etc) with the appropriate
1015 values for your configuration.
1017 Unfortunately, this method will not provide support for the
1018 ifup and ifdown scripts on the bond devices. To reload the bonding
1019 configuration, it is necessary to run the initialization script, e.g.,
1021 # /etc/init.d/boot.local
1025 # /etc/rc.d/rc.local
1027 It may be desirable in such a case to create a separate script
1028 which only initializes the bonding configuration, then call that
1029 separate script from within boot.local. This allows for bonding to be
1030 enabled without re-running the entire global init script.
1032 To shut down the bonding devices, it is necessary to first
1033 mark the bonding device itself as being down, then remove the
1034 appropriate device driver modules. For our example above, you can do
1037 # ifconfig bond0 down
1041 Again, for convenience, it may be desirable to create a script
1042 with these commands.
1045 3.3.1 Configuring Multiple Bonds Manually
1046 -----------------------------------------
1048 This section contains information on configuring multiple
1049 bonding devices with differing options for those systems whose network
1050 initialization scripts lack support for configuring multiple bonds.
1052 If you require multiple bonding devices, but all with the same
1053 options, you may wish to use the "max_bonds" module parameter,
1056 To create multiple bonding devices with differing options, it is
1057 preferrable to use bonding parameters exported by sysfs, documented in the
1060 For versions of bonding without sysfs support, the only means to
1061 provide multiple instances of bonding with differing options is to load
1062 the bonding driver multiple times. Note that current versions of the
1063 sysconfig network initialization scripts handle this automatically; if
1064 your distro uses these scripts, no special action is needed. See the
1065 section Configuring Bonding Devices, above, if you're not sure about your
1066 network initialization scripts.
1068 To load multiple instances of the module, it is necessary to
1069 specify a different name for each instance (the module loading system
1070 requires that every loaded module, even multiple instances of the same
1071 module, have a unique name). This is accomplished by supplying multiple
1072 sets of bonding options in /etc/modprobe.conf, for example:
1075 options bond0 -o bond0 mode=balance-rr miimon=100
1078 options bond1 -o bond1 mode=balance-alb miimon=50
1080 will load the bonding module two times. The first instance is
1081 named "bond0" and creates the bond0 device in balance-rr mode with an
1082 miimon of 100. The second instance is named "bond1" and creates the
1083 bond1 device in balance-alb mode with an miimon of 50.
1085 In some circumstances (typically with older distributions),
1086 the above does not work, and the second bonding instance never sees
1087 its options. In that case, the second options line can be substituted
1090 install bond1 /sbin/modprobe --ignore-install bonding -o bond1 \
1091 mode=balance-alb miimon=50
1093 This may be repeated any number of times, specifying a new and
1094 unique name in place of bond1 for each subsequent instance.
1096 It has been observed that some Red Hat supplied kernels are unable
1097 to rename modules at load time (the "-o bond1" part). Attempts to pass
1098 that option to modprobe will produce an "Operation not permitted" error.
1099 This has been reported on some Fedora Core kernels, and has been seen on
1100 RHEL 4 as well. On kernels exhibiting this problem, it will be impossible
1101 to configure multiple bonds with differing parameters (as they are older
1102 kernels, and also lack sysfs support).
1104 3.4 Configuring Bonding Manually via Sysfs
1105 ------------------------------------------
1107 Starting with version 3.0.0, Channel Bonding may be configured
1108 via the sysfs interface. This interface allows dynamic configuration
1109 of all bonds in the system without unloading the module. It also
1110 allows for adding and removing bonds at runtime. Ifenslave is no
1111 longer required, though it is still supported.
1113 Use of the sysfs interface allows you to use multiple bonds
1114 with different configurations without having to reload the module.
1115 It also allows you to use multiple, differently configured bonds when
1116 bonding is compiled into the kernel.
1118 You must have the sysfs filesystem mounted to configure
1119 bonding this way. The examples in this document assume that you
1120 are using the standard mount point for sysfs, e.g. /sys. If your
1121 sysfs filesystem is mounted elsewhere, you will need to adjust the
1122 example paths accordingly.
1124 Creating and Destroying Bonds
1125 -----------------------------
1126 To add a new bond foo:
1127 # echo +foo > /sys/class/net/bonding_masters
1129 To remove an existing bond bar:
1130 # echo -bar > /sys/class/net/bonding_masters
1132 To show all existing bonds:
1133 # cat /sys/class/net/bonding_masters
1135 NOTE: due to 4K size limitation of sysfs files, this list may be
1136 truncated if you have more than a few hundred bonds. This is unlikely
1137 to occur under normal operating conditions.
1139 Adding and Removing Slaves
1140 --------------------------
1141 Interfaces may be enslaved to a bond using the file
1142 /sys/class/net/<bond>/bonding/slaves. The semantics for this file
1143 are the same as for the bonding_masters file.
1145 To enslave interface eth0 to bond bond0:
1147 # echo +eth0 > /sys/class/net/bond0/bonding/slaves
1149 To free slave eth0 from bond bond0:
1150 # echo -eth0 > /sys/class/net/bond0/bonding/slaves
1152 When an interface is enslaved to a bond, symlinks between the
1153 two are created in the sysfs filesystem. In this case, you would get
1154 /sys/class/net/bond0/slave_eth0 pointing to /sys/class/net/eth0, and
1155 /sys/class/net/eth0/master pointing to /sys/class/net/bond0.
1157 This means that you can tell quickly whether or not an
1158 interface is enslaved by looking for the master symlink. Thus:
1159 # echo -eth0 > /sys/class/net/eth0/master/bonding/slaves
1160 will free eth0 from whatever bond it is enslaved to, regardless of
1161 the name of the bond interface.
1163 Changing a Bond's Configuration
1164 -------------------------------
1165 Each bond may be configured individually by manipulating the
1166 files located in /sys/class/net/<bond name>/bonding
1168 The names of these files correspond directly with the command-
1169 line parameters described elsewhere in this file, and, with the
1170 exception of arp_ip_target, they accept the same values. To see the
1171 current setting, simply cat the appropriate file.
1173 A few examples will be given here; for specific usage
1174 guidelines for each parameter, see the appropriate section in this
1177 To configure bond0 for balance-alb mode:
1178 # ifconfig bond0 down
1179 # echo 6 > /sys/class/net/bond0/bonding/mode
1181 # echo balance-alb > /sys/class/net/bond0/bonding/mode
1182 NOTE: The bond interface must be down before the mode can be
1185 To enable MII monitoring on bond0 with a 1 second interval:
1186 # echo 1000 > /sys/class/net/bond0/bonding/miimon
1187 NOTE: If ARP monitoring is enabled, it will disabled when MII
1188 monitoring is enabled, and vice-versa.
1191 # echo +192.168.0.100 > /sys/class/net/bond0/bonding/arp_ip_target
1192 # echo +192.168.0.101 > /sys/class/net/bond0/bonding/arp_ip_target
1193 NOTE: up to 10 target addresses may be specified.
1195 To remove an ARP target:
1196 # echo -192.168.0.100 > /sys/class/net/bond0/bonding/arp_ip_target
1198 Example Configuration
1199 ---------------------
1200 We begin with the same example that is shown in section 3.3,
1201 executed with sysfs, and without using ifenslave.
1203 To make a simple bond of two e100 devices (presumed to be eth0
1204 and eth1), and have it persist across reboots, edit the appropriate
1205 file (/etc/init.d/boot.local or /etc/rc.d/rc.local), and add the
1210 echo balance-alb > /sys/class/net/bond0/bonding/mode
1211 ifconfig bond0 192.168.1.1 netmask 255.255.255.0 up
1212 echo 100 > /sys/class/net/bond0/bonding/miimon
1213 echo +eth0 > /sys/class/net/bond0/bonding/slaves
1214 echo +eth1 > /sys/class/net/bond0/bonding/slaves
1216 To add a second bond, with two e1000 interfaces in
1217 active-backup mode, using ARP monitoring, add the following lines to
1221 echo +bond1 > /sys/class/net/bonding_masters
1222 echo active-backup > /sys/class/net/bond1/bonding/mode
1223 ifconfig bond1 192.168.2.1 netmask 255.255.255.0 up
1224 echo +192.168.2.100 /sys/class/net/bond1/bonding/arp_ip_target
1225 echo 2000 > /sys/class/net/bond1/bonding/arp_interval
1226 echo +eth2 > /sys/class/net/bond1/bonding/slaves
1227 echo +eth3 > /sys/class/net/bond1/bonding/slaves
1230 4. Querying Bonding Configuration
1231 =================================
1233 4.1 Bonding Configuration
1234 -------------------------
1236 Each bonding device has a read-only file residing in the
1237 /proc/net/bonding directory. The file contents include information
1238 about the bonding configuration, options and state of each slave.
1240 For example, the contents of /proc/net/bonding/bond0 after the
1241 driver is loaded with parameters of mode=0 and miimon=1000 is
1242 generally as follows:
1244 Ethernet Channel Bonding Driver: 2.6.1 (October 29, 2004)
1245 Bonding Mode: load balancing (round-robin)
1246 Currently Active Slave: eth0
1248 MII Polling Interval (ms): 1000
1252 Slave Interface: eth1
1254 Link Failure Count: 1
1256 Slave Interface: eth0
1258 Link Failure Count: 1
1260 The precise format and contents will change depending upon the
1261 bonding configuration, state, and version of the bonding driver.
1263 4.2 Network configuration
1264 -------------------------
1266 The network configuration can be inspected using the ifconfig
1267 command. Bonding devices will have the MASTER flag set; Bonding slave
1268 devices will have the SLAVE flag set. The ifconfig output does not
1269 contain information on which slaves are associated with which masters.
1271 In the example below, the bond0 interface is the master
1272 (MASTER) while eth0 and eth1 are slaves (SLAVE). Notice all slaves of
1273 bond0 have the same MAC address (HWaddr) as bond0 for all modes except
1274 TLB and ALB that require a unique MAC address for each slave.
1277 bond0 Link encap:Ethernet HWaddr 00:C0:F0:1F:37:B4
1278 inet addr:XXX.XXX.XXX.YYY Bcast:XXX.XXX.XXX.255 Mask:255.255.252.0
1279 UP BROADCAST RUNNING MASTER MULTICAST MTU:1500 Metric:1
1280 RX packets:7224794 errors:0 dropped:0 overruns:0 frame:0
1281 TX packets:3286647 errors:1 dropped:0 overruns:1 carrier:0
1282 collisions:0 txqueuelen:0
1284 eth0 Link encap:Ethernet HWaddr 00:C0:F0:1F:37:B4
1285 UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
1286 RX packets:3573025 errors:0 dropped:0 overruns:0 frame:0
1287 TX packets:1643167 errors:1 dropped:0 overruns:1 carrier:0
1288 collisions:0 txqueuelen:100
1289 Interrupt:10 Base address:0x1080
1291 eth1 Link encap:Ethernet HWaddr 00:C0:F0:1F:37:B4
1292 UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
1293 RX packets:3651769 errors:0 dropped:0 overruns:0 frame:0
1294 TX packets:1643480 errors:0 dropped:0 overruns:0 carrier:0
1295 collisions:0 txqueuelen:100
1296 Interrupt:9 Base address:0x1400
1298 5. Switch Configuration
1299 =======================
1301 For this section, "switch" refers to whatever system the
1302 bonded devices are directly connected to (i.e., where the other end of
1303 the cable plugs into). This may be an actual dedicated switch device,
1304 or it may be another regular system (e.g., another computer running
1307 The active-backup, balance-tlb and balance-alb modes do not
1308 require any specific configuration of the switch.
1310 The 802.3ad mode requires that the switch have the appropriate
1311 ports configured as an 802.3ad aggregation. The precise method used
1312 to configure this varies from switch to switch, but, for example, a
1313 Cisco 3550 series switch requires that the appropriate ports first be
1314 grouped together in a single etherchannel instance, then that
1315 etherchannel is set to mode "lacp" to enable 802.3ad (instead of
1316 standard EtherChannel).
1318 The balance-rr, balance-xor and broadcast modes generally
1319 require that the switch have the appropriate ports grouped together.
1320 The nomenclature for such a group differs between switches, it may be
1321 called an "etherchannel" (as in the Cisco example, above), a "trunk
1322 group" or some other similar variation. For these modes, each switch
1323 will also have its own configuration options for the switch's transmit
1324 policy to the bond. Typical choices include XOR of either the MAC or
1325 IP addresses. The transmit policy of the two peers does not need to
1326 match. For these three modes, the bonding mode really selects a
1327 transmit policy for an EtherChannel group; all three will interoperate
1328 with another EtherChannel group.
1331 6. 802.1q VLAN Support
1332 ======================
1334 It is possible to configure VLAN devices over a bond interface
1335 using the 8021q driver. However, only packets coming from the 8021q
1336 driver and passing through bonding will be tagged by default. Self
1337 generated packets, for example, bonding's learning packets or ARP
1338 packets generated by either ALB mode or the ARP monitor mechanism, are
1339 tagged internally by bonding itself. As a result, bonding must
1340 "learn" the VLAN IDs configured above it, and use those IDs to tag
1341 self generated packets.
1343 For reasons of simplicity, and to support the use of adapters
1344 that can do VLAN hardware acceleration offloading, the bonding
1345 interface declares itself as fully hardware offloading capable, it gets
1346 the add_vid/kill_vid notifications to gather the necessary
1347 information, and it propagates those actions to the slaves. In case
1348 of mixed adapter types, hardware accelerated tagged packets that
1349 should go through an adapter that is not offloading capable are
1350 "un-accelerated" by the bonding driver so the VLAN tag sits in the
1353 VLAN interfaces *must* be added on top of a bonding interface
1354 only after enslaving at least one slave. The bonding interface has a
1355 hardware address of 00:00:00:00:00:00 until the first slave is added.
1356 If the VLAN interface is created prior to the first enslavement, it
1357 would pick up the all-zeroes hardware address. Once the first slave
1358 is attached to the bond, the bond device itself will pick up the
1359 slave's hardware address, which is then available for the VLAN device.
1361 Also, be aware that a similar problem can occur if all slaves
1362 are released from a bond that still has one or more VLAN interfaces on
1363 top of it. When a new slave is added, the bonding interface will
1364 obtain its hardware address from the first slave, which might not
1365 match the hardware address of the VLAN interfaces (which was
1366 ultimately copied from an earlier slave).
1368 There are two methods to insure that the VLAN device operates
1369 with the correct hardware address if all slaves are removed from a
1372 1. Remove all VLAN interfaces then recreate them
1374 2. Set the bonding interface's hardware address so that it
1375 matches the hardware address of the VLAN interfaces.
1377 Note that changing a VLAN interface's HW address would set the
1378 underlying device -- i.e. the bonding interface -- to promiscuous
1379 mode, which might not be what you want.
1385 The bonding driver at present supports two schemes for
1386 monitoring a slave device's link state: the ARP monitor and the MII
1389 At the present time, due to implementation restrictions in the
1390 bonding driver itself, it is not possible to enable both ARP and MII
1391 monitoring simultaneously.
1393 7.1 ARP Monitor Operation
1394 -------------------------
1396 The ARP monitor operates as its name suggests: it sends ARP
1397 queries to one or more designated peer systems on the network, and
1398 uses the response as an indication that the link is operating. This
1399 gives some assurance that traffic is actually flowing to and from one
1400 or more peers on the local network.
1402 The ARP monitor relies on the device driver itself to verify
1403 that traffic is flowing. In particular, the driver must keep up to
1404 date the last receive time, dev->last_rx, and transmit start time,
1405 dev->trans_start. If these are not updated by the driver, then the
1406 ARP monitor will immediately fail any slaves using that driver, and
1407 those slaves will stay down. If networking monitoring (tcpdump, etc)
1408 shows the ARP requests and replies on the network, then it may be that
1409 your device driver is not updating last_rx and trans_start.
1411 7.2 Configuring Multiple ARP Targets
1412 ------------------------------------
1414 While ARP monitoring can be done with just one target, it can
1415 be useful in a High Availability setup to have several targets to
1416 monitor. In the case of just one target, the target itself may go
1417 down or have a problem making it unresponsive to ARP requests. Having
1418 an additional target (or several) increases the reliability of the ARP
1421 Multiple ARP targets must be separated by commas as follows:
1423 # example options for ARP monitoring with three targets
1425 options bond0 arp_interval=60 arp_ip_target=192.168.0.1,192.168.0.3,192.168.0.9
1427 For just a single target the options would resemble:
1429 # example options for ARP monitoring with one target
1431 options bond0 arp_interval=60 arp_ip_target=192.168.0.100
1434 7.3 MII Monitor Operation
1435 -------------------------
1437 The MII monitor monitors only the carrier state of the local
1438 network interface. It accomplishes this in one of three ways: by
1439 depending upon the device driver to maintain its carrier state, by
1440 querying the device's MII registers, or by making an ethtool query to
1443 If the use_carrier module parameter is 1 (the default value),
1444 then the MII monitor will rely on the driver for carrier state
1445 information (via the netif_carrier subsystem). As explained in the
1446 use_carrier parameter information, above, if the MII monitor fails to
1447 detect carrier loss on the device (e.g., when the cable is physically
1448 disconnected), it may be that the driver does not support
1451 If use_carrier is 0, then the MII monitor will first query the
1452 device's (via ioctl) MII registers and check the link state. If that
1453 request fails (not just that it returns carrier down), then the MII
1454 monitor will make an ethtool ETHOOL_GLINK request to attempt to obtain
1455 the same information. If both methods fail (i.e., the driver either
1456 does not support or had some error in processing both the MII register
1457 and ethtool requests), then the MII monitor will assume the link is
1460 8. Potential Sources of Trouble
1461 ===============================
1463 8.1 Adventures in Routing
1464 -------------------------
1466 When bonding is configured, it is important that the slave
1467 devices not have routes that supersede routes of the master (or,
1468 generally, not have routes at all). For example, suppose the bonding
1469 device bond0 has two slaves, eth0 and eth1, and the routing table is
1472 Kernel IP routing table
1473 Destination Gateway Genmask Flags MSS Window irtt Iface
1474 10.0.0.0 0.0.0.0 255.255.0.0 U 40 0 0 eth0
1475 10.0.0.0 0.0.0.0 255.255.0.0 U 40 0 0 eth1
1476 10.0.0.0 0.0.0.0 255.255.0.0 U 40 0 0 bond0
1477 127.0.0.0 0.0.0.0 255.0.0.0 U 40 0 0 lo
1479 This routing configuration will likely still update the
1480 receive/transmit times in the driver (needed by the ARP monitor), but
1481 may bypass the bonding driver (because outgoing traffic to, in this
1482 case, another host on network 10 would use eth0 or eth1 before bond0).
1484 The ARP monitor (and ARP itself) may become confused by this
1485 configuration, because ARP requests (generated by the ARP monitor)
1486 will be sent on one interface (bond0), but the corresponding reply
1487 will arrive on a different interface (eth0). This reply looks to ARP
1488 as an unsolicited ARP reply (because ARP matches replies on an
1489 interface basis), and is discarded. The MII monitor is not affected
1490 by the state of the routing table.
1492 The solution here is simply to insure that slaves do not have
1493 routes of their own, and if for some reason they must, those routes do
1494 not supersede routes of their master. This should generally be the
1495 case, but unusual configurations or errant manual or automatic static
1496 route additions may cause trouble.
1498 8.2 Ethernet Device Renaming
1499 ----------------------------
1501 On systems with network configuration scripts that do not
1502 associate physical devices directly with network interface names (so
1503 that the same physical device always has the same "ethX" name), it may
1504 be necessary to add some special logic to either /etc/modules.conf or
1505 /etc/modprobe.conf (depending upon which is installed on the system).
1507 For example, given a modules.conf containing the following:
1510 options bond0 mode=some-mode miimon=50
1516 If neither eth0 and eth1 are slaves to bond0, then when the
1517 bond0 interface comes up, the devices may end up reordered. This
1518 happens because bonding is loaded first, then its slave device's
1519 drivers are loaded next. Since no other drivers have been loaded,
1520 when the e1000 driver loads, it will receive eth0 and eth1 for its
1521 devices, but the bonding configuration tries to enslave eth2 and eth3
1522 (which may later be assigned to the tg3 devices).
1524 Adding the following:
1526 add above bonding e1000 tg3
1528 causes modprobe to load e1000 then tg3, in that order, when
1529 bonding is loaded. This command is fully documented in the
1530 modules.conf manual page.
1532 On systems utilizing modprobe.conf (or modprobe.conf.local),
1533 an equivalent problem can occur. In this case, the following can be
1534 added to modprobe.conf (or modprobe.conf.local, as appropriate), as
1535 follows (all on one line; it has been split here for clarity):
1537 install bonding /sbin/modprobe tg3; /sbin/modprobe e1000;
1538 /sbin/modprobe --ignore-install bonding
1540 This will, when loading the bonding module, rather than
1541 performing the normal action, instead execute the provided command.
1542 This command loads the device drivers in the order needed, then calls
1543 modprobe with --ignore-install to cause the normal action to then take
1544 place. Full documentation on this can be found in the modprobe.conf
1545 and modprobe manual pages.
1547 8.3. Painfully Slow Or No Failed Link Detection By Miimon
1548 ---------------------------------------------------------
1550 By default, bonding enables the use_carrier option, which
1551 instructs bonding to trust the driver to maintain carrier state.
1553 As discussed in the options section, above, some drivers do
1554 not support the netif_carrier_on/_off link state tracking system.
1555 With use_carrier enabled, bonding will always see these links as up,
1556 regardless of their actual state.
1558 Additionally, other drivers do support netif_carrier, but do
1559 not maintain it in real time, e.g., only polling the link state at
1560 some fixed interval. In this case, miimon will detect failures, but
1561 only after some long period of time has expired. If it appears that
1562 miimon is very slow in detecting link failures, try specifying
1563 use_carrier=0 to see if that improves the failure detection time. If
1564 it does, then it may be that the driver checks the carrier state at a
1565 fixed interval, but does not cache the MII register values (so the
1566 use_carrier=0 method of querying the registers directly works). If
1567 use_carrier=0 does not improve the failover, then the driver may cache
1568 the registers, or the problem may be elsewhere.
1570 Also, remember that miimon only checks for the device's
1571 carrier state. It has no way to determine the state of devices on or
1572 beyond other ports of a switch, or if a switch is refusing to pass
1573 traffic while still maintaining carrier on.
1578 If running SNMP agents, the bonding driver should be loaded
1579 before any network drivers participating in a bond. This requirement
1580 is due to the interface index (ipAdEntIfIndex) being associated to
1581 the first interface found with a given IP address. That is, there is
1582 only one ipAdEntIfIndex for each IP address. For example, if eth0 and
1583 eth1 are slaves of bond0 and the driver for eth0 is loaded before the
1584 bonding driver, the interface for the IP address will be associated
1585 with the eth0 interface. This configuration is shown below, the IP
1586 address 192.168.1.1 has an interface index of 2 which indexes to eth0
1587 in the ifDescr table (ifDescr.2).
1589 interfaces.ifTable.ifEntry.ifDescr.1 = lo
1590 interfaces.ifTable.ifEntry.ifDescr.2 = eth0
1591 interfaces.ifTable.ifEntry.ifDescr.3 = eth1
1592 interfaces.ifTable.ifEntry.ifDescr.4 = eth2
1593 interfaces.ifTable.ifEntry.ifDescr.5 = eth3
1594 interfaces.ifTable.ifEntry.ifDescr.6 = bond0
1595 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.10.10.10 = 5
1596 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.192.168.1.1 = 2
1597 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.74.20.94 = 4
1598 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.127.0.0.1 = 1
1600 This problem is avoided by loading the bonding driver before
1601 any network drivers participating in a bond. Below is an example of
1602 loading the bonding driver first, the IP address 192.168.1.1 is
1603 correctly associated with ifDescr.2.
1605 interfaces.ifTable.ifEntry.ifDescr.1 = lo
1606 interfaces.ifTable.ifEntry.ifDescr.2 = bond0
1607 interfaces.ifTable.ifEntry.ifDescr.3 = eth0
1608 interfaces.ifTable.ifEntry.ifDescr.4 = eth1
1609 interfaces.ifTable.ifEntry.ifDescr.5 = eth2
1610 interfaces.ifTable.ifEntry.ifDescr.6 = eth3
1611 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.10.10.10 = 6
1612 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.192.168.1.1 = 2
1613 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.74.20.94 = 5
1614 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.127.0.0.1 = 1
1616 While some distributions may not report the interface name in
1617 ifDescr, the association between the IP address and IfIndex remains
1618 and SNMP functions such as Interface_Scan_Next will report that
1621 10. Promiscuous mode
1622 ====================
1624 When running network monitoring tools, e.g., tcpdump, it is
1625 common to enable promiscuous mode on the device, so that all traffic
1626 is seen (instead of seeing only traffic destined for the local host).
1627 The bonding driver handles promiscuous mode changes to the bonding
1628 master device (e.g., bond0), and propagates the setting to the slave
1631 For the balance-rr, balance-xor, broadcast, and 802.3ad modes,
1632 the promiscuous mode setting is propagated to all slaves.
1634 For the active-backup, balance-tlb and balance-alb modes, the
1635 promiscuous mode setting is propagated only to the active slave.
1637 For balance-tlb mode, the active slave is the slave currently
1638 receiving inbound traffic.
1640 For balance-alb mode, the active slave is the slave used as a
1641 "primary." This slave is used for mode-specific control traffic, for
1642 sending to peers that are unassigned or if the load is unbalanced.
1644 For the active-backup, balance-tlb and balance-alb modes, when
1645 the active slave changes (e.g., due to a link failure), the
1646 promiscuous setting will be propagated to the new active slave.
1648 11. Configuring Bonding for High Availability
1649 =============================================
1651 High Availability refers to configurations that provide
1652 maximum network availability by having redundant or backup devices,
1653 links or switches between the host and the rest of the world. The
1654 goal is to provide the maximum availability of network connectivity
1655 (i.e., the network always works), even though other configurations
1656 could provide higher throughput.
1658 11.1 High Availability in a Single Switch Topology
1659 --------------------------------------------------
1661 If two hosts (or a host and a single switch) are directly
1662 connected via multiple physical links, then there is no availability
1663 penalty to optimizing for maximum bandwidth. In this case, there is
1664 only one switch (or peer), so if it fails, there is no alternative
1665 access to fail over to. Additionally, the bonding load balance modes
1666 support link monitoring of their members, so if individual links fail,
1667 the load will be rebalanced across the remaining devices.
1669 See Section 13, "Configuring Bonding for Maximum Throughput"
1670 for information on configuring bonding with one peer device.
1672 11.2 High Availability in a Multiple Switch Topology
1673 ----------------------------------------------------
1675 With multiple switches, the configuration of bonding and the
1676 network changes dramatically. In multiple switch topologies, there is
1677 a trade off between network availability and usable bandwidth.
1679 Below is a sample network, configured to maximize the
1680 availability of the network:
1684 +-----+----+ +-----+----+
1685 | |port2 ISL port2| |
1686 | switch A +--------------------------+ switch B |
1688 +-----+----+ +-----++---+
1691 +-------------+ host1 +---------------+
1694 In this configuration, there is a link between the two
1695 switches (ISL, or inter switch link), and multiple ports connecting to
1696 the outside world ("port3" on each switch). There is no technical
1697 reason that this could not be extended to a third switch.
1699 11.2.1 HA Bonding Mode Selection for Multiple Switch Topology
1700 -------------------------------------------------------------
1702 In a topology such as the example above, the active-backup and
1703 broadcast modes are the only useful bonding modes when optimizing for
1704 availability; the other modes require all links to terminate on the
1705 same peer for them to behave rationally.
1707 active-backup: This is generally the preferred mode, particularly if
1708 the switches have an ISL and play together well. If the
1709 network configuration is such that one switch is specifically
1710 a backup switch (e.g., has lower capacity, higher cost, etc),
1711 then the primary option can be used to insure that the
1712 preferred link is always used when it is available.
1714 broadcast: This mode is really a special purpose mode, and is suitable
1715 only for very specific needs. For example, if the two
1716 switches are not connected (no ISL), and the networks beyond
1717 them are totally independent. In this case, if it is
1718 necessary for some specific one-way traffic to reach both
1719 independent networks, then the broadcast mode may be suitable.
1721 11.2.2 HA Link Monitoring Selection for Multiple Switch Topology
1722 ----------------------------------------------------------------
1724 The choice of link monitoring ultimately depends upon your
1725 switch. If the switch can reliably fail ports in response to other
1726 failures, then either the MII or ARP monitors should work. For
1727 example, in the above example, if the "port3" link fails at the remote
1728 end, the MII monitor has no direct means to detect this. The ARP
1729 monitor could be configured with a target at the remote end of port3,
1730 thus detecting that failure without switch support.
1732 In general, however, in a multiple switch topology, the ARP
1733 monitor can provide a higher level of reliability in detecting end to
1734 end connectivity failures (which may be caused by the failure of any
1735 individual component to pass traffic for any reason). Additionally,
1736 the ARP monitor should be configured with multiple targets (at least
1737 one for each switch in the network). This will insure that,
1738 regardless of which switch is active, the ARP monitor has a suitable
1741 Note, also, that of late many switches now support a functionality
1742 generally referred to as "trunk failover." This is a feature of the
1743 switch that causes the link state of a particular switch port to be set
1744 down (or up) when the state of another switch port goes down (or up).
1745 It's purpose is to propogate link failures from logically "exterior" ports
1746 to the logically "interior" ports that bonding is able to monitor via
1747 miimon. Availability and configuration for trunk failover varies by
1748 switch, but this can be a viable alternative to the ARP monitor when using
1751 12. Configuring Bonding for Maximum Throughput
1752 ==============================================
1754 12.1 Maximizing Throughput in a Single Switch Topology
1755 ------------------------------------------------------
1757 In a single switch configuration, the best method to maximize
1758 throughput depends upon the application and network environment. The
1759 various load balancing modes each have strengths and weaknesses in
1760 different environments, as detailed below.
1762 For this discussion, we will break down the topologies into
1763 two categories. Depending upon the destination of most traffic, we
1764 categorize them into either "gatewayed" or "local" configurations.
1766 In a gatewayed configuration, the "switch" is acting primarily
1767 as a router, and the majority of traffic passes through this router to
1768 other networks. An example would be the following:
1771 +----------+ +----------+
1772 | |eth0 port1| | to other networks
1773 | Host A +---------------------+ router +------------------->
1774 | +---------------------+ | Hosts B and C are out
1775 | |eth1 port2| | here somewhere
1776 +----------+ +----------+
1778 The router may be a dedicated router device, or another host
1779 acting as a gateway. For our discussion, the important point is that
1780 the majority of traffic from Host A will pass through the router to
1781 some other network before reaching its final destination.
1783 In a gatewayed network configuration, although Host A may
1784 communicate with many other systems, all of its traffic will be sent
1785 and received via one other peer on the local network, the router.
1787 Note that the case of two systems connected directly via
1788 multiple physical links is, for purposes of configuring bonding, the
1789 same as a gatewayed configuration. In that case, it happens that all
1790 traffic is destined for the "gateway" itself, not some other network
1793 In a local configuration, the "switch" is acting primarily as
1794 a switch, and the majority of traffic passes through this switch to
1795 reach other stations on the same network. An example would be the
1798 +----------+ +----------+ +--------+
1799 | |eth0 port1| +-------+ Host B |
1800 | Host A +------------+ switch |port3 +--------+
1801 | +------------+ | +--------+
1802 | |eth1 port2| +------------------+ Host C |
1803 +----------+ +----------+port4 +--------+
1806 Again, the switch may be a dedicated switch device, or another
1807 host acting as a gateway. For our discussion, the important point is
1808 that the majority of traffic from Host A is destined for other hosts
1809 on the same local network (Hosts B and C in the above example).
1811 In summary, in a gatewayed configuration, traffic to and from
1812 the bonded device will be to the same MAC level peer on the network
1813 (the gateway itself, i.e., the router), regardless of its final
1814 destination. In a local configuration, traffic flows directly to and
1815 from the final destinations, thus, each destination (Host B, Host C)
1816 will be addressed directly by their individual MAC addresses.
1818 This distinction between a gatewayed and a local network
1819 configuration is important because many of the load balancing modes
1820 available use the MAC addresses of the local network source and
1821 destination to make load balancing decisions. The behavior of each
1822 mode is described below.
1825 12.1.1 MT Bonding Mode Selection for Single Switch Topology
1826 -----------------------------------------------------------
1828 This configuration is the easiest to set up and to understand,
1829 although you will have to decide which bonding mode best suits your
1830 needs. The trade offs for each mode are detailed below:
1832 balance-rr: This mode is the only mode that will permit a single
1833 TCP/IP connection to stripe traffic across multiple
1834 interfaces. It is therefore the only mode that will allow a
1835 single TCP/IP stream to utilize more than one interface's
1836 worth of throughput. This comes at a cost, however: the
1837 striping generally results in peer systems receiving packets out
1838 of order, causing TCP/IP's congestion control system to kick
1839 in, often by retransmitting segments.
1841 It is possible to adjust TCP/IP's congestion limits by
1842 altering the net.ipv4.tcp_reordering sysctl parameter. The
1843 usual default value is 3, and the maximum useful value is 127.
1844 For a four interface balance-rr bond, expect that a single
1845 TCP/IP stream will utilize no more than approximately 2.3
1846 interface's worth of throughput, even after adjusting
1849 Note that the fraction of packets that will be delivered out of
1850 order is highly variable, and is unlikely to be zero. The level
1851 of reordering depends upon a variety of factors, including the
1852 networking interfaces, the switch, and the topology of the
1853 configuration. Speaking in general terms, higher speed network
1854 cards produce more reordering (due to factors such as packet
1855 coalescing), and a "many to many" topology will reorder at a
1856 higher rate than a "many slow to one fast" configuration.
1858 Many switches do not support any modes that stripe traffic
1859 (instead choosing a port based upon IP or MAC level addresses);
1860 for those devices, traffic for a particular connection flowing
1861 through the switch to a balance-rr bond will not utilize greater
1862 than one interface's worth of bandwidth.
1864 If you are utilizing protocols other than TCP/IP, UDP for
1865 example, and your application can tolerate out of order
1866 delivery, then this mode can allow for single stream datagram
1867 performance that scales near linearly as interfaces are added
1870 This mode requires the switch to have the appropriate ports
1871 configured for "etherchannel" or "trunking."
1873 active-backup: There is not much advantage in this network topology to
1874 the active-backup mode, as the inactive backup devices are all
1875 connected to the same peer as the primary. In this case, a
1876 load balancing mode (with link monitoring) will provide the
1877 same level of network availability, but with increased
1878 available bandwidth. On the plus side, active-backup mode
1879 does not require any configuration of the switch, so it may
1880 have value if the hardware available does not support any of
1881 the load balance modes.
1883 balance-xor: This mode will limit traffic such that packets destined
1884 for specific peers will always be sent over the same
1885 interface. Since the destination is determined by the MAC
1886 addresses involved, this mode works best in a "local" network
1887 configuration (as described above), with destinations all on
1888 the same local network. This mode is likely to be suboptimal
1889 if all your traffic is passed through a single router (i.e., a
1890 "gatewayed" network configuration, as described above).
1892 As with balance-rr, the switch ports need to be configured for
1893 "etherchannel" or "trunking."
1895 broadcast: Like active-backup, there is not much advantage to this
1896 mode in this type of network topology.
1898 802.3ad: This mode can be a good choice for this type of network
1899 topology. The 802.3ad mode is an IEEE standard, so all peers
1900 that implement 802.3ad should interoperate well. The 802.3ad
1901 protocol includes automatic configuration of the aggregates,
1902 so minimal manual configuration of the switch is needed
1903 (typically only to designate that some set of devices is
1904 available for 802.3ad). The 802.3ad standard also mandates
1905 that frames be delivered in order (within certain limits), so
1906 in general single connections will not see misordering of
1907 packets. The 802.3ad mode does have some drawbacks: the
1908 standard mandates that all devices in the aggregate operate at
1909 the same speed and duplex. Also, as with all bonding load
1910 balance modes other than balance-rr, no single connection will
1911 be able to utilize more than a single interface's worth of
1914 Additionally, the linux bonding 802.3ad implementation
1915 distributes traffic by peer (using an XOR of MAC addresses),
1916 so in a "gatewayed" configuration, all outgoing traffic will
1917 generally use the same device. Incoming traffic may also end
1918 up on a single device, but that is dependent upon the
1919 balancing policy of the peer's 8023.ad implementation. In a
1920 "local" configuration, traffic will be distributed across the
1921 devices in the bond.
1923 Finally, the 802.3ad mode mandates the use of the MII monitor,
1924 therefore, the ARP monitor is not available in this mode.
1926 balance-tlb: The balance-tlb mode balances outgoing traffic by peer.
1927 Since the balancing is done according to MAC address, in a
1928 "gatewayed" configuration (as described above), this mode will
1929 send all traffic across a single device. However, in a
1930 "local" network configuration, this mode balances multiple
1931 local network peers across devices in a vaguely intelligent
1932 manner (not a simple XOR as in balance-xor or 802.3ad mode),
1933 so that mathematically unlucky MAC addresses (i.e., ones that
1934 XOR to the same value) will not all "bunch up" on a single
1937 Unlike 802.3ad, interfaces may be of differing speeds, and no
1938 special switch configuration is required. On the down side,
1939 in this mode all incoming traffic arrives over a single
1940 interface, this mode requires certain ethtool support in the
1941 network device driver of the slave interfaces, and the ARP
1942 monitor is not available.
1944 balance-alb: This mode is everything that balance-tlb is, and more.
1945 It has all of the features (and restrictions) of balance-tlb,
1946 and will also balance incoming traffic from local network
1947 peers (as described in the Bonding Module Options section,
1950 The only additional down side to this mode is that the network
1951 device driver must support changing the hardware address while
1954 12.1.2 MT Link Monitoring for Single Switch Topology
1955 ----------------------------------------------------
1957 The choice of link monitoring may largely depend upon which
1958 mode you choose to use. The more advanced load balancing modes do not
1959 support the use of the ARP monitor, and are thus restricted to using
1960 the MII monitor (which does not provide as high a level of end to end
1961 assurance as the ARP monitor).
1963 12.2 Maximum Throughput in a Multiple Switch Topology
1964 -----------------------------------------------------
1966 Multiple switches may be utilized to optimize for throughput
1967 when they are configured in parallel as part of an isolated network
1968 between two or more systems, for example:
1974 +--------+ | +---------+
1976 +------+---+ +-----+----+ +-----+----+
1977 | Switch A | | Switch B | | Switch C |
1978 +------+---+ +-----+----+ +-----+----+
1980 +--------+ | +---------+
1986 In this configuration, the switches are isolated from one
1987 another. One reason to employ a topology such as this is for an
1988 isolated network with many hosts (a cluster configured for high
1989 performance, for example), using multiple smaller switches can be more
1990 cost effective than a single larger switch, e.g., on a network with 24
1991 hosts, three 24 port switches can be significantly less expensive than
1992 a single 72 port switch.
1994 If access beyond the network is required, an individual host
1995 can be equipped with an additional network device connected to an
1996 external network; this host then additionally acts as a gateway.
1998 12.2.1 MT Bonding Mode Selection for Multiple Switch Topology
1999 -------------------------------------------------------------
2001 In actual practice, the bonding mode typically employed in
2002 configurations of this type is balance-rr. Historically, in this
2003 network configuration, the usual caveats about out of order packet
2004 delivery are mitigated by the use of network adapters that do not do
2005 any kind of packet coalescing (via the use of NAPI, or because the
2006 device itself does not generate interrupts until some number of
2007 packets has arrived). When employed in this fashion, the balance-rr
2008 mode allows individual connections between two hosts to effectively
2009 utilize greater than one interface's bandwidth.
2011 12.2.2 MT Link Monitoring for Multiple Switch Topology
2012 ------------------------------------------------------
2014 Again, in actual practice, the MII monitor is most often used
2015 in this configuration, as performance is given preference over
2016 availability. The ARP monitor will function in this topology, but its
2017 advantages over the MII monitor are mitigated by the volume of probes
2018 needed as the number of systems involved grows (remember that each
2019 host in the network is configured with bonding).
2021 13. Switch Behavior Issues
2022 ==========================
2024 13.1 Link Establishment and Failover Delays
2025 -------------------------------------------
2027 Some switches exhibit undesirable behavior with regard to the
2028 timing of link up and down reporting by the switch.
2030 First, when a link comes up, some switches may indicate that
2031 the link is up (carrier available), but not pass traffic over the
2032 interface for some period of time. This delay is typically due to
2033 some type of autonegotiation or routing protocol, but may also occur
2034 during switch initialization (e.g., during recovery after a switch
2035 failure). If you find this to be a problem, specify an appropriate
2036 value to the updelay bonding module option to delay the use of the
2037 relevant interface(s).
2039 Second, some switches may "bounce" the link state one or more
2040 times while a link is changing state. This occurs most commonly while
2041 the switch is initializing. Again, an appropriate updelay value may
2044 Note that when a bonding interface has no active links, the
2045 driver will immediately reuse the first link that goes up, even if the
2046 updelay parameter has been specified (the updelay is ignored in this
2047 case). If there are slave interfaces waiting for the updelay timeout
2048 to expire, the interface that first went into that state will be
2049 immediately reused. This reduces down time of the network if the
2050 value of updelay has been overestimated, and since this occurs only in
2051 cases with no connectivity, there is no additional penalty for
2052 ignoring the updelay.
2054 In addition to the concerns about switch timings, if your
2055 switches take a long time to go into backup mode, it may be desirable
2056 to not activate a backup interface immediately after a link goes down.
2057 Failover may be delayed via the downdelay bonding module option.
2059 13.2 Duplicated Incoming Packets
2060 --------------------------------
2062 NOTE: Starting with version 3.0.2, the bonding driver has logic to
2063 suppress duplicate packets, which should largely eliminate this problem.
2064 The following description is kept for reference.
2066 It is not uncommon to observe a short burst of duplicated
2067 traffic when the bonding device is first used, or after it has been
2068 idle for some period of time. This is most easily observed by issuing
2069 a "ping" to some other host on the network, and noticing that the
2070 output from ping flags duplicates (typically one per slave).
2072 For example, on a bond in active-backup mode with five slaves
2073 all connected to one switch, the output may appear as follows:
2076 PING 10.0.4.2 (10.0.4.2) from 10.0.3.10 : 56(84) bytes of data.
2077 64 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.7 ms
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=1 ttl=64 time=13.8 ms (DUP!)
2080 64 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.8 ms (DUP!)
2081 64 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.8 ms (DUP!)
2082 64 bytes from 10.0.4.2: icmp_seq=2 ttl=64 time=0.216 ms
2083 64 bytes from 10.0.4.2: icmp_seq=3 ttl=64 time=0.267 ms
2084 64 bytes from 10.0.4.2: icmp_seq=4 ttl=64 time=0.222 ms
2086 This is not due to an error in the bonding driver, rather, it
2087 is a side effect of how many switches update their MAC forwarding
2088 tables. Initially, the switch does not associate the MAC address in
2089 the packet with a particular switch port, and so it may send the
2090 traffic to all ports until its MAC forwarding table is updated. Since
2091 the interfaces attached to the bond may occupy multiple ports on a
2092 single switch, when the switch (temporarily) floods the traffic to all
2093 ports, the bond device receives multiple copies of the same packet
2094 (one per slave device).
2096 The duplicated packet behavior is switch dependent, some
2097 switches exhibit this, and some do not. On switches that display this
2098 behavior, it can be induced by clearing the MAC forwarding table (on
2099 most Cisco switches, the privileged command "clear mac address-table
2100 dynamic" will accomplish this).
2102 14. Hardware Specific Considerations
2103 ====================================
2105 This section contains additional information for configuring
2106 bonding on specific hardware platforms, or for interfacing bonding
2107 with particular switches or other devices.
2109 14.1 IBM BladeCenter
2110 --------------------
2112 This applies to the JS20 and similar systems.
2114 On the JS20 blades, the bonding driver supports only
2115 balance-rr, active-backup, balance-tlb and balance-alb modes. This is
2116 largely due to the network topology inside the BladeCenter, detailed
2119 JS20 network adapter information
2120 --------------------------------
2122 All JS20s come with two Broadcom Gigabit Ethernet ports
2123 integrated on the planar (that's "motherboard" in IBM-speak). In the
2124 BladeCenter chassis, the eth0 port of all JS20 blades is hard wired to
2125 I/O Module #1; similarly, all eth1 ports are wired to I/O Module #2.
2126 An add-on Broadcom daughter card can be installed on a JS20 to provide
2127 two more Gigabit Ethernet ports. These ports, eth2 and eth3, are
2128 wired to I/O Modules 3 and 4, respectively.
2130 Each I/O Module may contain either a switch or a passthrough
2131 module (which allows ports to be directly connected to an external
2132 switch). Some bonding modes require a specific BladeCenter internal
2133 network topology in order to function; these are detailed below.
2135 Additional BladeCenter-specific networking information can be
2136 found in two IBM Redbooks (www.ibm.com/redbooks):
2138 "IBM eServer BladeCenter Networking Options"
2139 "IBM eServer BladeCenter Layer 2-7 Network Switching"
2141 BladeCenter networking configuration
2142 ------------------------------------
2144 Because a BladeCenter can be configured in a very large number
2145 of ways, this discussion will be confined to describing basic
2148 Normally, Ethernet Switch Modules (ESMs) are used in I/O
2149 modules 1 and 2. In this configuration, the eth0 and eth1 ports of a
2150 JS20 will be connected to different internal switches (in the
2151 respective I/O modules).
2153 A passthrough module (OPM or CPM, optical or copper,
2154 passthrough module) connects the I/O module directly to an external
2155 switch. By using PMs in I/O module #1 and #2, the eth0 and eth1
2156 interfaces of a JS20 can be redirected to the outside world and
2157 connected to a common external switch.
2159 Depending upon the mix of ESMs and PMs, the network will
2160 appear to bonding as either a single switch topology (all PMs) or as a
2161 multiple switch topology (one or more ESMs, zero or more PMs). It is
2162 also possible to connect ESMs together, resulting in a configuration
2163 much like the example in "High Availability in a Multiple Switch
2166 Requirements for specific modes
2167 -------------------------------
2169 The balance-rr mode requires the use of passthrough modules
2170 for devices in the bond, all connected to an common external switch.
2171 That switch must be configured for "etherchannel" or "trunking" on the
2172 appropriate ports, as is usual for balance-rr.
2174 The balance-alb and balance-tlb modes will function with
2175 either switch modules or passthrough modules (or a mix). The only
2176 specific requirement for these modes is that all network interfaces
2177 must be able to reach all destinations for traffic sent over the
2178 bonding device (i.e., the network must converge at some point outside
2181 The active-backup mode has no additional requirements.
2183 Link monitoring issues
2184 ----------------------
2186 When an Ethernet Switch Module is in place, only the ARP
2187 monitor will reliably detect link loss to an external switch. This is
2188 nothing unusual, but examination of the BladeCenter cabinet would
2189 suggest that the "external" network ports are the ethernet ports for
2190 the system, when it fact there is a switch between these "external"
2191 ports and the devices on the JS20 system itself. The MII monitor is
2192 only able to detect link failures between the ESM and the JS20 system.
2194 When a passthrough module is in place, the MII monitor does
2195 detect failures to the "external" port, which is then directly
2196 connected to the JS20 system.
2201 The Serial Over LAN (SoL) link is established over the primary
2202 ethernet (eth0) only, therefore, any loss of link to eth0 will result
2203 in losing your SoL connection. It will not fail over with other
2204 network traffic, as the SoL system is beyond the control of the
2207 It may be desirable to disable spanning tree on the switch
2208 (either the internal Ethernet Switch Module, or an external switch) to
2209 avoid fail-over delay issues when using bonding.
2212 15. Frequently Asked Questions
2213 ==============================
2217 Yes. The old 2.0.xx channel bonding patch was not SMP safe.
2218 The new driver was designed to be SMP safe from the start.
2220 2. What type of cards will work with it?
2222 Any Ethernet type cards (you can even mix cards - a Intel
2223 EtherExpress PRO/100 and a 3com 3c905b, for example). For most modes,
2224 devices need not be of the same speed.
2226 Starting with version 3.2.1, bonding also supports Infiniband
2227 slaves in active-backup mode.
2229 3. How many bonding devices can I have?
2233 4. How many slaves can a bonding device have?
2235 This is limited only by the number of network interfaces Linux
2236 supports and/or the number of network cards you can place in your
2239 5. What happens when a slave link dies?
2241 If link monitoring is enabled, then the failing device will be
2242 disabled. The active-backup mode will fail over to a backup link, and
2243 other modes will ignore the failed link. The link will continue to be
2244 monitored, and should it recover, it will rejoin the bond (in whatever
2245 manner is appropriate for the mode). See the sections on High
2246 Availability and the documentation for each mode for additional
2249 Link monitoring can be enabled via either the miimon or
2250 arp_interval parameters (described in the module parameters section,
2251 above). In general, miimon monitors the carrier state as sensed by
2252 the underlying network device, and the arp monitor (arp_interval)
2253 monitors connectivity to another host on the local network.
2255 If no link monitoring is configured, the bonding driver will
2256 be unable to detect link failures, and will assume that all links are
2257 always available. This will likely result in lost packets, and a
2258 resulting degradation of performance. The precise performance loss
2259 depends upon the bonding mode and network configuration.
2261 6. Can bonding be used for High Availability?
2263 Yes. See the section on High Availability for details.
2265 7. Which switches/systems does it work with?
2267 The full answer to this depends upon the desired mode.
2269 In the basic balance modes (balance-rr and balance-xor), it
2270 works with any system that supports etherchannel (also called
2271 trunking). Most managed switches currently available have such
2272 support, and many unmanaged switches as well.
2274 The advanced balance modes (balance-tlb and balance-alb) do
2275 not have special switch requirements, but do need device drivers that
2276 support specific features (described in the appropriate section under
2277 module parameters, above).
2279 In 802.3ad mode, it works with systems that support IEEE
2280 802.3ad Dynamic Link Aggregation. Most managed and many unmanaged
2281 switches currently available support 802.3ad.
2283 The active-backup mode should work with any Layer-II switch.
2285 8. Where does a bonding device get its MAC address from?
2287 When using slave devices that have fixed MAC addresses, or when
2288 the fail_over_mac option is enabled, the bonding device's MAC address is
2289 the MAC address of the active slave.
2291 For other configurations, if not explicitly configured (with
2292 ifconfig or ip link), the MAC address of the bonding device is taken from
2293 its first slave device. This MAC address is then passed to all following
2294 slaves and remains persistent (even if the first slave is removed) until
2295 the bonding device is brought down or reconfigured.
2297 If you wish to change the MAC address, you can set it with
2298 ifconfig or ip link:
2300 # ifconfig bond0 hw ether 00:11:22:33:44:55
2302 # ip link set bond0 address 66:77:88:99:aa:bb
2304 The MAC address can be also changed by bringing down/up the
2305 device and then changing its slaves (or their order):
2307 # ifconfig bond0 down ; modprobe -r bonding
2308 # ifconfig bond0 .... up
2309 # ifenslave bond0 eth...
2311 This method will automatically take the address from the next
2312 slave that is added.
2314 To restore your slaves' MAC addresses, you need to detach them
2315 from the bond (`ifenslave -d bond0 eth0'). The bonding driver will
2316 then restore the MAC addresses that the slaves had before they were
2319 16. Resources and Links
2320 =======================
2322 The latest version of the bonding driver can be found in the latest
2323 version of the linux kernel, found on http://kernel.org
2325 The latest version of this document can be found in either the latest
2326 kernel source (named Documentation/networking/bonding.txt), or on the
2327 bonding sourceforge site:
2329 http://www.sourceforge.net/projects/bonding
2331 Discussions regarding the bonding driver take place primarily on the
2332 bonding-devel mailing list, hosted at sourceforge.net. If you have
2333 questions or problems, post them to the list. The list address is:
2335 bonding-devel@lists.sourceforge.net
2337 The administrative interface (to subscribe or unsubscribe) can
2340 https://lists.sourceforge.net/lists/listinfo/bonding-devel
2342 Donald Becker's Ethernet Drivers and diag programs may be found at :
2343 - http://www.scyld.com/network/
2345 You will also find a lot of information regarding Ethernet, NWay, MII,
2346 etc. at www.scyld.com.