2 Linux Ethernet Bonding Driver HOWTO
4 Latest update: 24 April 2006
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
170 the bonding module at load time. They may be given as command line
171 arguments to the insmod or modprobe command, but are usually specified
172 in either the /etc/modules.conf or /etc/modprobe.conf configuration
173 file, or in a distro-specific configuration file (some of which are
174 detailed in the next section).
176 The available bonding driver parameters are listed below. If a
177 parameter is not specified the default value is used. When initially
178 configuring a bond, it is recommended "tail -f /var/log/messages" be
179 run in a separate window to watch for bonding driver error messages.
181 It is critical that either the miimon or arp_interval and
182 arp_ip_target parameters be specified, otherwise serious network
183 degradation will occur during link failures. Very few devices do not
184 support at least miimon, so there is really no reason not to use it.
186 Options with textual values will accept either the text name
187 or, for backwards compatibility, the option value. E.g.,
188 "mode=802.3ad" and "mode=4" set the same mode.
190 The parameters are as follows:
194 Specifies the ARP link monitoring frequency in milliseconds.
196 The ARP monitor works by periodically checking the slave
197 devices to determine whether they have sent or received
198 traffic recently (the precise criteria depends upon the
199 bonding mode, and the state of the slave). Regular traffic is
200 generated via ARP probes issued for the addresses specified by
201 the arp_ip_target option.
203 This behavior can be modified by the arp_validate option,
206 If ARP monitoring is used in an etherchannel compatible mode
207 (modes 0 and 2), the switch should be configured in a mode
208 that evenly distributes packets across all links. If the
209 switch is configured to distribute the packets in an XOR
210 fashion, all replies from the ARP targets will be received on
211 the same link which could cause the other team members to
212 fail. ARP monitoring should not be used in conjunction with
213 miimon. A value of 0 disables ARP monitoring. The default
218 Specifies the IP addresses to use as ARP monitoring peers when
219 arp_interval is > 0. These are the targets of the ARP request
220 sent to determine the health of the link to the targets.
221 Specify these values in ddd.ddd.ddd.ddd format. Multiple IP
222 addresses must be separated by a comma. At least one IP
223 address must be given for ARP monitoring to function. The
224 maximum number of targets that can be specified is 16. The
225 default value is no IP addresses.
229 Specifies whether or not ARP probes and replies should be
230 validated in the active-backup mode. This causes the ARP
231 monitor to examine the incoming ARP requests and replies, and
232 only consider a slave to be up if it is receiving the
233 appropriate ARP traffic.
239 No validation is performed. This is the default.
243 Validation is performed only for the active slave.
247 Validation is performed only for backup slaves.
251 Validation is performed for all slaves.
253 For the active slave, the validation checks ARP replies to
254 confirm that they were generated by an arp_ip_target. Since
255 backup slaves do not typically receive these replies, the
256 validation performed for backup slaves is on the ARP request
257 sent out via the active slave. It is possible that some
258 switch or network configurations may result in situations
259 wherein the backup slaves do not receive the ARP requests; in
260 such a situation, validation of backup slaves must be
263 This option is useful in network configurations in which
264 multiple bonding hosts are concurrently issuing ARPs to one or
265 more targets beyond a common switch. Should the link between
266 the switch and target fail (but not the switch itself), the
267 probe traffic generated by the multiple bonding instances will
268 fool the standard ARP monitor into considering the links as
269 still up. Use of the arp_validate option can resolve this, as
270 the ARP monitor will only consider ARP requests and replies
271 associated with its own instance of bonding.
273 This option was added in bonding version 3.1.0.
277 Specifies the time, in milliseconds, to wait before disabling
278 a slave after a link failure has been detected. This option
279 is only valid for the miimon link monitor. The downdelay
280 value should be a multiple of the miimon value; if not, it
281 will be rounded down to the nearest multiple. The default
286 Specifies whether active-backup mode should set all slaves to
287 the same MAC address (the traditional behavior), or, when
288 enabled, change the bond's MAC address when changing the
289 active interface (i.e., fail over the MAC address itself).
291 Fail over MAC is useful for devices that cannot ever alter
292 their MAC address, or for devices that refuse incoming
293 broadcasts with their own source MAC (which interferes with
296 The down side of fail over MAC is that every device on the
297 network must be updated via gratuitous ARP, vs. just updating
298 a switch or set of switches (which often takes place for any
299 traffic, not just ARP traffic, if the switch snoops incoming
300 traffic to update its tables) for the traditional method. If
301 the gratuitous ARP is lost, communication may be disrupted.
303 When fail over MAC is used in conjuction with the mii monitor,
304 devices which assert link up prior to being able to actually
305 transmit and receive are particularly susecptible to loss of
306 the gratuitous ARP, and an appropriate updelay setting may be
309 A value of 0 disables fail over MAC, and is the default. A
310 value of 1 enables fail over MAC. This option is enabled
311 automatically if the first slave added cannot change its MAC
312 address. This option may be modified via sysfs only when no
313 slaves are present in the bond.
315 This option was added in bonding version 3.2.0.
319 Option specifying the rate in which we'll ask our link partner
320 to transmit LACPDU packets in 802.3ad mode. Possible values
324 Request partner to transmit LACPDUs every 30 seconds
327 Request partner to transmit LACPDUs every 1 second
333 Specifies the number of bonding devices to create for this
334 instance of the bonding driver. E.g., if max_bonds is 3, and
335 the bonding driver is not already loaded, then bond0, bond1
336 and bond2 will be created. The default value is 1.
340 Specifies the MII link monitoring frequency in milliseconds.
341 This determines how often the link state of each slave is
342 inspected for link failures. A value of zero disables MII
343 link monitoring. A value of 100 is a good starting point.
344 The use_carrier option, below, affects how the link state is
345 determined. See the High Availability section for additional
346 information. The default value is 0.
350 Specifies one of the bonding policies. The default is
351 balance-rr (round robin). Possible values are:
355 Round-robin policy: Transmit packets in sequential
356 order from the first available slave through the
357 last. This mode provides load balancing and fault
362 Active-backup policy: Only one slave in the bond is
363 active. A different slave becomes active if, and only
364 if, the active slave fails. The bond's MAC address is
365 externally visible on only one port (network adapter)
366 to avoid confusing the switch.
368 In bonding version 2.6.2 or later, when a failover
369 occurs in active-backup mode, bonding will issue one
370 or more gratuitous ARPs on the newly active slave.
371 One gratuitous ARP is issued for the bonding master
372 interface and each VLAN interfaces configured above
373 it, provided that the interface has at least one IP
374 address configured. Gratuitous ARPs issued for VLAN
375 interfaces are tagged with the appropriate VLAN id.
377 This mode provides fault tolerance. The primary
378 option, documented below, affects the behavior of this
383 XOR policy: Transmit based on the selected transmit
384 hash policy. The default policy is a simple [(source
385 MAC address XOR'd with destination MAC address) modulo
386 slave count]. Alternate transmit policies may be
387 selected via the xmit_hash_policy option, described
390 This mode provides load balancing and fault tolerance.
394 Broadcast policy: transmits everything on all slave
395 interfaces. This mode provides fault tolerance.
399 IEEE 802.3ad Dynamic link aggregation. Creates
400 aggregation groups that share the same speed and
401 duplex settings. Utilizes all slaves in the active
402 aggregator according to the 802.3ad specification.
404 Slave selection for outgoing traffic is done according
405 to the transmit hash policy, which may be changed from
406 the default simple XOR policy via the xmit_hash_policy
407 option, documented below. Note that not all transmit
408 policies may be 802.3ad compliant, particularly in
409 regards to the packet mis-ordering requirements of
410 section 43.2.4 of the 802.3ad standard. Differing
411 peer implementations will have varying tolerances for
416 1. Ethtool support in the base drivers for retrieving
417 the speed and duplex of each slave.
419 2. A switch that supports IEEE 802.3ad Dynamic link
422 Most switches will require some type of configuration
423 to enable 802.3ad mode.
427 Adaptive transmit load balancing: channel bonding that
428 does not require any special switch support. The
429 outgoing traffic is distributed according to the
430 current load (computed relative to the speed) on each
431 slave. Incoming traffic is received by the current
432 slave. If the receiving slave fails, another slave
433 takes over the MAC address of the failed receiving
438 Ethtool support in the base drivers for retrieving the
443 Adaptive load balancing: includes balance-tlb plus
444 receive load balancing (rlb) for IPV4 traffic, and
445 does not require any special switch support. The
446 receive load balancing is achieved by ARP negotiation.
447 The bonding driver intercepts the ARP Replies sent by
448 the local system on their way out and overwrites the
449 source hardware address with the unique hardware
450 address of one of the slaves in the bond such that
451 different peers use different hardware addresses for
454 Receive traffic from connections created by the server
455 is also balanced. When the local system sends an ARP
456 Request the bonding driver copies and saves the peer's
457 IP information from the ARP packet. When the ARP
458 Reply arrives from the peer, its hardware address is
459 retrieved and the bonding driver initiates an ARP
460 reply to this peer assigning it to one of the slaves
461 in the bond. A problematic outcome of using ARP
462 negotiation for balancing is that each time that an
463 ARP request is broadcast it uses the hardware address
464 of the bond. Hence, peers learn the hardware address
465 of the bond and the balancing of receive traffic
466 collapses to the current slave. This is handled by
467 sending updates (ARP Replies) to all the peers with
468 their individually assigned hardware address such that
469 the traffic is redistributed. Receive traffic is also
470 redistributed when a new slave is added to the bond
471 and when an inactive slave is re-activated. The
472 receive load is distributed sequentially (round robin)
473 among the group of highest speed slaves in the bond.
475 When a link is reconnected or a new slave joins the
476 bond the receive traffic is redistributed among all
477 active slaves in the bond by initiating ARP Replies
478 with the selected MAC address to each of the
479 clients. The updelay parameter (detailed below) must
480 be set to a value equal or greater than the switch's
481 forwarding delay so that the ARP Replies sent to the
482 peers will not be blocked by the switch.
486 1. Ethtool support in the base drivers for retrieving
487 the speed of each slave.
489 2. Base driver support for setting the hardware
490 address of a device while it is open. This is
491 required so that there will always be one slave in the
492 team using the bond hardware address (the
493 curr_active_slave) while having a unique hardware
494 address for each slave in the bond. If the
495 curr_active_slave fails its hardware address is
496 swapped with the new curr_active_slave that was
501 A string (eth0, eth2, etc) specifying which slave is the
502 primary device. The specified device will always be the
503 active slave while it is available. Only when the primary is
504 off-line will alternate devices be used. This is useful when
505 one slave is preferred over another, e.g., when one slave has
506 higher throughput than another.
508 The primary option is only valid for active-backup mode.
512 Specifies the time, in milliseconds, to wait before enabling a
513 slave after a link recovery has been detected. This option is
514 only valid for the miimon link monitor. The updelay value
515 should be a multiple of the miimon value; if not, it will be
516 rounded down to the nearest multiple. The default value is 0.
520 Specifies whether or not miimon should use MII or ETHTOOL
521 ioctls vs. netif_carrier_ok() to determine the link
522 status. The MII or ETHTOOL ioctls are less efficient and
523 utilize a deprecated calling sequence within the kernel. The
524 netif_carrier_ok() relies on the device driver to maintain its
525 state with netif_carrier_on/off; at this writing, most, but
526 not all, device drivers support this facility.
528 If bonding insists that the link is up when it should not be,
529 it may be that your network device driver does not support
530 netif_carrier_on/off. The default state for netif_carrier is
531 "carrier on," so if a driver does not support netif_carrier,
532 it will appear as if the link is always up. In this case,
533 setting use_carrier to 0 will cause bonding to revert to the
534 MII / ETHTOOL ioctl method to determine the link state.
536 A value of 1 enables the use of netif_carrier_ok(), a value of
537 0 will use the deprecated MII / ETHTOOL ioctls. The default
542 Selects the transmit hash policy to use for slave selection in
543 balance-xor and 802.3ad modes. Possible values are:
547 Uses XOR of hardware MAC addresses to generate the
550 (source MAC XOR destination MAC) modulo slave count
552 This algorithm will place all traffic to a particular
553 network peer on the same slave.
555 This algorithm is 802.3ad compliant.
559 This policy uses upper layer protocol information,
560 when available, to generate the hash. This allows for
561 traffic to a particular network peer to span multiple
562 slaves, although a single connection will not span
565 The formula for unfragmented TCP and UDP packets is
567 ((source port XOR dest port) XOR
568 ((source IP XOR dest IP) AND 0xffff)
571 For fragmented TCP or UDP packets and all other IP
572 protocol traffic, the source and destination port
573 information is omitted. For non-IP traffic, the
574 formula is the same as for the layer2 transmit hash
577 This policy is intended to mimic the behavior of
578 certain switches, notably Cisco switches with PFC2 as
579 well as some Foundry and IBM products.
581 This algorithm is not fully 802.3ad compliant. A
582 single TCP or UDP conversation containing both
583 fragmented and unfragmented packets will see packets
584 striped across two interfaces. This may result in out
585 of order delivery. Most traffic types will not meet
586 this criteria, as TCP rarely fragments traffic, and
587 most UDP traffic is not involved in extended
588 conversations. Other implementations of 802.3ad may
589 or may not tolerate this noncompliance.
591 The default value is layer2. This option was added in bonding
592 version 2.6.3. In earlier versions of bonding, this parameter does
593 not exist, and the layer2 policy is the only policy.
596 3. Configuring Bonding Devices
597 ==============================
599 You can configure bonding using either your distro's network
600 initialization scripts, or manually using either ifenslave or the
601 sysfs interface. Distros generally use one of two packages for the
602 network initialization scripts: initscripts or sysconfig. Recent
603 versions of these packages have support for bonding, while older
606 We will first describe the options for configuring bonding for
607 distros using versions of initscripts and sysconfig with full or
608 partial support for bonding, then provide information on enabling
609 bonding without support from the network initialization scripts (i.e.,
610 older versions of initscripts or sysconfig).
612 If you're unsure whether your distro uses sysconfig or
613 initscripts, or don't know if it's new enough, have no fear.
614 Determining this is fairly straightforward.
616 First, issue the command:
620 It will respond with a line of text starting with either
621 "initscripts" or "sysconfig," followed by some numbers. This is the
622 package that provides your network initialization scripts.
624 Next, to determine if your installation supports bonding,
627 $ grep ifenslave /sbin/ifup
629 If this returns any matches, then your initscripts or
630 sysconfig has support for bonding.
632 3.1 Configuration with Sysconfig Support
633 ----------------------------------------
635 This section applies to distros using a version of sysconfig
636 with bonding support, for example, SuSE Linux Enterprise Server 9.
638 SuSE SLES 9's networking configuration system does support
639 bonding, however, at this writing, the YaST system configuration
640 front end does not provide any means to work with bonding devices.
641 Bonding devices can be managed by hand, however, as follows.
643 First, if they have not already been configured, configure the
644 slave devices. On SLES 9, this is most easily done by running the
645 yast2 sysconfig configuration utility. The goal is for to create an
646 ifcfg-id file for each slave device. The simplest way to accomplish
647 this is to configure the devices for DHCP (this is only to get the
648 file ifcfg-id file created; see below for some issues with DHCP). The
649 name of the configuration file for each device will be of the form:
651 ifcfg-id-xx:xx:xx:xx:xx:xx
653 Where the "xx" portion will be replaced with the digits from
654 the device's permanent MAC address.
656 Once the set of ifcfg-id-xx:xx:xx:xx:xx:xx files has been
657 created, it is necessary to edit the configuration files for the slave
658 devices (the MAC addresses correspond to those of the slave devices).
659 Before editing, the file will contain multiple lines, and will look
665 UNIQUE='XNzu.WeZGOGF+4wE'
666 _nm_name='bus-pci-0001:61:01.0'
668 Change the BOOTPROTO and STARTMODE lines to the following:
673 Do not alter the UNIQUE or _nm_name lines. Remove any other
674 lines (USERCTL, etc).
676 Once the ifcfg-id-xx:xx:xx:xx:xx:xx files have been modified,
677 it's time to create the configuration file for the bonding device
678 itself. This file is named ifcfg-bondX, where X is the number of the
679 bonding device to create, starting at 0. The first such file is
680 ifcfg-bond0, the second is ifcfg-bond1, and so on. The sysconfig
681 network configuration system will correctly start multiple instances
684 The contents of the ifcfg-bondX file is as follows:
687 BROADCAST="10.0.2.255"
689 NETMASK="255.255.0.0"
694 BONDING_MODULE_OPTS="mode=active-backup miimon=100"
695 BONDING_SLAVE0="eth0"
696 BONDING_SLAVE1="bus-pci-0000:06:08.1"
698 Replace the sample BROADCAST, IPADDR, NETMASK and NETWORK
699 values with the appropriate values for your network.
701 The STARTMODE specifies when the device is brought online.
702 The possible values are:
704 onboot: The device is started at boot time. If you're not
705 sure, this is probably what you want.
707 manual: The device is started only when ifup is called
708 manually. Bonding devices may be configured this
709 way if you do not wish them to start automatically
710 at boot for some reason.
712 hotplug: The device is started by a hotplug event. This is not
713 a valid choice for a bonding device.
715 off or ignore: The device configuration is ignored.
717 The line BONDING_MASTER='yes' indicates that the device is a
718 bonding master device. The only useful value is "yes."
720 The contents of BONDING_MODULE_OPTS are supplied to the
721 instance of the bonding module for this device. Specify the options
722 for the bonding mode, link monitoring, and so on here. Do not include
723 the max_bonds bonding parameter; this will confuse the configuration
724 system if you have multiple bonding devices.
726 Finally, supply one BONDING_SLAVEn="slave device" for each
727 slave. where "n" is an increasing value, one for each slave. The
728 "slave device" is either an interface name, e.g., "eth0", or a device
729 specifier for the network device. The interface name is easier to
730 find, but the ethN names are subject to change at boot time if, e.g.,
731 a device early in the sequence has failed. The device specifiers
732 (bus-pci-0000:06:08.1 in the example above) specify the physical
733 network device, and will not change unless the device's bus location
734 changes (for example, it is moved from one PCI slot to another). The
735 example above uses one of each type for demonstration purposes; most
736 configurations will choose one or the other for all slave devices.
738 When all configuration files have been modified or created,
739 networking must be restarted for the configuration changes to take
740 effect. This can be accomplished via the following:
742 # /etc/init.d/network restart
744 Note that the network control script (/sbin/ifdown) will
745 remove the bonding module as part of the network shutdown processing,
746 so it is not necessary to remove the module by hand if, e.g., the
747 module parameters have changed.
749 Also, at this writing, YaST/YaST2 will not manage bonding
750 devices (they do not show bonding interfaces on its list of network
751 devices). It is necessary to edit the configuration file by hand to
752 change the bonding configuration.
754 Additional general options and details of the ifcfg file
755 format can be found in an example ifcfg template file:
757 /etc/sysconfig/network/ifcfg.template
759 Note that the template does not document the various BONDING_
760 settings described above, but does describe many of the other options.
762 3.1.1 Using DHCP with Sysconfig
763 -------------------------------
765 Under sysconfig, configuring a device with BOOTPROTO='dhcp'
766 will cause it to query DHCP for its IP address information. At this
767 writing, this does not function for bonding devices; the scripts
768 attempt to obtain the device address from DHCP prior to adding any of
769 the slave devices. Without active slaves, the DHCP requests are not
772 3.1.2 Configuring Multiple Bonds with Sysconfig
773 -----------------------------------------------
775 The sysconfig network initialization system is capable of
776 handling multiple bonding devices. All that is necessary is for each
777 bonding instance to have an appropriately configured ifcfg-bondX file
778 (as described above). Do not specify the "max_bonds" parameter to any
779 instance of bonding, as this will confuse sysconfig. If you require
780 multiple bonding devices with identical parameters, create multiple
783 Because the sysconfig scripts supply the bonding module
784 options in the ifcfg-bondX file, it is not necessary to add them to
785 the system /etc/modules.conf or /etc/modprobe.conf configuration file.
787 3.2 Configuration with Initscripts Support
788 ------------------------------------------
790 This section applies to distros using a version of initscripts
791 with bonding support, for example, Red Hat Linux 9 or Red Hat
792 Enterprise Linux version 3 or 4. On these systems, the network
793 initialization scripts have some knowledge of bonding, and can be
794 configured to control bonding devices.
796 These distros will not automatically load the network adapter
797 driver unless the ethX device is configured with an IP address.
798 Because of this constraint, users must manually configure a
799 network-script file for all physical adapters that will be members of
800 a bondX link. Network script files are located in the directory:
802 /etc/sysconfig/network-scripts
804 The file name must be prefixed with "ifcfg-eth" and suffixed
805 with the adapter's physical adapter number. For example, the script
806 for eth0 would be named /etc/sysconfig/network-scripts/ifcfg-eth0.
807 Place the following text in the file:
816 The DEVICE= line will be different for every ethX device and
817 must correspond with the name of the file, i.e., ifcfg-eth1 must have
818 a device line of DEVICE=eth1. The setting of the MASTER= line will
819 also depend on the final bonding interface name chosen for your bond.
820 As with other network devices, these typically start at 0, and go up
821 one for each device, i.e., the first bonding instance is bond0, the
822 second is bond1, and so on.
824 Next, create a bond network script. The file name for this
825 script will be /etc/sysconfig/network-scripts/ifcfg-bondX where X is
826 the number of the bond. For bond0 the file is named "ifcfg-bond0",
827 for bond1 it is named "ifcfg-bond1", and so on. Within that file,
828 place the following text:
832 NETMASK=255.255.255.0
834 BROADCAST=192.168.1.255
839 Be sure to change the networking specific lines (IPADDR,
840 NETMASK, NETWORK and BROADCAST) to match your network configuration.
842 Finally, it is necessary to edit /etc/modules.conf (or
843 /etc/modprobe.conf, depending upon your distro) to load the bonding
844 module with your desired options when the bond0 interface is brought
845 up. The following lines in /etc/modules.conf (or modprobe.conf) will
846 load the bonding module, and select its options:
849 options bond0 mode=balance-alb miimon=100
851 Replace the sample parameters with the appropriate set of
852 options for your configuration.
854 Finally run "/etc/rc.d/init.d/network restart" as root. This
855 will restart the networking subsystem and your bond link should be now
858 3.2.1 Using DHCP with Initscripts
859 ---------------------------------
861 Recent versions of initscripts (the version supplied with
862 Fedora Core 3 and Red Hat Enterprise Linux 4 is reported to work) do
863 have support for assigning IP information to bonding devices via DHCP.
865 To configure bonding for DHCP, configure it as described
866 above, except replace the line "BOOTPROTO=none" with "BOOTPROTO=dhcp"
867 and add a line consisting of "TYPE=Bonding". Note that the TYPE value
870 3.2.2 Configuring Multiple Bonds with Initscripts
871 -------------------------------------------------
873 At this writing, the initscripts package does not directly
874 support loading the bonding driver multiple times, so the process for
875 doing so is the same as described in the "Configuring Multiple Bonds
876 Manually" section, below.
878 NOTE: It has been observed that some Red Hat supplied kernels
879 are apparently unable to rename modules at load time (the "-o bond1"
880 part). Attempts to pass that option to modprobe will produce an
881 "Operation not permitted" error. This has been reported on some
882 Fedora Core kernels, and has been seen on RHEL 4 as well. On kernels
883 exhibiting this problem, it will be impossible to configure multiple
884 bonds with differing parameters.
886 3.3 Configuring Bonding Manually with Ifenslave
887 -----------------------------------------------
889 This section applies to distros whose network initialization
890 scripts (the sysconfig or initscripts package) do not have specific
891 knowledge of bonding. One such distro is SuSE Linux Enterprise Server
894 The general method for these systems is to place the bonding
895 module parameters into /etc/modules.conf or /etc/modprobe.conf (as
896 appropriate for the installed distro), then add modprobe and/or
897 ifenslave commands to the system's global init script. The name of
898 the global init script differs; for sysconfig, it is
899 /etc/init.d/boot.local and for initscripts it is /etc/rc.d/rc.local.
901 For example, if you wanted to make a simple bond of two e100
902 devices (presumed to be eth0 and eth1), and have it persist across
903 reboots, edit the appropriate file (/etc/init.d/boot.local or
904 /etc/rc.d/rc.local), and add the following:
906 modprobe bonding mode=balance-alb miimon=100
908 ifconfig bond0 192.168.1.1 netmask 255.255.255.0 up
912 Replace the example bonding module parameters and bond0
913 network configuration (IP address, netmask, etc) with the appropriate
914 values for your configuration.
916 Unfortunately, this method will not provide support for the
917 ifup and ifdown scripts on the bond devices. To reload the bonding
918 configuration, it is necessary to run the initialization script, e.g.,
920 # /etc/init.d/boot.local
926 It may be desirable in such a case to create a separate script
927 which only initializes the bonding configuration, then call that
928 separate script from within boot.local. This allows for bonding to be
929 enabled without re-running the entire global init script.
931 To shut down the bonding devices, it is necessary to first
932 mark the bonding device itself as being down, then remove the
933 appropriate device driver modules. For our example above, you can do
936 # ifconfig bond0 down
940 Again, for convenience, it may be desirable to create a script
944 3.3.1 Configuring Multiple Bonds Manually
945 -----------------------------------------
947 This section contains information on configuring multiple
948 bonding devices with differing options for those systems whose network
949 initialization scripts lack support for configuring multiple bonds.
951 If you require multiple bonding devices, but all with the same
952 options, you may wish to use the "max_bonds" module parameter,
955 To create multiple bonding devices with differing options, it
956 is necessary to use bonding parameters exported by sysfs, documented
957 in the section below.
960 3.4 Configuring Bonding Manually via Sysfs
961 ------------------------------------------
963 Starting with version 3.0, Channel Bonding may be configured
964 via the sysfs interface. This interface allows dynamic configuration
965 of all bonds in the system without unloading the module. It also
966 allows for adding and removing bonds at runtime. Ifenslave is no
967 longer required, though it is still supported.
969 Use of the sysfs interface allows you to use multiple bonds
970 with different configurations without having to reload the module.
971 It also allows you to use multiple, differently configured bonds when
972 bonding is compiled into the kernel.
974 You must have the sysfs filesystem mounted to configure
975 bonding this way. The examples in this document assume that you
976 are using the standard mount point for sysfs, e.g. /sys. If your
977 sysfs filesystem is mounted elsewhere, you will need to adjust the
978 example paths accordingly.
980 Creating and Destroying Bonds
981 -----------------------------
982 To add a new bond foo:
983 # echo +foo > /sys/class/net/bonding_masters
985 To remove an existing bond bar:
986 # echo -bar > /sys/class/net/bonding_masters
988 To show all existing bonds:
989 # cat /sys/class/net/bonding_masters
991 NOTE: due to 4K size limitation of sysfs files, this list may be
992 truncated if you have more than a few hundred bonds. This is unlikely
993 to occur under normal operating conditions.
995 Adding and Removing Slaves
996 --------------------------
997 Interfaces may be enslaved to a bond using the file
998 /sys/class/net/<bond>/bonding/slaves. The semantics for this file
999 are the same as for the bonding_masters file.
1001 To enslave interface eth0 to bond bond0:
1003 # echo +eth0 > /sys/class/net/bond0/bonding/slaves
1005 To free slave eth0 from bond bond0:
1006 # echo -eth0 > /sys/class/net/bond0/bonding/slaves
1008 NOTE: The bond must be up before slaves can be added. All
1009 slaves are freed when the interface is brought down.
1011 When an interface is enslaved to a bond, symlinks between the
1012 two are created in the sysfs filesystem. In this case, you would get
1013 /sys/class/net/bond0/slave_eth0 pointing to /sys/class/net/eth0, and
1014 /sys/class/net/eth0/master pointing to /sys/class/net/bond0.
1016 This means that you can tell quickly whether or not an
1017 interface is enslaved by looking for the master symlink. Thus:
1018 # echo -eth0 > /sys/class/net/eth0/master/bonding/slaves
1019 will free eth0 from whatever bond it is enslaved to, regardless of
1020 the name of the bond interface.
1022 Changing a Bond's Configuration
1023 -------------------------------
1024 Each bond may be configured individually by manipulating the
1025 files located in /sys/class/net/<bond name>/bonding
1027 The names of these files correspond directly with the command-
1028 line parameters described elsewhere in this file, and, with the
1029 exception of arp_ip_target, they accept the same values. To see the
1030 current setting, simply cat the appropriate file.
1032 A few examples will be given here; for specific usage
1033 guidelines for each parameter, see the appropriate section in this
1036 To configure bond0 for balance-alb mode:
1037 # ifconfig bond0 down
1038 # echo 6 > /sys/class/net/bond0/bonding/mode
1040 # echo balance-alb > /sys/class/net/bond0/bonding/mode
1041 NOTE: The bond interface must be down before the mode can be
1044 To enable MII monitoring on bond0 with a 1 second interval:
1045 # echo 1000 > /sys/class/net/bond0/bonding/miimon
1046 NOTE: If ARP monitoring is enabled, it will disabled when MII
1047 monitoring is enabled, and vice-versa.
1050 # echo +192.168.0.100 > /sys/class/net/bond0/bonding/arp_ip_target
1051 # echo +192.168.0.101 > /sys/class/net/bond0/bonding/arp_ip_target
1052 NOTE: up to 10 target addresses may be specified.
1054 To remove an ARP target:
1055 # echo -192.168.0.100 > /sys/class/net/bond0/bonding/arp_ip_target
1057 Example Configuration
1058 ---------------------
1059 We begin with the same example that is shown in section 3.3,
1060 executed with sysfs, and without using ifenslave.
1062 To make a simple bond of two e100 devices (presumed to be eth0
1063 and eth1), and have it persist across reboots, edit the appropriate
1064 file (/etc/init.d/boot.local or /etc/rc.d/rc.local), and add the
1069 echo balance-alb > /sys/class/net/bond0/bonding/mode
1070 ifconfig bond0 192.168.1.1 netmask 255.255.255.0 up
1071 echo 100 > /sys/class/net/bond0/bonding/miimon
1072 echo +eth0 > /sys/class/net/bond0/bonding/slaves
1073 echo +eth1 > /sys/class/net/bond0/bonding/slaves
1075 To add a second bond, with two e1000 interfaces in
1076 active-backup mode, using ARP monitoring, add the following lines to
1080 echo +bond1 > /sys/class/net/bonding_masters
1081 echo active-backup > /sys/class/net/bond1/bonding/mode
1082 ifconfig bond1 192.168.2.1 netmask 255.255.255.0 up
1083 echo +192.168.2.100 /sys/class/net/bond1/bonding/arp_ip_target
1084 echo 2000 > /sys/class/net/bond1/bonding/arp_interval
1085 echo +eth2 > /sys/class/net/bond1/bonding/slaves
1086 echo +eth3 > /sys/class/net/bond1/bonding/slaves
1089 4. Querying Bonding Configuration
1090 =================================
1092 4.1 Bonding Configuration
1093 -------------------------
1095 Each bonding device has a read-only file residing in the
1096 /proc/net/bonding directory. The file contents include information
1097 about the bonding configuration, options and state of each slave.
1099 For example, the contents of /proc/net/bonding/bond0 after the
1100 driver is loaded with parameters of mode=0 and miimon=1000 is
1101 generally as follows:
1103 Ethernet Channel Bonding Driver: 2.6.1 (October 29, 2004)
1104 Bonding Mode: load balancing (round-robin)
1105 Currently Active Slave: eth0
1107 MII Polling Interval (ms): 1000
1111 Slave Interface: eth1
1113 Link Failure Count: 1
1115 Slave Interface: eth0
1117 Link Failure Count: 1
1119 The precise format and contents will change depending upon the
1120 bonding configuration, state, and version of the bonding driver.
1122 4.2 Network configuration
1123 -------------------------
1125 The network configuration can be inspected using the ifconfig
1126 command. Bonding devices will have the MASTER flag set; Bonding slave
1127 devices will have the SLAVE flag set. The ifconfig output does not
1128 contain information on which slaves are associated with which masters.
1130 In the example below, the bond0 interface is the master
1131 (MASTER) while eth0 and eth1 are slaves (SLAVE). Notice all slaves of
1132 bond0 have the same MAC address (HWaddr) as bond0 for all modes except
1133 TLB and ALB that require a unique MAC address for each slave.
1136 bond0 Link encap:Ethernet HWaddr 00:C0:F0:1F:37:B4
1137 inet addr:XXX.XXX.XXX.YYY Bcast:XXX.XXX.XXX.255 Mask:255.255.252.0
1138 UP BROADCAST RUNNING MASTER MULTICAST MTU:1500 Metric:1
1139 RX packets:7224794 errors:0 dropped:0 overruns:0 frame:0
1140 TX packets:3286647 errors:1 dropped:0 overruns:1 carrier:0
1141 collisions:0 txqueuelen:0
1143 eth0 Link encap:Ethernet HWaddr 00:C0:F0:1F:37:B4
1144 UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
1145 RX packets:3573025 errors:0 dropped:0 overruns:0 frame:0
1146 TX packets:1643167 errors:1 dropped:0 overruns:1 carrier:0
1147 collisions:0 txqueuelen:100
1148 Interrupt:10 Base address:0x1080
1150 eth1 Link encap:Ethernet HWaddr 00:C0:F0:1F:37:B4
1151 UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
1152 RX packets:3651769 errors:0 dropped:0 overruns:0 frame:0
1153 TX packets:1643480 errors:0 dropped:0 overruns:0 carrier:0
1154 collisions:0 txqueuelen:100
1155 Interrupt:9 Base address:0x1400
1157 5. Switch Configuration
1158 =======================
1160 For this section, "switch" refers to whatever system the
1161 bonded devices are directly connected to (i.e., where the other end of
1162 the cable plugs into). This may be an actual dedicated switch device,
1163 or it may be another regular system (e.g., another computer running
1166 The active-backup, balance-tlb and balance-alb modes do not
1167 require any specific configuration of the switch.
1169 The 802.3ad mode requires that the switch have the appropriate
1170 ports configured as an 802.3ad aggregation. The precise method used
1171 to configure this varies from switch to switch, but, for example, a
1172 Cisco 3550 series switch requires that the appropriate ports first be
1173 grouped together in a single etherchannel instance, then that
1174 etherchannel is set to mode "lacp" to enable 802.3ad (instead of
1175 standard EtherChannel).
1177 The balance-rr, balance-xor and broadcast modes generally
1178 require that the switch have the appropriate ports grouped together.
1179 The nomenclature for such a group differs between switches, it may be
1180 called an "etherchannel" (as in the Cisco example, above), a "trunk
1181 group" or some other similar variation. For these modes, each switch
1182 will also have its own configuration options for the switch's transmit
1183 policy to the bond. Typical choices include XOR of either the MAC or
1184 IP addresses. The transmit policy of the two peers does not need to
1185 match. For these three modes, the bonding mode really selects a
1186 transmit policy for an EtherChannel group; all three will interoperate
1187 with another EtherChannel group.
1190 6. 802.1q VLAN Support
1191 ======================
1193 It is possible to configure VLAN devices over a bond interface
1194 using the 8021q driver. However, only packets coming from the 8021q
1195 driver and passing through bonding will be tagged by default. Self
1196 generated packets, for example, bonding's learning packets or ARP
1197 packets generated by either ALB mode or the ARP monitor mechanism, are
1198 tagged internally by bonding itself. As a result, bonding must
1199 "learn" the VLAN IDs configured above it, and use those IDs to tag
1200 self generated packets.
1202 For reasons of simplicity, and to support the use of adapters
1203 that can do VLAN hardware acceleration offloading, the bonding
1204 interface declares itself as fully hardware offloading capable, it gets
1205 the add_vid/kill_vid notifications to gather the necessary
1206 information, and it propagates those actions to the slaves. In case
1207 of mixed adapter types, hardware accelerated tagged packets that
1208 should go through an adapter that is not offloading capable are
1209 "un-accelerated" by the bonding driver so the VLAN tag sits in the
1212 VLAN interfaces *must* be added on top of a bonding interface
1213 only after enslaving at least one slave. The bonding interface has a
1214 hardware address of 00:00:00:00:00:00 until the first slave is added.
1215 If the VLAN interface is created prior to the first enslavement, it
1216 would pick up the all-zeroes hardware address. Once the first slave
1217 is attached to the bond, the bond device itself will pick up the
1218 slave's hardware address, which is then available for the VLAN device.
1220 Also, be aware that a similar problem can occur if all slaves
1221 are released from a bond that still has one or more VLAN interfaces on
1222 top of it. When a new slave is added, the bonding interface will
1223 obtain its hardware address from the first slave, which might not
1224 match the hardware address of the VLAN interfaces (which was
1225 ultimately copied from an earlier slave).
1227 There are two methods to insure that the VLAN device operates
1228 with the correct hardware address if all slaves are removed from a
1231 1. Remove all VLAN interfaces then recreate them
1233 2. Set the bonding interface's hardware address so that it
1234 matches the hardware address of the VLAN interfaces.
1236 Note that changing a VLAN interface's HW address would set the
1237 underlying device -- i.e. the bonding interface -- to promiscuous
1238 mode, which might not be what you want.
1244 The bonding driver at present supports two schemes for
1245 monitoring a slave device's link state: the ARP monitor and the MII
1248 At the present time, due to implementation restrictions in the
1249 bonding driver itself, it is not possible to enable both ARP and MII
1250 monitoring simultaneously.
1252 7.1 ARP Monitor Operation
1253 -------------------------
1255 The ARP monitor operates as its name suggests: it sends ARP
1256 queries to one or more designated peer systems on the network, and
1257 uses the response as an indication that the link is operating. This
1258 gives some assurance that traffic is actually flowing to and from one
1259 or more peers on the local network.
1261 The ARP monitor relies on the device driver itself to verify
1262 that traffic is flowing. In particular, the driver must keep up to
1263 date the last receive time, dev->last_rx, and transmit start time,
1264 dev->trans_start. If these are not updated by the driver, then the
1265 ARP monitor will immediately fail any slaves using that driver, and
1266 those slaves will stay down. If networking monitoring (tcpdump, etc)
1267 shows the ARP requests and replies on the network, then it may be that
1268 your device driver is not updating last_rx and trans_start.
1270 7.2 Configuring Multiple ARP Targets
1271 ------------------------------------
1273 While ARP monitoring can be done with just one target, it can
1274 be useful in a High Availability setup to have several targets to
1275 monitor. In the case of just one target, the target itself may go
1276 down or have a problem making it unresponsive to ARP requests. Having
1277 an additional target (or several) increases the reliability of the ARP
1280 Multiple ARP targets must be separated by commas as follows:
1282 # example options for ARP monitoring with three targets
1284 options bond0 arp_interval=60 arp_ip_target=192.168.0.1,192.168.0.3,192.168.0.9
1286 For just a single target the options would resemble:
1288 # example options for ARP monitoring with one target
1290 options bond0 arp_interval=60 arp_ip_target=192.168.0.100
1293 7.3 MII Monitor Operation
1294 -------------------------
1296 The MII monitor monitors only the carrier state of the local
1297 network interface. It accomplishes this in one of three ways: by
1298 depending upon the device driver to maintain its carrier state, by
1299 querying the device's MII registers, or by making an ethtool query to
1302 If the use_carrier module parameter is 1 (the default value),
1303 then the MII monitor will rely on the driver for carrier state
1304 information (via the netif_carrier subsystem). As explained in the
1305 use_carrier parameter information, above, if the MII monitor fails to
1306 detect carrier loss on the device (e.g., when the cable is physically
1307 disconnected), it may be that the driver does not support
1310 If use_carrier is 0, then the MII monitor will first query the
1311 device's (via ioctl) MII registers and check the link state. If that
1312 request fails (not just that it returns carrier down), then the MII
1313 monitor will make an ethtool ETHOOL_GLINK request to attempt to obtain
1314 the same information. If both methods fail (i.e., the driver either
1315 does not support or had some error in processing both the MII register
1316 and ethtool requests), then the MII monitor will assume the link is
1319 8. Potential Sources of Trouble
1320 ===============================
1322 8.1 Adventures in Routing
1323 -------------------------
1325 When bonding is configured, it is important that the slave
1326 devices not have routes that supersede routes of the master (or,
1327 generally, not have routes at all). For example, suppose the bonding
1328 device bond0 has two slaves, eth0 and eth1, and the routing table is
1331 Kernel IP routing table
1332 Destination Gateway Genmask Flags MSS Window irtt Iface
1333 10.0.0.0 0.0.0.0 255.255.0.0 U 40 0 0 eth0
1334 10.0.0.0 0.0.0.0 255.255.0.0 U 40 0 0 eth1
1335 10.0.0.0 0.0.0.0 255.255.0.0 U 40 0 0 bond0
1336 127.0.0.0 0.0.0.0 255.0.0.0 U 40 0 0 lo
1338 This routing configuration will likely still update the
1339 receive/transmit times in the driver (needed by the ARP monitor), but
1340 may bypass the bonding driver (because outgoing traffic to, in this
1341 case, another host on network 10 would use eth0 or eth1 before bond0).
1343 The ARP monitor (and ARP itself) may become confused by this
1344 configuration, because ARP requests (generated by the ARP monitor)
1345 will be sent on one interface (bond0), but the corresponding reply
1346 will arrive on a different interface (eth0). This reply looks to ARP
1347 as an unsolicited ARP reply (because ARP matches replies on an
1348 interface basis), and is discarded. The MII monitor is not affected
1349 by the state of the routing table.
1351 The solution here is simply to insure that slaves do not have
1352 routes of their own, and if for some reason they must, those routes do
1353 not supersede routes of their master. This should generally be the
1354 case, but unusual configurations or errant manual or automatic static
1355 route additions may cause trouble.
1357 8.2 Ethernet Device Renaming
1358 ----------------------------
1360 On systems with network configuration scripts that do not
1361 associate physical devices directly with network interface names (so
1362 that the same physical device always has the same "ethX" name), it may
1363 be necessary to add some special logic to either /etc/modules.conf or
1364 /etc/modprobe.conf (depending upon which is installed on the system).
1366 For example, given a modules.conf containing the following:
1369 options bond0 mode=some-mode miimon=50
1375 If neither eth0 and eth1 are slaves to bond0, then when the
1376 bond0 interface comes up, the devices may end up reordered. This
1377 happens because bonding is loaded first, then its slave device's
1378 drivers are loaded next. Since no other drivers have been loaded,
1379 when the e1000 driver loads, it will receive eth0 and eth1 for its
1380 devices, but the bonding configuration tries to enslave eth2 and eth3
1381 (which may later be assigned to the tg3 devices).
1383 Adding the following:
1385 add above bonding e1000 tg3
1387 causes modprobe to load e1000 then tg3, in that order, when
1388 bonding is loaded. This command is fully documented in the
1389 modules.conf manual page.
1391 On systems utilizing modprobe.conf (or modprobe.conf.local),
1392 an equivalent problem can occur. In this case, the following can be
1393 added to modprobe.conf (or modprobe.conf.local, as appropriate), as
1394 follows (all on one line; it has been split here for clarity):
1396 install bonding /sbin/modprobe tg3; /sbin/modprobe e1000;
1397 /sbin/modprobe --ignore-install bonding
1399 This will, when loading the bonding module, rather than
1400 performing the normal action, instead execute the provided command.
1401 This command loads the device drivers in the order needed, then calls
1402 modprobe with --ignore-install to cause the normal action to then take
1403 place. Full documentation on this can be found in the modprobe.conf
1404 and modprobe manual pages.
1406 8.3. Painfully Slow Or No Failed Link Detection By Miimon
1407 ---------------------------------------------------------
1409 By default, bonding enables the use_carrier option, which
1410 instructs bonding to trust the driver to maintain carrier state.
1412 As discussed in the options section, above, some drivers do
1413 not support the netif_carrier_on/_off link state tracking system.
1414 With use_carrier enabled, bonding will always see these links as up,
1415 regardless of their actual state.
1417 Additionally, other drivers do support netif_carrier, but do
1418 not maintain it in real time, e.g., only polling the link state at
1419 some fixed interval. In this case, miimon will detect failures, but
1420 only after some long period of time has expired. If it appears that
1421 miimon is very slow in detecting link failures, try specifying
1422 use_carrier=0 to see if that improves the failure detection time. If
1423 it does, then it may be that the driver checks the carrier state at a
1424 fixed interval, but does not cache the MII register values (so the
1425 use_carrier=0 method of querying the registers directly works). If
1426 use_carrier=0 does not improve the failover, then the driver may cache
1427 the registers, or the problem may be elsewhere.
1429 Also, remember that miimon only checks for the device's
1430 carrier state. It has no way to determine the state of devices on or
1431 beyond other ports of a switch, or if a switch is refusing to pass
1432 traffic while still maintaining carrier on.
1437 If running SNMP agents, the bonding driver should be loaded
1438 before any network drivers participating in a bond. This requirement
1439 is due to the interface index (ipAdEntIfIndex) being associated to
1440 the first interface found with a given IP address. That is, there is
1441 only one ipAdEntIfIndex for each IP address. For example, if eth0 and
1442 eth1 are slaves of bond0 and the driver for eth0 is loaded before the
1443 bonding driver, the interface for the IP address will be associated
1444 with the eth0 interface. This configuration is shown below, the IP
1445 address 192.168.1.1 has an interface index of 2 which indexes to eth0
1446 in the ifDescr table (ifDescr.2).
1448 interfaces.ifTable.ifEntry.ifDescr.1 = lo
1449 interfaces.ifTable.ifEntry.ifDescr.2 = eth0
1450 interfaces.ifTable.ifEntry.ifDescr.3 = eth1
1451 interfaces.ifTable.ifEntry.ifDescr.4 = eth2
1452 interfaces.ifTable.ifEntry.ifDescr.5 = eth3
1453 interfaces.ifTable.ifEntry.ifDescr.6 = bond0
1454 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.10.10.10 = 5
1455 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.192.168.1.1 = 2
1456 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.74.20.94 = 4
1457 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.127.0.0.1 = 1
1459 This problem is avoided by loading the bonding driver before
1460 any network drivers participating in a bond. Below is an example of
1461 loading the bonding driver first, the IP address 192.168.1.1 is
1462 correctly associated with ifDescr.2.
1464 interfaces.ifTable.ifEntry.ifDescr.1 = lo
1465 interfaces.ifTable.ifEntry.ifDescr.2 = bond0
1466 interfaces.ifTable.ifEntry.ifDescr.3 = eth0
1467 interfaces.ifTable.ifEntry.ifDescr.4 = eth1
1468 interfaces.ifTable.ifEntry.ifDescr.5 = eth2
1469 interfaces.ifTable.ifEntry.ifDescr.6 = eth3
1470 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.10.10.10 = 6
1471 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.192.168.1.1 = 2
1472 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.74.20.94 = 5
1473 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.127.0.0.1 = 1
1475 While some distributions may not report the interface name in
1476 ifDescr, the association between the IP address and IfIndex remains
1477 and SNMP functions such as Interface_Scan_Next will report that
1480 10. Promiscuous mode
1481 ====================
1483 When running network monitoring tools, e.g., tcpdump, it is
1484 common to enable promiscuous mode on the device, so that all traffic
1485 is seen (instead of seeing only traffic destined for the local host).
1486 The bonding driver handles promiscuous mode changes to the bonding
1487 master device (e.g., bond0), and propagates the setting to the slave
1490 For the balance-rr, balance-xor, broadcast, and 802.3ad modes,
1491 the promiscuous mode setting is propagated to all slaves.
1493 For the active-backup, balance-tlb and balance-alb modes, the
1494 promiscuous mode setting is propagated only to the active slave.
1496 For balance-tlb mode, the active slave is the slave currently
1497 receiving inbound traffic.
1499 For balance-alb mode, the active slave is the slave used as a
1500 "primary." This slave is used for mode-specific control traffic, for
1501 sending to peers that are unassigned or if the load is unbalanced.
1503 For the active-backup, balance-tlb and balance-alb modes, when
1504 the active slave changes (e.g., due to a link failure), the
1505 promiscuous setting will be propagated to the new active slave.
1507 11. Configuring Bonding for High Availability
1508 =============================================
1510 High Availability refers to configurations that provide
1511 maximum network availability by having redundant or backup devices,
1512 links or switches between the host and the rest of the world. The
1513 goal is to provide the maximum availability of network connectivity
1514 (i.e., the network always works), even though other configurations
1515 could provide higher throughput.
1517 11.1 High Availability in a Single Switch Topology
1518 --------------------------------------------------
1520 If two hosts (or a host and a single switch) are directly
1521 connected via multiple physical links, then there is no availability
1522 penalty to optimizing for maximum bandwidth. In this case, there is
1523 only one switch (or peer), so if it fails, there is no alternative
1524 access to fail over to. Additionally, the bonding load balance modes
1525 support link monitoring of their members, so if individual links fail,
1526 the load will be rebalanced across the remaining devices.
1528 See Section 13, "Configuring Bonding for Maximum Throughput"
1529 for information on configuring bonding with one peer device.
1531 11.2 High Availability in a Multiple Switch Topology
1532 ----------------------------------------------------
1534 With multiple switches, the configuration of bonding and the
1535 network changes dramatically. In multiple switch topologies, there is
1536 a trade off between network availability and usable bandwidth.
1538 Below is a sample network, configured to maximize the
1539 availability of the network:
1543 +-----+----+ +-----+----+
1544 | |port2 ISL port2| |
1545 | switch A +--------------------------+ switch B |
1547 +-----+----+ +-----++---+
1550 +-------------+ host1 +---------------+
1553 In this configuration, there is a link between the two
1554 switches (ISL, or inter switch link), and multiple ports connecting to
1555 the outside world ("port3" on each switch). There is no technical
1556 reason that this could not be extended to a third switch.
1558 11.2.1 HA Bonding Mode Selection for Multiple Switch Topology
1559 -------------------------------------------------------------
1561 In a topology such as the example above, the active-backup and
1562 broadcast modes are the only useful bonding modes when optimizing for
1563 availability; the other modes require all links to terminate on the
1564 same peer for them to behave rationally.
1566 active-backup: This is generally the preferred mode, particularly if
1567 the switches have an ISL and play together well. If the
1568 network configuration is such that one switch is specifically
1569 a backup switch (e.g., has lower capacity, higher cost, etc),
1570 then the primary option can be used to insure that the
1571 preferred link is always used when it is available.
1573 broadcast: This mode is really a special purpose mode, and is suitable
1574 only for very specific needs. For example, if the two
1575 switches are not connected (no ISL), and the networks beyond
1576 them are totally independent. In this case, if it is
1577 necessary for some specific one-way traffic to reach both
1578 independent networks, then the broadcast mode may be suitable.
1580 11.2.2 HA Link Monitoring Selection for Multiple Switch Topology
1581 ----------------------------------------------------------------
1583 The choice of link monitoring ultimately depends upon your
1584 switch. If the switch can reliably fail ports in response to other
1585 failures, then either the MII or ARP monitors should work. For
1586 example, in the above example, if the "port3" link fails at the remote
1587 end, the MII monitor has no direct means to detect this. The ARP
1588 monitor could be configured with a target at the remote end of port3,
1589 thus detecting that failure without switch support.
1591 In general, however, in a multiple switch topology, the ARP
1592 monitor can provide a higher level of reliability in detecting end to
1593 end connectivity failures (which may be caused by the failure of any
1594 individual component to pass traffic for any reason). Additionally,
1595 the ARP monitor should be configured with multiple targets (at least
1596 one for each switch in the network). This will insure that,
1597 regardless of which switch is active, the ARP monitor has a suitable
1601 12. Configuring Bonding for Maximum Throughput
1602 ==============================================
1604 12.1 Maximizing Throughput in a Single Switch Topology
1605 ------------------------------------------------------
1607 In a single switch configuration, the best method to maximize
1608 throughput depends upon the application and network environment. The
1609 various load balancing modes each have strengths and weaknesses in
1610 different environments, as detailed below.
1612 For this discussion, we will break down the topologies into
1613 two categories. Depending upon the destination of most traffic, we
1614 categorize them into either "gatewayed" or "local" configurations.
1616 In a gatewayed configuration, the "switch" is acting primarily
1617 as a router, and the majority of traffic passes through this router to
1618 other networks. An example would be the following:
1621 +----------+ +----------+
1622 | |eth0 port1| | to other networks
1623 | Host A +---------------------+ router +------------------->
1624 | +---------------------+ | Hosts B and C are out
1625 | |eth1 port2| | here somewhere
1626 +----------+ +----------+
1628 The router may be a dedicated router device, or another host
1629 acting as a gateway. For our discussion, the important point is that
1630 the majority of traffic from Host A will pass through the router to
1631 some other network before reaching its final destination.
1633 In a gatewayed network configuration, although Host A may
1634 communicate with many other systems, all of its traffic will be sent
1635 and received via one other peer on the local network, the router.
1637 Note that the case of two systems connected directly via
1638 multiple physical links is, for purposes of configuring bonding, the
1639 same as a gatewayed configuration. In that case, it happens that all
1640 traffic is destined for the "gateway" itself, not some other network
1643 In a local configuration, the "switch" is acting primarily as
1644 a switch, and the majority of traffic passes through this switch to
1645 reach other stations on the same network. An example would be the
1648 +----------+ +----------+ +--------+
1649 | |eth0 port1| +-------+ Host B |
1650 | Host A +------------+ switch |port3 +--------+
1651 | +------------+ | +--------+
1652 | |eth1 port2| +------------------+ Host C |
1653 +----------+ +----------+port4 +--------+
1656 Again, the switch may be a dedicated switch device, or another
1657 host acting as a gateway. For our discussion, the important point is
1658 that the majority of traffic from Host A is destined for other hosts
1659 on the same local network (Hosts B and C in the above example).
1661 In summary, in a gatewayed configuration, traffic to and from
1662 the bonded device will be to the same MAC level peer on the network
1663 (the gateway itself, i.e., the router), regardless of its final
1664 destination. In a local configuration, traffic flows directly to and
1665 from the final destinations, thus, each destination (Host B, Host C)
1666 will be addressed directly by their individual MAC addresses.
1668 This distinction between a gatewayed and a local network
1669 configuration is important because many of the load balancing modes
1670 available use the MAC addresses of the local network source and
1671 destination to make load balancing decisions. The behavior of each
1672 mode is described below.
1675 12.1.1 MT Bonding Mode Selection for Single Switch Topology
1676 -----------------------------------------------------------
1678 This configuration is the easiest to set up and to understand,
1679 although you will have to decide which bonding mode best suits your
1680 needs. The trade offs for each mode are detailed below:
1682 balance-rr: This mode is the only mode that will permit a single
1683 TCP/IP connection to stripe traffic across multiple
1684 interfaces. It is therefore the only mode that will allow a
1685 single TCP/IP stream to utilize more than one interface's
1686 worth of throughput. This comes at a cost, however: the
1687 striping often results in peer systems receiving packets out
1688 of order, causing TCP/IP's congestion control system to kick
1689 in, often by retransmitting segments.
1691 It is possible to adjust TCP/IP's congestion limits by
1692 altering the net.ipv4.tcp_reordering sysctl parameter. The
1693 usual default value is 3, and the maximum useful value is 127.
1694 For a four interface balance-rr bond, expect that a single
1695 TCP/IP stream will utilize no more than approximately 2.3
1696 interface's worth of throughput, even after adjusting
1699 Note that this out of order delivery occurs when both the
1700 sending and receiving systems are utilizing a multiple
1701 interface bond. Consider a configuration in which a
1702 balance-rr bond feeds into a single higher capacity network
1703 channel (e.g., multiple 100Mb/sec ethernets feeding a single
1704 gigabit ethernet via an etherchannel capable switch). In this
1705 configuration, traffic sent from the multiple 100Mb devices to
1706 a destination connected to the gigabit device will not see
1707 packets out of order. However, traffic sent from the gigabit
1708 device to the multiple 100Mb devices may or may not see
1709 traffic out of order, depending upon the balance policy of the
1710 switch. Many switches do not support any modes that stripe
1711 traffic (instead choosing a port based upon IP or MAC level
1712 addresses); for those devices, traffic flowing from the
1713 gigabit device to the many 100Mb devices will only utilize one
1716 If you are utilizing protocols other than TCP/IP, UDP for
1717 example, and your application can tolerate out of order
1718 delivery, then this mode can allow for single stream datagram
1719 performance that scales near linearly as interfaces are added
1722 This mode requires the switch to have the appropriate ports
1723 configured for "etherchannel" or "trunking."
1725 active-backup: There is not much advantage in this network topology to
1726 the active-backup mode, as the inactive backup devices are all
1727 connected to the same peer as the primary. In this case, a
1728 load balancing mode (with link monitoring) will provide the
1729 same level of network availability, but with increased
1730 available bandwidth. On the plus side, active-backup mode
1731 does not require any configuration of the switch, so it may
1732 have value if the hardware available does not support any of
1733 the load balance modes.
1735 balance-xor: This mode will limit traffic such that packets destined
1736 for specific peers will always be sent over the same
1737 interface. Since the destination is determined by the MAC
1738 addresses involved, this mode works best in a "local" network
1739 configuration (as described above), with destinations all on
1740 the same local network. This mode is likely to be suboptimal
1741 if all your traffic is passed through a single router (i.e., a
1742 "gatewayed" network configuration, as described above).
1744 As with balance-rr, the switch ports need to be configured for
1745 "etherchannel" or "trunking."
1747 broadcast: Like active-backup, there is not much advantage to this
1748 mode in this type of network topology.
1750 802.3ad: This mode can be a good choice for this type of network
1751 topology. The 802.3ad mode is an IEEE standard, so all peers
1752 that implement 802.3ad should interoperate well. The 802.3ad
1753 protocol includes automatic configuration of the aggregates,
1754 so minimal manual configuration of the switch is needed
1755 (typically only to designate that some set of devices is
1756 available for 802.3ad). The 802.3ad standard also mandates
1757 that frames be delivered in order (within certain limits), so
1758 in general single connections will not see misordering of
1759 packets. The 802.3ad mode does have some drawbacks: the
1760 standard mandates that all devices in the aggregate operate at
1761 the same speed and duplex. Also, as with all bonding load
1762 balance modes other than balance-rr, no single connection will
1763 be able to utilize more than a single interface's worth of
1766 Additionally, the linux bonding 802.3ad implementation
1767 distributes traffic by peer (using an XOR of MAC addresses),
1768 so in a "gatewayed" configuration, all outgoing traffic will
1769 generally use the same device. Incoming traffic may also end
1770 up on a single device, but that is dependent upon the
1771 balancing policy of the peer's 8023.ad implementation. In a
1772 "local" configuration, traffic will be distributed across the
1773 devices in the bond.
1775 Finally, the 802.3ad mode mandates the use of the MII monitor,
1776 therefore, the ARP monitor is not available in this mode.
1778 balance-tlb: The balance-tlb mode balances outgoing traffic by peer.
1779 Since the balancing is done according to MAC address, in a
1780 "gatewayed" configuration (as described above), this mode will
1781 send all traffic across a single device. However, in a
1782 "local" network configuration, this mode balances multiple
1783 local network peers across devices in a vaguely intelligent
1784 manner (not a simple XOR as in balance-xor or 802.3ad mode),
1785 so that mathematically unlucky MAC addresses (i.e., ones that
1786 XOR to the same value) will not all "bunch up" on a single
1789 Unlike 802.3ad, interfaces may be of differing speeds, and no
1790 special switch configuration is required. On the down side,
1791 in this mode all incoming traffic arrives over a single
1792 interface, this mode requires certain ethtool support in the
1793 network device driver of the slave interfaces, and the ARP
1794 monitor is not available.
1796 balance-alb: This mode is everything that balance-tlb is, and more.
1797 It has all of the features (and restrictions) of balance-tlb,
1798 and will also balance incoming traffic from local network
1799 peers (as described in the Bonding Module Options section,
1802 The only additional down side to this mode is that the network
1803 device driver must support changing the hardware address while
1806 12.1.2 MT Link Monitoring for Single Switch Topology
1807 ----------------------------------------------------
1809 The choice of link monitoring may largely depend upon which
1810 mode you choose to use. The more advanced load balancing modes do not
1811 support the use of the ARP monitor, and are thus restricted to using
1812 the MII monitor (which does not provide as high a level of end to end
1813 assurance as the ARP monitor).
1815 12.2 Maximum Throughput in a Multiple Switch Topology
1816 -----------------------------------------------------
1818 Multiple switches may be utilized to optimize for throughput
1819 when they are configured in parallel as part of an isolated network
1820 between two or more systems, for example:
1826 +--------+ | +---------+
1828 +------+---+ +-----+----+ +-----+----+
1829 | Switch A | | Switch B | | Switch C |
1830 +------+---+ +-----+----+ +-----+----+
1832 +--------+ | +---------+
1838 In this configuration, the switches are isolated from one
1839 another. One reason to employ a topology such as this is for an
1840 isolated network with many hosts (a cluster configured for high
1841 performance, for example), using multiple smaller switches can be more
1842 cost effective than a single larger switch, e.g., on a network with 24
1843 hosts, three 24 port switches can be significantly less expensive than
1844 a single 72 port switch.
1846 If access beyond the network is required, an individual host
1847 can be equipped with an additional network device connected to an
1848 external network; this host then additionally acts as a gateway.
1850 12.2.1 MT Bonding Mode Selection for Multiple Switch Topology
1851 -------------------------------------------------------------
1853 In actual practice, the bonding mode typically employed in
1854 configurations of this type is balance-rr. Historically, in this
1855 network configuration, the usual caveats about out of order packet
1856 delivery are mitigated by the use of network adapters that do not do
1857 any kind of packet coalescing (via the use of NAPI, or because the
1858 device itself does not generate interrupts until some number of
1859 packets has arrived). When employed in this fashion, the balance-rr
1860 mode allows individual connections between two hosts to effectively
1861 utilize greater than one interface's bandwidth.
1863 12.2.2 MT Link Monitoring for Multiple Switch Topology
1864 ------------------------------------------------------
1866 Again, in actual practice, the MII monitor is most often used
1867 in this configuration, as performance is given preference over
1868 availability. The ARP monitor will function in this topology, but its
1869 advantages over the MII monitor are mitigated by the volume of probes
1870 needed as the number of systems involved grows (remember that each
1871 host in the network is configured with bonding).
1873 13. Switch Behavior Issues
1874 ==========================
1876 13.1 Link Establishment and Failover Delays
1877 -------------------------------------------
1879 Some switches exhibit undesirable behavior with regard to the
1880 timing of link up and down reporting by the switch.
1882 First, when a link comes up, some switches may indicate that
1883 the link is up (carrier available), but not pass traffic over the
1884 interface for some period of time. This delay is typically due to
1885 some type of autonegotiation or routing protocol, but may also occur
1886 during switch initialization (e.g., during recovery after a switch
1887 failure). If you find this to be a problem, specify an appropriate
1888 value to the updelay bonding module option to delay the use of the
1889 relevant interface(s).
1891 Second, some switches may "bounce" the link state one or more
1892 times while a link is changing state. This occurs most commonly while
1893 the switch is initializing. Again, an appropriate updelay value may
1896 Note that when a bonding interface has no active links, the
1897 driver will immediately reuse the first link that goes up, even if the
1898 updelay parameter has been specified (the updelay is ignored in this
1899 case). If there are slave interfaces waiting for the updelay timeout
1900 to expire, the interface that first went into that state will be
1901 immediately reused. This reduces down time of the network if the
1902 value of updelay has been overestimated, and since this occurs only in
1903 cases with no connectivity, there is no additional penalty for
1904 ignoring the updelay.
1906 In addition to the concerns about switch timings, if your
1907 switches take a long time to go into backup mode, it may be desirable
1908 to not activate a backup interface immediately after a link goes down.
1909 Failover may be delayed via the downdelay bonding module option.
1911 13.2 Duplicated Incoming Packets
1912 --------------------------------
1914 It is not uncommon to observe a short burst of duplicated
1915 traffic when the bonding device is first used, or after it has been
1916 idle for some period of time. This is most easily observed by issuing
1917 a "ping" to some other host on the network, and noticing that the
1918 output from ping flags duplicates (typically one per slave).
1920 For example, on a bond in active-backup mode with five slaves
1921 all connected to one switch, the output may appear as follows:
1924 PING 10.0.4.2 (10.0.4.2) from 10.0.3.10 : 56(84) bytes of data.
1925 64 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.7 ms
1926 64 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.8 ms (DUP!)
1927 64 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.8 ms (DUP!)
1928 64 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.8 ms (DUP!)
1929 64 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.8 ms (DUP!)
1930 64 bytes from 10.0.4.2: icmp_seq=2 ttl=64 time=0.216 ms
1931 64 bytes from 10.0.4.2: icmp_seq=3 ttl=64 time=0.267 ms
1932 64 bytes from 10.0.4.2: icmp_seq=4 ttl=64 time=0.222 ms
1934 This is not due to an error in the bonding driver, rather, it
1935 is a side effect of how many switches update their MAC forwarding
1936 tables. Initially, the switch does not associate the MAC address in
1937 the packet with a particular switch port, and so it may send the
1938 traffic to all ports until its MAC forwarding table is updated. Since
1939 the interfaces attached to the bond may occupy multiple ports on a
1940 single switch, when the switch (temporarily) floods the traffic to all
1941 ports, the bond device receives multiple copies of the same packet
1942 (one per slave device).
1944 The duplicated packet behavior is switch dependent, some
1945 switches exhibit this, and some do not. On switches that display this
1946 behavior, it can be induced by clearing the MAC forwarding table (on
1947 most Cisco switches, the privileged command "clear mac address-table
1948 dynamic" will accomplish this).
1950 14. Hardware Specific Considerations
1951 ====================================
1953 This section contains additional information for configuring
1954 bonding on specific hardware platforms, or for interfacing bonding
1955 with particular switches or other devices.
1957 14.1 IBM BladeCenter
1958 --------------------
1960 This applies to the JS20 and similar systems.
1962 On the JS20 blades, the bonding driver supports only
1963 balance-rr, active-backup, balance-tlb and balance-alb modes. This is
1964 largely due to the network topology inside the BladeCenter, detailed
1967 JS20 network adapter information
1968 --------------------------------
1970 All JS20s come with two Broadcom Gigabit Ethernet ports
1971 integrated on the planar (that's "motherboard" in IBM-speak). In the
1972 BladeCenter chassis, the eth0 port of all JS20 blades is hard wired to
1973 I/O Module #1; similarly, all eth1 ports are wired to I/O Module #2.
1974 An add-on Broadcom daughter card can be installed on a JS20 to provide
1975 two more Gigabit Ethernet ports. These ports, eth2 and eth3, are
1976 wired to I/O Modules 3 and 4, respectively.
1978 Each I/O Module may contain either a switch or a passthrough
1979 module (which allows ports to be directly connected to an external
1980 switch). Some bonding modes require a specific BladeCenter internal
1981 network topology in order to function; these are detailed below.
1983 Additional BladeCenter-specific networking information can be
1984 found in two IBM Redbooks (www.ibm.com/redbooks):
1986 "IBM eServer BladeCenter Networking Options"
1987 "IBM eServer BladeCenter Layer 2-7 Network Switching"
1989 BladeCenter networking configuration
1990 ------------------------------------
1992 Because a BladeCenter can be configured in a very large number
1993 of ways, this discussion will be confined to describing basic
1996 Normally, Ethernet Switch Modules (ESMs) are used in I/O
1997 modules 1 and 2. In this configuration, the eth0 and eth1 ports of a
1998 JS20 will be connected to different internal switches (in the
1999 respective I/O modules).
2001 A passthrough module (OPM or CPM, optical or copper,
2002 passthrough module) connects the I/O module directly to an external
2003 switch. By using PMs in I/O module #1 and #2, the eth0 and eth1
2004 interfaces of a JS20 can be redirected to the outside world and
2005 connected to a common external switch.
2007 Depending upon the mix of ESMs and PMs, the network will
2008 appear to bonding as either a single switch topology (all PMs) or as a
2009 multiple switch topology (one or more ESMs, zero or more PMs). It is
2010 also possible to connect ESMs together, resulting in a configuration
2011 much like the example in "High Availability in a Multiple Switch
2014 Requirements for specific modes
2015 -------------------------------
2017 The balance-rr mode requires the use of passthrough modules
2018 for devices in the bond, all connected to an common external switch.
2019 That switch must be configured for "etherchannel" or "trunking" on the
2020 appropriate ports, as is usual for balance-rr.
2022 The balance-alb and balance-tlb modes will function with
2023 either switch modules or passthrough modules (or a mix). The only
2024 specific requirement for these modes is that all network interfaces
2025 must be able to reach all destinations for traffic sent over the
2026 bonding device (i.e., the network must converge at some point outside
2029 The active-backup mode has no additional requirements.
2031 Link monitoring issues
2032 ----------------------
2034 When an Ethernet Switch Module is in place, only the ARP
2035 monitor will reliably detect link loss to an external switch. This is
2036 nothing unusual, but examination of the BladeCenter cabinet would
2037 suggest that the "external" network ports are the ethernet ports for
2038 the system, when it fact there is a switch between these "external"
2039 ports and the devices on the JS20 system itself. The MII monitor is
2040 only able to detect link failures between the ESM and the JS20 system.
2042 When a passthrough module is in place, the MII monitor does
2043 detect failures to the "external" port, which is then directly
2044 connected to the JS20 system.
2049 The Serial Over LAN (SoL) link is established over the primary
2050 ethernet (eth0) only, therefore, any loss of link to eth0 will result
2051 in losing your SoL connection. It will not fail over with other
2052 network traffic, as the SoL system is beyond the control of the
2055 It may be desirable to disable spanning tree on the switch
2056 (either the internal Ethernet Switch Module, or an external switch) to
2057 avoid fail-over delay issues when using bonding.
2060 15. Frequently Asked Questions
2061 ==============================
2065 Yes. The old 2.0.xx channel bonding patch was not SMP safe.
2066 The new driver was designed to be SMP safe from the start.
2068 2. What type of cards will work with it?
2070 Any Ethernet type cards (you can even mix cards - a Intel
2071 EtherExpress PRO/100 and a 3com 3c905b, for example). For most modes,
2072 devices need not be of the same speed.
2074 3. How many bonding devices can I have?
2078 4. How many slaves can a bonding device have?
2080 This is limited only by the number of network interfaces Linux
2081 supports and/or the number of network cards you can place in your
2084 5. What happens when a slave link dies?
2086 If link monitoring is enabled, then the failing device will be
2087 disabled. The active-backup mode will fail over to a backup link, and
2088 other modes will ignore the failed link. The link will continue to be
2089 monitored, and should it recover, it will rejoin the bond (in whatever
2090 manner is appropriate for the mode). See the sections on High
2091 Availability and the documentation for each mode for additional
2094 Link monitoring can be enabled via either the miimon or
2095 arp_interval parameters (described in the module parameters section,
2096 above). In general, miimon monitors the carrier state as sensed by
2097 the underlying network device, and the arp monitor (arp_interval)
2098 monitors connectivity to another host on the local network.
2100 If no link monitoring is configured, the bonding driver will
2101 be unable to detect link failures, and will assume that all links are
2102 always available. This will likely result in lost packets, and a
2103 resulting degradation of performance. The precise performance loss
2104 depends upon the bonding mode and network configuration.
2106 6. Can bonding be used for High Availability?
2108 Yes. See the section on High Availability for details.
2110 7. Which switches/systems does it work with?
2112 The full answer to this depends upon the desired mode.
2114 In the basic balance modes (balance-rr and balance-xor), it
2115 works with any system that supports etherchannel (also called
2116 trunking). Most managed switches currently available have such
2117 support, and many unmanaged switches as well.
2119 The advanced balance modes (balance-tlb and balance-alb) do
2120 not have special switch requirements, but do need device drivers that
2121 support specific features (described in the appropriate section under
2122 module parameters, above).
2124 In 802.3ad mode, it works with systems that support IEEE
2125 802.3ad Dynamic Link Aggregation. Most managed and many unmanaged
2126 switches currently available support 802.3ad.
2128 The active-backup mode should work with any Layer-II switch.
2130 8. Where does a bonding device get its MAC address from?
2132 If not explicitly configured (with ifconfig or ip link), the
2133 MAC address of the bonding device is taken from its first slave
2134 device. This MAC address is then passed to all following slaves and
2135 remains persistent (even if the first slave is removed) until the
2136 bonding device is brought down or reconfigured.
2138 If you wish to change the MAC address, you can set it with
2139 ifconfig or ip link:
2141 # ifconfig bond0 hw ether 00:11:22:33:44:55
2143 # ip link set bond0 address 66:77:88:99:aa:bb
2145 The MAC address can be also changed by bringing down/up the
2146 device and then changing its slaves (or their order):
2148 # ifconfig bond0 down ; modprobe -r bonding
2149 # ifconfig bond0 .... up
2150 # ifenslave bond0 eth...
2152 This method will automatically take the address from the next
2153 slave that is added.
2155 To restore your slaves' MAC addresses, you need to detach them
2156 from the bond (`ifenslave -d bond0 eth0'). The bonding driver will
2157 then restore the MAC addresses that the slaves had before they were
2160 16. Resources and Links
2161 =======================
2163 The latest version of the bonding driver can be found in the latest
2164 version of the linux kernel, found on http://kernel.org
2166 The latest version of this document can be found in either the latest
2167 kernel source (named Documentation/networking/bonding.txt), or on the
2168 bonding sourceforge site:
2170 http://www.sourceforge.net/projects/bonding
2172 Discussions regarding the bonding driver take place primarily on the
2173 bonding-devel mailing list, hosted at sourceforge.net. If you have
2174 questions or problems, post them to the list. The list address is:
2176 bonding-devel@lists.sourceforge.net
2178 The administrative interface (to subscribe or unsubscribe) can
2181 https://lists.sourceforge.net/lists/listinfo/bonding-devel
2183 Donald Becker's Ethernet Drivers and diag programs may be found at :
2184 - http://www.scyld.com/network/
2186 You will also find a lot of information regarding Ethernet, NWay, MII,
2187 etc. at www.scyld.com.