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 Option specifying the rate in which we'll ask our link partner
287 to transmit LACPDU packets in 802.3ad mode. Possible values
291 Request partner to transmit LACPDUs every 30 seconds
294 Request partner to transmit LACPDUs every 1 second
300 Specifies the number of bonding devices to create for this
301 instance of the bonding driver. E.g., if max_bonds is 3, and
302 the bonding driver is not already loaded, then bond0, bond1
303 and bond2 will be created. The default value is 1.
307 Specifies the MII link monitoring frequency in milliseconds.
308 This determines how often the link state of each slave is
309 inspected for link failures. A value of zero disables MII
310 link monitoring. A value of 100 is a good starting point.
311 The use_carrier option, below, affects how the link state is
312 determined. See the High Availability section for additional
313 information. The default value is 0.
317 Specifies one of the bonding policies. The default is
318 balance-rr (round robin). Possible values are:
322 Round-robin policy: Transmit packets in sequential
323 order from the first available slave through the
324 last. This mode provides load balancing and fault
329 Active-backup policy: Only one slave in the bond is
330 active. A different slave becomes active if, and only
331 if, the active slave fails. The bond's MAC address is
332 externally visible on only one port (network adapter)
333 to avoid confusing the switch.
335 In bonding version 2.6.2 or later, when a failover
336 occurs in active-backup mode, bonding will issue one
337 or more gratuitous ARPs on the newly active slave.
338 One gratuitous ARP is issued for the bonding master
339 interface and each VLAN interfaces configured above
340 it, provided that the interface has at least one IP
341 address configured. Gratuitous ARPs issued for VLAN
342 interfaces are tagged with the appropriate VLAN id.
344 This mode provides fault tolerance. The primary
345 option, documented below, affects the behavior of this
350 XOR policy: Transmit based on the selected transmit
351 hash policy. The default policy is a simple [(source
352 MAC address XOR'd with destination MAC address) modulo
353 slave count]. Alternate transmit policies may be
354 selected via the xmit_hash_policy option, described
357 This mode provides load balancing and fault tolerance.
361 Broadcast policy: transmits everything on all slave
362 interfaces. This mode provides fault tolerance.
366 IEEE 802.3ad Dynamic link aggregation. Creates
367 aggregation groups that share the same speed and
368 duplex settings. Utilizes all slaves in the active
369 aggregator according to the 802.3ad specification.
371 Slave selection for outgoing traffic is done according
372 to the transmit hash policy, which may be changed from
373 the default simple XOR policy via the xmit_hash_policy
374 option, documented below. Note that not all transmit
375 policies may be 802.3ad compliant, particularly in
376 regards to the packet mis-ordering requirements of
377 section 43.2.4 of the 802.3ad standard. Differing
378 peer implementations will have varying tolerances for
383 1. Ethtool support in the base drivers for retrieving
384 the speed and duplex of each slave.
386 2. A switch that supports IEEE 802.3ad Dynamic link
389 Most switches will require some type of configuration
390 to enable 802.3ad mode.
394 Adaptive transmit load balancing: channel bonding that
395 does not require any special switch support. The
396 outgoing traffic is distributed according to the
397 current load (computed relative to the speed) on each
398 slave. Incoming traffic is received by the current
399 slave. If the receiving slave fails, another slave
400 takes over the MAC address of the failed receiving
405 Ethtool support in the base drivers for retrieving the
410 Adaptive load balancing: includes balance-tlb plus
411 receive load balancing (rlb) for IPV4 traffic, and
412 does not require any special switch support. The
413 receive load balancing is achieved by ARP negotiation.
414 The bonding driver intercepts the ARP Replies sent by
415 the local system on their way out and overwrites the
416 source hardware address with the unique hardware
417 address of one of the slaves in the bond such that
418 different peers use different hardware addresses for
421 Receive traffic from connections created by the server
422 is also balanced. When the local system sends an ARP
423 Request the bonding driver copies and saves the peer's
424 IP information from the ARP packet. When the ARP
425 Reply arrives from the peer, its hardware address is
426 retrieved and the bonding driver initiates an ARP
427 reply to this peer assigning it to one of the slaves
428 in the bond. A problematic outcome of using ARP
429 negotiation for balancing is that each time that an
430 ARP request is broadcast it uses the hardware address
431 of the bond. Hence, peers learn the hardware address
432 of the bond and the balancing of receive traffic
433 collapses to the current slave. This is handled by
434 sending updates (ARP Replies) to all the peers with
435 their individually assigned hardware address such that
436 the traffic is redistributed. Receive traffic is also
437 redistributed when a new slave is added to the bond
438 and when an inactive slave is re-activated. The
439 receive load is distributed sequentially (round robin)
440 among the group of highest speed slaves in the bond.
442 When a link is reconnected or a new slave joins the
443 bond the receive traffic is redistributed among all
444 active slaves in the bond by initiating ARP Replies
445 with the selected MAC address to each of the
446 clients. The updelay parameter (detailed below) must
447 be set to a value equal or greater than the switch's
448 forwarding delay so that the ARP Replies sent to the
449 peers will not be blocked by the switch.
453 1. Ethtool support in the base drivers for retrieving
454 the speed of each slave.
456 2. Base driver support for setting the hardware
457 address of a device while it is open. This is
458 required so that there will always be one slave in the
459 team using the bond hardware address (the
460 curr_active_slave) while having a unique hardware
461 address for each slave in the bond. If the
462 curr_active_slave fails its hardware address is
463 swapped with the new curr_active_slave that was
468 A string (eth0, eth2, etc) specifying which slave is the
469 primary device. The specified device will always be the
470 active slave while it is available. Only when the primary is
471 off-line will alternate devices be used. This is useful when
472 one slave is preferred over another, e.g., when one slave has
473 higher throughput than another.
475 The primary option is only valid for active-backup mode.
479 Specifies the time, in milliseconds, to wait before enabling a
480 slave after a link recovery has been detected. This option is
481 only valid for the miimon link monitor. The updelay value
482 should be a multiple of the miimon value; if not, it will be
483 rounded down to the nearest multiple. The default value is 0.
487 Specifies whether or not miimon should use MII or ETHTOOL
488 ioctls vs. netif_carrier_ok() to determine the link
489 status. The MII or ETHTOOL ioctls are less efficient and
490 utilize a deprecated calling sequence within the kernel. The
491 netif_carrier_ok() relies on the device driver to maintain its
492 state with netif_carrier_on/off; at this writing, most, but
493 not all, device drivers support this facility.
495 If bonding insists that the link is up when it should not be,
496 it may be that your network device driver does not support
497 netif_carrier_on/off. The default state for netif_carrier is
498 "carrier on," so if a driver does not support netif_carrier,
499 it will appear as if the link is always up. In this case,
500 setting use_carrier to 0 will cause bonding to revert to the
501 MII / ETHTOOL ioctl method to determine the link state.
503 A value of 1 enables the use of netif_carrier_ok(), a value of
504 0 will use the deprecated MII / ETHTOOL ioctls. The default
509 Selects the transmit hash policy to use for slave selection in
510 balance-xor and 802.3ad modes. Possible values are:
514 Uses XOR of hardware MAC addresses to generate the
517 (source MAC XOR destination MAC) modulo slave count
519 This algorithm will place all traffic to a particular
520 network peer on the same slave.
522 This algorithm is 802.3ad compliant.
526 This policy uses upper layer protocol information,
527 when available, to generate the hash. This allows for
528 traffic to a particular network peer to span multiple
529 slaves, although a single connection will not span
532 The formula for unfragmented TCP and UDP packets is
534 ((source port XOR dest port) XOR
535 ((source IP XOR dest IP) AND 0xffff)
538 For fragmented TCP or UDP packets and all other IP
539 protocol traffic, the source and destination port
540 information is omitted. For non-IP traffic, the
541 formula is the same as for the layer2 transmit hash
544 This policy is intended to mimic the behavior of
545 certain switches, notably Cisco switches with PFC2 as
546 well as some Foundry and IBM products.
548 This algorithm is not fully 802.3ad compliant. A
549 single TCP or UDP conversation containing both
550 fragmented and unfragmented packets will see packets
551 striped across two interfaces. This may result in out
552 of order delivery. Most traffic types will not meet
553 this criteria, as TCP rarely fragments traffic, and
554 most UDP traffic is not involved in extended
555 conversations. Other implementations of 802.3ad may
556 or may not tolerate this noncompliance.
558 The default value is layer2. This option was added in bonding
559 version 2.6.3. In earlier versions of bonding, this parameter does
560 not exist, and the layer2 policy is the only policy.
563 3. Configuring Bonding Devices
564 ==============================
566 You can configure bonding using either your distro's network
567 initialization scripts, or manually using either ifenslave or the
568 sysfs interface. Distros generally use one of two packages for the
569 network initialization scripts: initscripts or sysconfig. Recent
570 versions of these packages have support for bonding, while older
573 We will first describe the options for configuring bonding for
574 distros using versions of initscripts and sysconfig with full or
575 partial support for bonding, then provide information on enabling
576 bonding without support from the network initialization scripts (i.e.,
577 older versions of initscripts or sysconfig).
579 If you're unsure whether your distro uses sysconfig or
580 initscripts, or don't know if it's new enough, have no fear.
581 Determining this is fairly straightforward.
583 First, issue the command:
587 It will respond with a line of text starting with either
588 "initscripts" or "sysconfig," followed by some numbers. This is the
589 package that provides your network initialization scripts.
591 Next, to determine if your installation supports bonding,
594 $ grep ifenslave /sbin/ifup
596 If this returns any matches, then your initscripts or
597 sysconfig has support for bonding.
599 3.1 Configuration with Sysconfig Support
600 ----------------------------------------
602 This section applies to distros using a version of sysconfig
603 with bonding support, for example, SuSE Linux Enterprise Server 9.
605 SuSE SLES 9's networking configuration system does support
606 bonding, however, at this writing, the YaST system configuration
607 front end does not provide any means to work with bonding devices.
608 Bonding devices can be managed by hand, however, as follows.
610 First, if they have not already been configured, configure the
611 slave devices. On SLES 9, this is most easily done by running the
612 yast2 sysconfig configuration utility. The goal is for to create an
613 ifcfg-id file for each slave device. The simplest way to accomplish
614 this is to configure the devices for DHCP (this is only to get the
615 file ifcfg-id file created; see below for some issues with DHCP). The
616 name of the configuration file for each device will be of the form:
618 ifcfg-id-xx:xx:xx:xx:xx:xx
620 Where the "xx" portion will be replaced with the digits from
621 the device's permanent MAC address.
623 Once the set of ifcfg-id-xx:xx:xx:xx:xx:xx files has been
624 created, it is necessary to edit the configuration files for the slave
625 devices (the MAC addresses correspond to those of the slave devices).
626 Before editing, the file will contain multiple lines, and will look
632 UNIQUE='XNzu.WeZGOGF+4wE'
633 _nm_name='bus-pci-0001:61:01.0'
635 Change the BOOTPROTO and STARTMODE lines to the following:
640 Do not alter the UNIQUE or _nm_name lines. Remove any other
641 lines (USERCTL, etc).
643 Once the ifcfg-id-xx:xx:xx:xx:xx:xx files have been modified,
644 it's time to create the configuration file for the bonding device
645 itself. This file is named ifcfg-bondX, where X is the number of the
646 bonding device to create, starting at 0. The first such file is
647 ifcfg-bond0, the second is ifcfg-bond1, and so on. The sysconfig
648 network configuration system will correctly start multiple instances
651 The contents of the ifcfg-bondX file is as follows:
654 BROADCAST="10.0.2.255"
656 NETMASK="255.255.0.0"
661 BONDING_MODULE_OPTS="mode=active-backup miimon=100"
662 BONDING_SLAVE0="eth0"
663 BONDING_SLAVE1="bus-pci-0000:06:08.1"
665 Replace the sample BROADCAST, IPADDR, NETMASK and NETWORK
666 values with the appropriate values for your network.
668 The STARTMODE specifies when the device is brought online.
669 The possible values are:
671 onboot: The device is started at boot time. If you're not
672 sure, this is probably what you want.
674 manual: The device is started only when ifup is called
675 manually. Bonding devices may be configured this
676 way if you do not wish them to start automatically
677 at boot for some reason.
679 hotplug: The device is started by a hotplug event. This is not
680 a valid choice for a bonding device.
682 off or ignore: The device configuration is ignored.
684 The line BONDING_MASTER='yes' indicates that the device is a
685 bonding master device. The only useful value is "yes."
687 The contents of BONDING_MODULE_OPTS are supplied to the
688 instance of the bonding module for this device. Specify the options
689 for the bonding mode, link monitoring, and so on here. Do not include
690 the max_bonds bonding parameter; this will confuse the configuration
691 system if you have multiple bonding devices.
693 Finally, supply one BONDING_SLAVEn="slave device" for each
694 slave. where "n" is an increasing value, one for each slave. The
695 "slave device" is either an interface name, e.g., "eth0", or a device
696 specifier for the network device. The interface name is easier to
697 find, but the ethN names are subject to change at boot time if, e.g.,
698 a device early in the sequence has failed. The device specifiers
699 (bus-pci-0000:06:08.1 in the example above) specify the physical
700 network device, and will not change unless the device's bus location
701 changes (for example, it is moved from one PCI slot to another). The
702 example above uses one of each type for demonstration purposes; most
703 configurations will choose one or the other for all slave devices.
705 When all configuration files have been modified or created,
706 networking must be restarted for the configuration changes to take
707 effect. This can be accomplished via the following:
709 # /etc/init.d/network restart
711 Note that the network control script (/sbin/ifdown) will
712 remove the bonding module as part of the network shutdown processing,
713 so it is not necessary to remove the module by hand if, e.g., the
714 module parameters have changed.
716 Also, at this writing, YaST/YaST2 will not manage bonding
717 devices (they do not show bonding interfaces on its list of network
718 devices). It is necessary to edit the configuration file by hand to
719 change the bonding configuration.
721 Additional general options and details of the ifcfg file
722 format can be found in an example ifcfg template file:
724 /etc/sysconfig/network/ifcfg.template
726 Note that the template does not document the various BONDING_
727 settings described above, but does describe many of the other options.
729 3.1.1 Using DHCP with Sysconfig
730 -------------------------------
732 Under sysconfig, configuring a device with BOOTPROTO='dhcp'
733 will cause it to query DHCP for its IP address information. At this
734 writing, this does not function for bonding devices; the scripts
735 attempt to obtain the device address from DHCP prior to adding any of
736 the slave devices. Without active slaves, the DHCP requests are not
739 3.1.2 Configuring Multiple Bonds with Sysconfig
740 -----------------------------------------------
742 The sysconfig network initialization system is capable of
743 handling multiple bonding devices. All that is necessary is for each
744 bonding instance to have an appropriately configured ifcfg-bondX file
745 (as described above). Do not specify the "max_bonds" parameter to any
746 instance of bonding, as this will confuse sysconfig. If you require
747 multiple bonding devices with identical parameters, create multiple
750 Because the sysconfig scripts supply the bonding module
751 options in the ifcfg-bondX file, it is not necessary to add them to
752 the system /etc/modules.conf or /etc/modprobe.conf configuration file.
754 3.2 Configuration with Initscripts Support
755 ------------------------------------------
757 This section applies to distros using a version of initscripts
758 with bonding support, for example, Red Hat Linux 9 or Red Hat
759 Enterprise Linux version 3 or 4. On these systems, the network
760 initialization scripts have some knowledge of bonding, and can be
761 configured to control bonding devices.
763 These distros will not automatically load the network adapter
764 driver unless the ethX device is configured with an IP address.
765 Because of this constraint, users must manually configure a
766 network-script file for all physical adapters that will be members of
767 a bondX link. Network script files are located in the directory:
769 /etc/sysconfig/network-scripts
771 The file name must be prefixed with "ifcfg-eth" and suffixed
772 with the adapter's physical adapter number. For example, the script
773 for eth0 would be named /etc/sysconfig/network-scripts/ifcfg-eth0.
774 Place the following text in the file:
783 The DEVICE= line will be different for every ethX device and
784 must correspond with the name of the file, i.e., ifcfg-eth1 must have
785 a device line of DEVICE=eth1. The setting of the MASTER= line will
786 also depend on the final bonding interface name chosen for your bond.
787 As with other network devices, these typically start at 0, and go up
788 one for each device, i.e., the first bonding instance is bond0, the
789 second is bond1, and so on.
791 Next, create a bond network script. The file name for this
792 script will be /etc/sysconfig/network-scripts/ifcfg-bondX where X is
793 the number of the bond. For bond0 the file is named "ifcfg-bond0",
794 for bond1 it is named "ifcfg-bond1", and so on. Within that file,
795 place the following text:
799 NETMASK=255.255.255.0
801 BROADCAST=192.168.1.255
806 Be sure to change the networking specific lines (IPADDR,
807 NETMASK, NETWORK and BROADCAST) to match your network configuration.
809 Finally, it is necessary to edit /etc/modules.conf (or
810 /etc/modprobe.conf, depending upon your distro) to load the bonding
811 module with your desired options when the bond0 interface is brought
812 up. The following lines in /etc/modules.conf (or modprobe.conf) will
813 load the bonding module, and select its options:
816 options bond0 mode=balance-alb miimon=100
818 Replace the sample parameters with the appropriate set of
819 options for your configuration.
821 Finally run "/etc/rc.d/init.d/network restart" as root. This
822 will restart the networking subsystem and your bond link should be now
825 3.2.1 Using DHCP with Initscripts
826 ---------------------------------
828 Recent versions of initscripts (the version supplied with
829 Fedora Core 3 and Red Hat Enterprise Linux 4 is reported to work) do
830 have support for assigning IP information to bonding devices via DHCP.
832 To configure bonding for DHCP, configure it as described
833 above, except replace the line "BOOTPROTO=none" with "BOOTPROTO=dhcp"
834 and add a line consisting of "TYPE=Bonding". Note that the TYPE value
837 3.2.2 Configuring Multiple Bonds with Initscripts
838 -------------------------------------------------
840 At this writing, the initscripts package does not directly
841 support loading the bonding driver multiple times, so the process for
842 doing so is the same as described in the "Configuring Multiple Bonds
843 Manually" section, below.
845 NOTE: It has been observed that some Red Hat supplied kernels
846 are apparently unable to rename modules at load time (the "-o bond1"
847 part). Attempts to pass that option to modprobe will produce an
848 "Operation not permitted" error. This has been reported on some
849 Fedora Core kernels, and has been seen on RHEL 4 as well. On kernels
850 exhibiting this problem, it will be impossible to configure multiple
851 bonds with differing parameters.
853 3.3 Configuring Bonding Manually with Ifenslave
854 -----------------------------------------------
856 This section applies to distros whose network initialization
857 scripts (the sysconfig or initscripts package) do not have specific
858 knowledge of bonding. One such distro is SuSE Linux Enterprise Server
861 The general method for these systems is to place the bonding
862 module parameters into /etc/modules.conf or /etc/modprobe.conf (as
863 appropriate for the installed distro), then add modprobe and/or
864 ifenslave commands to the system's global init script. The name of
865 the global init script differs; for sysconfig, it is
866 /etc/init.d/boot.local and for initscripts it is /etc/rc.d/rc.local.
868 For example, if you wanted to make a simple bond of two e100
869 devices (presumed to be eth0 and eth1), and have it persist across
870 reboots, edit the appropriate file (/etc/init.d/boot.local or
871 /etc/rc.d/rc.local), and add the following:
873 modprobe bonding mode=balance-alb miimon=100
875 ifconfig bond0 192.168.1.1 netmask 255.255.255.0 up
879 Replace the example bonding module parameters and bond0
880 network configuration (IP address, netmask, etc) with the appropriate
881 values for your configuration.
883 Unfortunately, this method will not provide support for the
884 ifup and ifdown scripts on the bond devices. To reload the bonding
885 configuration, it is necessary to run the initialization script, e.g.,
887 # /etc/init.d/boot.local
893 It may be desirable in such a case to create a separate script
894 which only initializes the bonding configuration, then call that
895 separate script from within boot.local. This allows for bonding to be
896 enabled without re-running the entire global init script.
898 To shut down the bonding devices, it is necessary to first
899 mark the bonding device itself as being down, then remove the
900 appropriate device driver modules. For our example above, you can do
903 # ifconfig bond0 down
907 Again, for convenience, it may be desirable to create a script
911 3.3.1 Configuring Multiple Bonds Manually
912 -----------------------------------------
914 This section contains information on configuring multiple
915 bonding devices with differing options for those systems whose network
916 initialization scripts lack support for configuring multiple bonds.
918 If you require multiple bonding devices, but all with the same
919 options, you may wish to use the "max_bonds" module parameter,
922 To create multiple bonding devices with differing options, it
923 is necessary to load the bonding driver multiple times. Note that
924 current versions of the sysconfig network initialization scripts
925 handle this automatically; if your distro uses these scripts, no
926 special action is needed. See the section Configuring Bonding
927 Devices, above, if you're not sure about your network initialization
930 To load multiple instances of the module, it is necessary to
931 specify a different name for each instance (the module loading system
932 requires that every loaded module, even multiple instances of the same
933 module, have a unique name). This is accomplished by supplying
934 multiple sets of bonding options in /etc/modprobe.conf, for example:
937 options bond0 -o bond0 mode=balance-rr miimon=100
940 options bond1 -o bond1 mode=balance-alb miimon=50
942 will load the bonding module two times. The first instance is
943 named "bond0" and creates the bond0 device in balance-rr mode with an
944 miimon of 100. The second instance is named "bond1" and creates the
945 bond1 device in balance-alb mode with an miimon of 50.
947 In some circumstances (typically with older distributions),
948 the above does not work, and the second bonding instance never sees
949 its options. In that case, the second options line can be substituted
952 install bond1 /sbin/modprobe --ignore-install bonding -o bond1 \
953 mode=balance-alb miimon=50
955 This may be repeated any number of times, specifying a new and
956 unique name in place of bond1 for each subsequent instance.
958 3.4 Configuring Bonding Manually via Sysfs
959 ------------------------------------------
961 Starting with version 3.0, Channel Bonding may be configured
962 via the sysfs interface. This interface allows dynamic configuration
963 of all bonds in the system without unloading the module. It also
964 allows for adding and removing bonds at runtime. Ifenslave is no
965 longer required, though it is still supported.
967 Use of the sysfs interface allows you to use multiple bonds
968 with different configurations without having to reload the module.
969 It also allows you to use multiple, differently configured bonds when
970 bonding is compiled into the kernel.
972 You must have the sysfs filesystem mounted to configure
973 bonding this way. The examples in this document assume that you
974 are using the standard mount point for sysfs, e.g. /sys. If your
975 sysfs filesystem is mounted elsewhere, you will need to adjust the
976 example paths accordingly.
978 Creating and Destroying Bonds
979 -----------------------------
980 To add a new bond foo:
981 # echo +foo > /sys/class/net/bonding_masters
983 To remove an existing bond bar:
984 # echo -bar > /sys/class/net/bonding_masters
986 To show all existing bonds:
987 # cat /sys/class/net/bonding_masters
989 NOTE: due to 4K size limitation of sysfs files, this list may be
990 truncated if you have more than a few hundred bonds. This is unlikely
991 to occur under normal operating conditions.
993 Adding and Removing Slaves
994 --------------------------
995 Interfaces may be enslaved to a bond using the file
996 /sys/class/net/<bond>/bonding/slaves. The semantics for this file
997 are the same as for the bonding_masters file.
999 To enslave interface eth0 to bond bond0:
1001 # echo +eth0 > /sys/class/net/bond0/bonding/slaves
1003 To free slave eth0 from bond bond0:
1004 # echo -eth0 > /sys/class/net/bond0/bonding/slaves
1006 NOTE: The bond must be up before slaves can be added. All
1007 slaves are freed when the interface is brought down.
1009 When an interface is enslaved to a bond, symlinks between the
1010 two are created in the sysfs filesystem. In this case, you would get
1011 /sys/class/net/bond0/slave_eth0 pointing to /sys/class/net/eth0, and
1012 /sys/class/net/eth0/master pointing to /sys/class/net/bond0.
1014 This means that you can tell quickly whether or not an
1015 interface is enslaved by looking for the master symlink. Thus:
1016 # echo -eth0 > /sys/class/net/eth0/master/bonding/slaves
1017 will free eth0 from whatever bond it is enslaved to, regardless of
1018 the name of the bond interface.
1020 Changing a Bond's Configuration
1021 -------------------------------
1022 Each bond may be configured individually by manipulating the
1023 files located in /sys/class/net/<bond name>/bonding
1025 The names of these files correspond directly with the command-
1026 line parameters described elsewhere in in this file, and, with the
1027 exception of arp_ip_target, they accept the same values. To see the
1028 current setting, simply cat the appropriate file.
1030 A few examples will be given here; for specific usage
1031 guidelines for each parameter, see the appropriate section in this
1034 To configure bond0 for balance-alb mode:
1035 # ifconfig bond0 down
1036 # echo 6 > /sys/class/net/bond0/bonding/mode
1038 # echo balance-alb > /sys/class/net/bond0/bonding/mode
1039 NOTE: The bond interface must be down before the mode can be
1042 To enable MII monitoring on bond0 with a 1 second interval:
1043 # echo 1000 > /sys/class/net/bond0/bonding/miimon
1044 NOTE: If ARP monitoring is enabled, it will disabled when MII
1045 monitoring is enabled, and vice-versa.
1048 # echo +192.168.0.100 > /sys/class/net/bond0/bonding/arp_ip_target
1049 # echo +192.168.0.101 > /sys/class/net/bond0/bonding/arp_ip_target
1050 NOTE: up to 10 target addresses may be specified.
1052 To remove an ARP target:
1053 # echo -192.168.0.100 > /sys/class/net/bond0/bonding/arp_ip_target
1055 Example Configuration
1056 ---------------------
1057 We begin with the same example that is shown in section 3.3,
1058 executed with sysfs, and without using ifenslave.
1060 To make a simple bond of two e100 devices (presumed to be eth0
1061 and eth1), and have it persist across reboots, edit the appropriate
1062 file (/etc/init.d/boot.local or /etc/rc.d/rc.local), and add the
1067 echo balance-alb > /sys/class/net/bond0/bonding/mode
1068 ifconfig bond0 192.168.1.1 netmask 255.255.255.0 up
1069 echo 100 > /sys/class/net/bond0/bonding/miimon
1070 echo +eth0 > /sys/class/net/bond0/bonding/slaves
1071 echo +eth1 > /sys/class/net/bond0/bonding/slaves
1073 To add a second bond, with two e1000 interfaces in
1074 active-backup mode, using ARP monitoring, add the following lines to
1078 echo +bond1 > /sys/class/net/bonding_masters
1079 echo active-backup > /sys/class/net/bond1/bonding/mode
1080 ifconfig bond1 192.168.2.1 netmask 255.255.255.0 up
1081 echo +192.168.2.100 /sys/class/net/bond1/bonding/arp_ip_target
1082 echo 2000 > /sys/class/net/bond1/bonding/arp_interval
1083 echo +eth2 > /sys/class/net/bond1/bonding/slaves
1084 echo +eth3 > /sys/class/net/bond1/bonding/slaves
1087 4. Querying Bonding Configuration
1088 =================================
1090 4.1 Bonding Configuration
1091 -------------------------
1093 Each bonding device has a read-only file residing in the
1094 /proc/net/bonding directory. The file contents include information
1095 about the bonding configuration, options and state of each slave.
1097 For example, the contents of /proc/net/bonding/bond0 after the
1098 driver is loaded with parameters of mode=0 and miimon=1000 is
1099 generally as follows:
1101 Ethernet Channel Bonding Driver: 2.6.1 (October 29, 2004)
1102 Bonding Mode: load balancing (round-robin)
1103 Currently Active Slave: eth0
1105 MII Polling Interval (ms): 1000
1109 Slave Interface: eth1
1111 Link Failure Count: 1
1113 Slave Interface: eth0
1115 Link Failure Count: 1
1117 The precise format and contents will change depending upon the
1118 bonding configuration, state, and version of the bonding driver.
1120 4.2 Network configuration
1121 -------------------------
1123 The network configuration can be inspected using the ifconfig
1124 command. Bonding devices will have the MASTER flag set; Bonding slave
1125 devices will have the SLAVE flag set. The ifconfig output does not
1126 contain information on which slaves are associated with which masters.
1128 In the example below, the bond0 interface is the master
1129 (MASTER) while eth0 and eth1 are slaves (SLAVE). Notice all slaves of
1130 bond0 have the same MAC address (HWaddr) as bond0 for all modes except
1131 TLB and ALB that require a unique MAC address for each slave.
1134 bond0 Link encap:Ethernet HWaddr 00:C0:F0:1F:37:B4
1135 inet addr:XXX.XXX.XXX.YYY Bcast:XXX.XXX.XXX.255 Mask:255.255.252.0
1136 UP BROADCAST RUNNING MASTER MULTICAST MTU:1500 Metric:1
1137 RX packets:7224794 errors:0 dropped:0 overruns:0 frame:0
1138 TX packets:3286647 errors:1 dropped:0 overruns:1 carrier:0
1139 collisions:0 txqueuelen:0
1141 eth0 Link encap:Ethernet HWaddr 00:C0:F0:1F:37:B4
1142 UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
1143 RX packets:3573025 errors:0 dropped:0 overruns:0 frame:0
1144 TX packets:1643167 errors:1 dropped:0 overruns:1 carrier:0
1145 collisions:0 txqueuelen:100
1146 Interrupt:10 Base address:0x1080
1148 eth1 Link encap:Ethernet HWaddr 00:C0:F0:1F:37:B4
1149 UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
1150 RX packets:3651769 errors:0 dropped:0 overruns:0 frame:0
1151 TX packets:1643480 errors:0 dropped:0 overruns:0 carrier:0
1152 collisions:0 txqueuelen:100
1153 Interrupt:9 Base address:0x1400
1155 5. Switch Configuration
1156 =======================
1158 For this section, "switch" refers to whatever system the
1159 bonded devices are directly connected to (i.e., where the other end of
1160 the cable plugs into). This may be an actual dedicated switch device,
1161 or it may be another regular system (e.g., another computer running
1164 The active-backup, balance-tlb and balance-alb modes do not
1165 require any specific configuration of the switch.
1167 The 802.3ad mode requires that the switch have the appropriate
1168 ports configured as an 802.3ad aggregation. The precise method used
1169 to configure this varies from switch to switch, but, for example, a
1170 Cisco 3550 series switch requires that the appropriate ports first be
1171 grouped together in a single etherchannel instance, then that
1172 etherchannel is set to mode "lacp" to enable 802.3ad (instead of
1173 standard EtherChannel).
1175 The balance-rr, balance-xor and broadcast modes generally
1176 require that the switch have the appropriate ports grouped together.
1177 The nomenclature for such a group differs between switches, it may be
1178 called an "etherchannel" (as in the Cisco example, above), a "trunk
1179 group" or some other similar variation. For these modes, each switch
1180 will also have its own configuration options for the switch's transmit
1181 policy to the bond. Typical choices include XOR of either the MAC or
1182 IP addresses. The transmit policy of the two peers does not need to
1183 match. For these three modes, the bonding mode really selects a
1184 transmit policy for an EtherChannel group; all three will interoperate
1185 with another EtherChannel group.
1188 6. 802.1q VLAN Support
1189 ======================
1191 It is possible to configure VLAN devices over a bond interface
1192 using the 8021q driver. However, only packets coming from the 8021q
1193 driver and passing through bonding will be tagged by default. Self
1194 generated packets, for example, bonding's learning packets or ARP
1195 packets generated by either ALB mode or the ARP monitor mechanism, are
1196 tagged internally by bonding itself. As a result, bonding must
1197 "learn" the VLAN IDs configured above it, and use those IDs to tag
1198 self generated packets.
1200 For reasons of simplicity, and to support the use of adapters
1201 that can do VLAN hardware acceleration offloading, the bonding
1202 interface declares itself as fully hardware offloading capable, it gets
1203 the add_vid/kill_vid notifications to gather the necessary
1204 information, and it propagates those actions to the slaves. In case
1205 of mixed adapter types, hardware accelerated tagged packets that
1206 should go through an adapter that is not offloading capable are
1207 "un-accelerated" by the bonding driver so the VLAN tag sits in the
1210 VLAN interfaces *must* be added on top of a bonding interface
1211 only after enslaving at least one slave. The bonding interface has a
1212 hardware address of 00:00:00:00:00:00 until the first slave is added.
1213 If the VLAN interface is created prior to the first enslavement, it
1214 would pick up the all-zeroes hardware address. Once the first slave
1215 is attached to the bond, the bond device itself will pick up the
1216 slave's hardware address, which is then available for the VLAN device.
1218 Also, be aware that a similar problem can occur if all slaves
1219 are released from a bond that still has one or more VLAN interfaces on
1220 top of it. When a new slave is added, the bonding interface will
1221 obtain its hardware address from the first slave, which might not
1222 match the hardware address of the VLAN interfaces (which was
1223 ultimately copied from an earlier slave).
1225 There are two methods to insure that the VLAN device operates
1226 with the correct hardware address if all slaves are removed from a
1229 1. Remove all VLAN interfaces then recreate them
1231 2. Set the bonding interface's hardware address so that it
1232 matches the hardware address of the VLAN interfaces.
1234 Note that changing a VLAN interface's HW address would set the
1235 underlying device -- i.e. the bonding interface -- to promiscuous
1236 mode, which might not be what you want.
1242 The bonding driver at present supports two schemes for
1243 monitoring a slave device's link state: the ARP monitor and the MII
1246 At the present time, due to implementation restrictions in the
1247 bonding driver itself, it is not possible to enable both ARP and MII
1248 monitoring simultaneously.
1250 7.1 ARP Monitor Operation
1251 -------------------------
1253 The ARP monitor operates as its name suggests: it sends ARP
1254 queries to one or more designated peer systems on the network, and
1255 uses the response as an indication that the link is operating. This
1256 gives some assurance that traffic is actually flowing to and from one
1257 or more peers on the local network.
1259 The ARP monitor relies on the device driver itself to verify
1260 that traffic is flowing. In particular, the driver must keep up to
1261 date the last receive time, dev->last_rx, and transmit start time,
1262 dev->trans_start. If these are not updated by the driver, then the
1263 ARP monitor will immediately fail any slaves using that driver, and
1264 those slaves will stay down. If networking monitoring (tcpdump, etc)
1265 shows the ARP requests and replies on the network, then it may be that
1266 your device driver is not updating last_rx and trans_start.
1268 7.2 Configuring Multiple ARP Targets
1269 ------------------------------------
1271 While ARP monitoring can be done with just one target, it can
1272 be useful in a High Availability setup to have several targets to
1273 monitor. In the case of just one target, the target itself may go
1274 down or have a problem making it unresponsive to ARP requests. Having
1275 an additional target (or several) increases the reliability of the ARP
1278 Multiple ARP targets must be separated by commas as follows:
1280 # example options for ARP monitoring with three targets
1282 options bond0 arp_interval=60 arp_ip_target=192.168.0.1,192.168.0.3,192.168.0.9
1284 For just a single target the options would resemble:
1286 # example options for ARP monitoring with one target
1288 options bond0 arp_interval=60 arp_ip_target=192.168.0.100
1291 7.3 MII Monitor Operation
1292 -------------------------
1294 The MII monitor monitors only the carrier state of the local
1295 network interface. It accomplishes this in one of three ways: by
1296 depending upon the device driver to maintain its carrier state, by
1297 querying the device's MII registers, or by making an ethtool query to
1300 If the use_carrier module parameter is 1 (the default value),
1301 then the MII monitor will rely on the driver for carrier state
1302 information (via the netif_carrier subsystem). As explained in the
1303 use_carrier parameter information, above, if the MII monitor fails to
1304 detect carrier loss on the device (e.g., when the cable is physically
1305 disconnected), it may be that the driver does not support
1308 If use_carrier is 0, then the MII monitor will first query the
1309 device's (via ioctl) MII registers and check the link state. If that
1310 request fails (not just that it returns carrier down), then the MII
1311 monitor will make an ethtool ETHOOL_GLINK request to attempt to obtain
1312 the same information. If both methods fail (i.e., the driver either
1313 does not support or had some error in processing both the MII register
1314 and ethtool requests), then the MII monitor will assume the link is
1317 8. Potential Sources of Trouble
1318 ===============================
1320 8.1 Adventures in Routing
1321 -------------------------
1323 When bonding is configured, it is important that the slave
1324 devices not have routes that supersede routes of the master (or,
1325 generally, not have routes at all). For example, suppose the bonding
1326 device bond0 has two slaves, eth0 and eth1, and the routing table is
1329 Kernel IP routing table
1330 Destination Gateway Genmask Flags MSS Window irtt Iface
1331 10.0.0.0 0.0.0.0 255.255.0.0 U 40 0 0 eth0
1332 10.0.0.0 0.0.0.0 255.255.0.0 U 40 0 0 eth1
1333 10.0.0.0 0.0.0.0 255.255.0.0 U 40 0 0 bond0
1334 127.0.0.0 0.0.0.0 255.0.0.0 U 40 0 0 lo
1336 This routing configuration will likely still update the
1337 receive/transmit times in the driver (needed by the ARP monitor), but
1338 may bypass the bonding driver (because outgoing traffic to, in this
1339 case, another host on network 10 would use eth0 or eth1 before bond0).
1341 The ARP monitor (and ARP itself) may become confused by this
1342 configuration, because ARP requests (generated by the ARP monitor)
1343 will be sent on one interface (bond0), but the corresponding reply
1344 will arrive on a different interface (eth0). This reply looks to ARP
1345 as an unsolicited ARP reply (because ARP matches replies on an
1346 interface basis), and is discarded. The MII monitor is not affected
1347 by the state of the routing table.
1349 The solution here is simply to insure that slaves do not have
1350 routes of their own, and if for some reason they must, those routes do
1351 not supersede routes of their master. This should generally be the
1352 case, but unusual configurations or errant manual or automatic static
1353 route additions may cause trouble.
1355 8.2 Ethernet Device Renaming
1356 ----------------------------
1358 On systems with network configuration scripts that do not
1359 associate physical devices directly with network interface names (so
1360 that the same physical device always has the same "ethX" name), it may
1361 be necessary to add some special logic to either /etc/modules.conf or
1362 /etc/modprobe.conf (depending upon which is installed on the system).
1364 For example, given a modules.conf containing the following:
1367 options bond0 mode=some-mode miimon=50
1373 If neither eth0 and eth1 are slaves to bond0, then when the
1374 bond0 interface comes up, the devices may end up reordered. This
1375 happens because bonding is loaded first, then its slave device's
1376 drivers are loaded next. Since no other drivers have been loaded,
1377 when the e1000 driver loads, it will receive eth0 and eth1 for its
1378 devices, but the bonding configuration tries to enslave eth2 and eth3
1379 (which may later be assigned to the tg3 devices).
1381 Adding the following:
1383 add above bonding e1000 tg3
1385 causes modprobe to load e1000 then tg3, in that order, when
1386 bonding is loaded. This command is fully documented in the
1387 modules.conf manual page.
1389 On systems utilizing modprobe.conf (or modprobe.conf.local),
1390 an equivalent problem can occur. In this case, the following can be
1391 added to modprobe.conf (or modprobe.conf.local, as appropriate), as
1392 follows (all on one line; it has been split here for clarity):
1394 install bonding /sbin/modprobe tg3; /sbin/modprobe e1000;
1395 /sbin/modprobe --ignore-install bonding
1397 This will, when loading the bonding module, rather than
1398 performing the normal action, instead execute the provided command.
1399 This command loads the device drivers in the order needed, then calls
1400 modprobe with --ignore-install to cause the normal action to then take
1401 place. Full documentation on this can be found in the modprobe.conf
1402 and modprobe manual pages.
1404 8.3. Painfully Slow Or No Failed Link Detection By Miimon
1405 ---------------------------------------------------------
1407 By default, bonding enables the use_carrier option, which
1408 instructs bonding to trust the driver to maintain carrier state.
1410 As discussed in the options section, above, some drivers do
1411 not support the netif_carrier_on/_off link state tracking system.
1412 With use_carrier enabled, bonding will always see these links as up,
1413 regardless of their actual state.
1415 Additionally, other drivers do support netif_carrier, but do
1416 not maintain it in real time, e.g., only polling the link state at
1417 some fixed interval. In this case, miimon will detect failures, but
1418 only after some long period of time has expired. If it appears that
1419 miimon is very slow in detecting link failures, try specifying
1420 use_carrier=0 to see if that improves the failure detection time. If
1421 it does, then it may be that the driver checks the carrier state at a
1422 fixed interval, but does not cache the MII register values (so the
1423 use_carrier=0 method of querying the registers directly works). If
1424 use_carrier=0 does not improve the failover, then the driver may cache
1425 the registers, or the problem may be elsewhere.
1427 Also, remember that miimon only checks for the device's
1428 carrier state. It has no way to determine the state of devices on or
1429 beyond other ports of a switch, or if a switch is refusing to pass
1430 traffic while still maintaining carrier on.
1435 If running SNMP agents, the bonding driver should be loaded
1436 before any network drivers participating in a bond. This requirement
1437 is due to the interface index (ipAdEntIfIndex) being associated to
1438 the first interface found with a given IP address. That is, there is
1439 only one ipAdEntIfIndex for each IP address. For example, if eth0 and
1440 eth1 are slaves of bond0 and the driver for eth0 is loaded before the
1441 bonding driver, the interface for the IP address will be associated
1442 with the eth0 interface. This configuration is shown below, the IP
1443 address 192.168.1.1 has an interface index of 2 which indexes to eth0
1444 in the ifDescr table (ifDescr.2).
1446 interfaces.ifTable.ifEntry.ifDescr.1 = lo
1447 interfaces.ifTable.ifEntry.ifDescr.2 = eth0
1448 interfaces.ifTable.ifEntry.ifDescr.3 = eth1
1449 interfaces.ifTable.ifEntry.ifDescr.4 = eth2
1450 interfaces.ifTable.ifEntry.ifDescr.5 = eth3
1451 interfaces.ifTable.ifEntry.ifDescr.6 = bond0
1452 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.10.10.10 = 5
1453 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.192.168.1.1 = 2
1454 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.74.20.94 = 4
1455 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.127.0.0.1 = 1
1457 This problem is avoided by loading the bonding driver before
1458 any network drivers participating in a bond. Below is an example of
1459 loading the bonding driver first, the IP address 192.168.1.1 is
1460 correctly associated with ifDescr.2.
1462 interfaces.ifTable.ifEntry.ifDescr.1 = lo
1463 interfaces.ifTable.ifEntry.ifDescr.2 = bond0
1464 interfaces.ifTable.ifEntry.ifDescr.3 = eth0
1465 interfaces.ifTable.ifEntry.ifDescr.4 = eth1
1466 interfaces.ifTable.ifEntry.ifDescr.5 = eth2
1467 interfaces.ifTable.ifEntry.ifDescr.6 = eth3
1468 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.10.10.10 = 6
1469 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.192.168.1.1 = 2
1470 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.74.20.94 = 5
1471 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.127.0.0.1 = 1
1473 While some distributions may not report the interface name in
1474 ifDescr, the association between the IP address and IfIndex remains
1475 and SNMP functions such as Interface_Scan_Next will report that
1478 10. Promiscuous mode
1479 ====================
1481 When running network monitoring tools, e.g., tcpdump, it is
1482 common to enable promiscuous mode on the device, so that all traffic
1483 is seen (instead of seeing only traffic destined for the local host).
1484 The bonding driver handles promiscuous mode changes to the bonding
1485 master device (e.g., bond0), and propagates the setting to the slave
1488 For the balance-rr, balance-xor, broadcast, and 802.3ad modes,
1489 the promiscuous mode setting is propagated to all slaves.
1491 For the active-backup, balance-tlb and balance-alb modes, the
1492 promiscuous mode setting is propagated only to the active slave.
1494 For balance-tlb mode, the active slave is the slave currently
1495 receiving inbound traffic.
1497 For balance-alb mode, the active slave is the slave used as a
1498 "primary." This slave is used for mode-specific control traffic, for
1499 sending to peers that are unassigned or if the load is unbalanced.
1501 For the active-backup, balance-tlb and balance-alb modes, when
1502 the active slave changes (e.g., due to a link failure), the
1503 promiscuous setting will be propagated to the new active slave.
1505 11. Configuring Bonding for High Availability
1506 =============================================
1508 High Availability refers to configurations that provide
1509 maximum network availability by having redundant or backup devices,
1510 links or switches between the host and the rest of the world. The
1511 goal is to provide the maximum availability of network connectivity
1512 (i.e., the network always works), even though other configurations
1513 could provide higher throughput.
1515 11.1 High Availability in a Single Switch Topology
1516 --------------------------------------------------
1518 If two hosts (or a host and a single switch) are directly
1519 connected via multiple physical links, then there is no availability
1520 penalty to optimizing for maximum bandwidth. In this case, there is
1521 only one switch (or peer), so if it fails, there is no alternative
1522 access to fail over to. Additionally, the bonding load balance modes
1523 support link monitoring of their members, so if individual links fail,
1524 the load will be rebalanced across the remaining devices.
1526 See Section 13, "Configuring Bonding for Maximum Throughput"
1527 for information on configuring bonding with one peer device.
1529 11.2 High Availability in a Multiple Switch Topology
1530 ----------------------------------------------------
1532 With multiple switches, the configuration of bonding and the
1533 network changes dramatically. In multiple switch topologies, there is
1534 a trade off between network availability and usable bandwidth.
1536 Below is a sample network, configured to maximize the
1537 availability of the network:
1541 +-----+----+ +-----+----+
1542 | |port2 ISL port2| |
1543 | switch A +--------------------------+ switch B |
1545 +-----+----+ +-----++---+
1548 +-------------+ host1 +---------------+
1551 In this configuration, there is a link between the two
1552 switches (ISL, or inter switch link), and multiple ports connecting to
1553 the outside world ("port3" on each switch). There is no technical
1554 reason that this could not be extended to a third switch.
1556 11.2.1 HA Bonding Mode Selection for Multiple Switch Topology
1557 -------------------------------------------------------------
1559 In a topology such as the example above, the active-backup and
1560 broadcast modes are the only useful bonding modes when optimizing for
1561 availability; the other modes require all links to terminate on the
1562 same peer for them to behave rationally.
1564 active-backup: This is generally the preferred mode, particularly if
1565 the switches have an ISL and play together well. If the
1566 network configuration is such that one switch is specifically
1567 a backup switch (e.g., has lower capacity, higher cost, etc),
1568 then the primary option can be used to insure that the
1569 preferred link is always used when it is available.
1571 broadcast: This mode is really a special purpose mode, and is suitable
1572 only for very specific needs. For example, if the two
1573 switches are not connected (no ISL), and the networks beyond
1574 them are totally independent. In this case, if it is
1575 necessary for some specific one-way traffic to reach both
1576 independent networks, then the broadcast mode may be suitable.
1578 11.2.2 HA Link Monitoring Selection for Multiple Switch Topology
1579 ----------------------------------------------------------------
1581 The choice of link monitoring ultimately depends upon your
1582 switch. If the switch can reliably fail ports in response to other
1583 failures, then either the MII or ARP monitors should work. For
1584 example, in the above example, if the "port3" link fails at the remote
1585 end, the MII monitor has no direct means to detect this. The ARP
1586 monitor could be configured with a target at the remote end of port3,
1587 thus detecting that failure without switch support.
1589 In general, however, in a multiple switch topology, the ARP
1590 monitor can provide a higher level of reliability in detecting end to
1591 end connectivity failures (which may be caused by the failure of any
1592 individual component to pass traffic for any reason). Additionally,
1593 the ARP monitor should be configured with multiple targets (at least
1594 one for each switch in the network). This will insure that,
1595 regardless of which switch is active, the ARP monitor has a suitable
1599 12. Configuring Bonding for Maximum Throughput
1600 ==============================================
1602 12.1 Maximizing Throughput in a Single Switch Topology
1603 ------------------------------------------------------
1605 In a single switch configuration, the best method to maximize
1606 throughput depends upon the application and network environment. The
1607 various load balancing modes each have strengths and weaknesses in
1608 different environments, as detailed below.
1610 For this discussion, we will break down the topologies into
1611 two categories. Depending upon the destination of most traffic, we
1612 categorize them into either "gatewayed" or "local" configurations.
1614 In a gatewayed configuration, the "switch" is acting primarily
1615 as a router, and the majority of traffic passes through this router to
1616 other networks. An example would be the following:
1619 +----------+ +----------+
1620 | |eth0 port1| | to other networks
1621 | Host A +---------------------+ router +------------------->
1622 | +---------------------+ | Hosts B and C are out
1623 | |eth1 port2| | here somewhere
1624 +----------+ +----------+
1626 The router may be a dedicated router device, or another host
1627 acting as a gateway. For our discussion, the important point is that
1628 the majority of traffic from Host A will pass through the router to
1629 some other network before reaching its final destination.
1631 In a gatewayed network configuration, although Host A may
1632 communicate with many other systems, all of its traffic will be sent
1633 and received via one other peer on the local network, the router.
1635 Note that the case of two systems connected directly via
1636 multiple physical links is, for purposes of configuring bonding, the
1637 same as a gatewayed configuration. In that case, it happens that all
1638 traffic is destined for the "gateway" itself, not some other network
1641 In a local configuration, the "switch" is acting primarily as
1642 a switch, and the majority of traffic passes through this switch to
1643 reach other stations on the same network. An example would be the
1646 +----------+ +----------+ +--------+
1647 | |eth0 port1| +-------+ Host B |
1648 | Host A +------------+ switch |port3 +--------+
1649 | +------------+ | +--------+
1650 | |eth1 port2| +------------------+ Host C |
1651 +----------+ +----------+port4 +--------+
1654 Again, the switch may be a dedicated switch device, or another
1655 host acting as a gateway. For our discussion, the important point is
1656 that the majority of traffic from Host A is destined for other hosts
1657 on the same local network (Hosts B and C in the above example).
1659 In summary, in a gatewayed configuration, traffic to and from
1660 the bonded device will be to the same MAC level peer on the network
1661 (the gateway itself, i.e., the router), regardless of its final
1662 destination. In a local configuration, traffic flows directly to and
1663 from the final destinations, thus, each destination (Host B, Host C)
1664 will be addressed directly by their individual MAC addresses.
1666 This distinction between a gatewayed and a local network
1667 configuration is important because many of the load balancing modes
1668 available use the MAC addresses of the local network source and
1669 destination to make load balancing decisions. The behavior of each
1670 mode is described below.
1673 12.1.1 MT Bonding Mode Selection for Single Switch Topology
1674 -----------------------------------------------------------
1676 This configuration is the easiest to set up and to understand,
1677 although you will have to decide which bonding mode best suits your
1678 needs. The trade offs for each mode are detailed below:
1680 balance-rr: This mode is the only mode that will permit a single
1681 TCP/IP connection to stripe traffic across multiple
1682 interfaces. It is therefore the only mode that will allow a
1683 single TCP/IP stream to utilize more than one interface's
1684 worth of throughput. This comes at a cost, however: the
1685 striping often results in peer systems receiving packets out
1686 of order, causing TCP/IP's congestion control system to kick
1687 in, often by retransmitting segments.
1689 It is possible to adjust TCP/IP's congestion limits by
1690 altering the net.ipv4.tcp_reordering sysctl parameter. The
1691 usual default value is 3, and the maximum useful value is 127.
1692 For a four interface balance-rr bond, expect that a single
1693 TCP/IP stream will utilize no more than approximately 2.3
1694 interface's worth of throughput, even after adjusting
1697 Note that this out of order delivery occurs when both the
1698 sending and receiving systems are utilizing a multiple
1699 interface bond. Consider a configuration in which a
1700 balance-rr bond feeds into a single higher capacity network
1701 channel (e.g., multiple 100Mb/sec ethernets feeding a single
1702 gigabit ethernet via an etherchannel capable switch). In this
1703 configuration, traffic sent from the multiple 100Mb devices to
1704 a destination connected to the gigabit device will not see
1705 packets out of order. However, traffic sent from the gigabit
1706 device to the multiple 100Mb devices may or may not see
1707 traffic out of order, depending upon the balance policy of the
1708 switch. Many switches do not support any modes that stripe
1709 traffic (instead choosing a port based upon IP or MAC level
1710 addresses); for those devices, traffic flowing from the
1711 gigabit device to the many 100Mb devices will only utilize one
1714 If you are utilizing protocols other than TCP/IP, UDP for
1715 example, and your application can tolerate out of order
1716 delivery, then this mode can allow for single stream datagram
1717 performance that scales near linearly as interfaces are added
1720 This mode requires the switch to have the appropriate ports
1721 configured for "etherchannel" or "trunking."
1723 active-backup: There is not much advantage in this network topology to
1724 the active-backup mode, as the inactive backup devices are all
1725 connected to the same peer as the primary. In this case, a
1726 load balancing mode (with link monitoring) will provide the
1727 same level of network availability, but with increased
1728 available bandwidth. On the plus side, active-backup mode
1729 does not require any configuration of the switch, so it may
1730 have value if the hardware available does not support any of
1731 the load balance modes.
1733 balance-xor: This mode will limit traffic such that packets destined
1734 for specific peers will always be sent over the same
1735 interface. Since the destination is determined by the MAC
1736 addresses involved, this mode works best in a "local" network
1737 configuration (as described above), with destinations all on
1738 the same local network. This mode is likely to be suboptimal
1739 if all your traffic is passed through a single router (i.e., a
1740 "gatewayed" network configuration, as described above).
1742 As with balance-rr, the switch ports need to be configured for
1743 "etherchannel" or "trunking."
1745 broadcast: Like active-backup, there is not much advantage to this
1746 mode in this type of network topology.
1748 802.3ad: This mode can be a good choice for this type of network
1749 topology. The 802.3ad mode is an IEEE standard, so all peers
1750 that implement 802.3ad should interoperate well. The 802.3ad
1751 protocol includes automatic configuration of the aggregates,
1752 so minimal manual configuration of the switch is needed
1753 (typically only to designate that some set of devices is
1754 available for 802.3ad). The 802.3ad standard also mandates
1755 that frames be delivered in order (within certain limits), so
1756 in general single connections will not see misordering of
1757 packets. The 802.3ad mode does have some drawbacks: the
1758 standard mandates that all devices in the aggregate operate at
1759 the same speed and duplex. Also, as with all bonding load
1760 balance modes other than balance-rr, no single connection will
1761 be able to utilize more than a single interface's worth of
1764 Additionally, the linux bonding 802.3ad implementation
1765 distributes traffic by peer (using an XOR of MAC addresses),
1766 so in a "gatewayed" configuration, all outgoing traffic will
1767 generally use the same device. Incoming traffic may also end
1768 up on a single device, but that is dependent upon the
1769 balancing policy of the peer's 8023.ad implementation. In a
1770 "local" configuration, traffic will be distributed across the
1771 devices in the bond.
1773 Finally, the 802.3ad mode mandates the use of the MII monitor,
1774 therefore, the ARP monitor is not available in this mode.
1776 balance-tlb: The balance-tlb mode balances outgoing traffic by peer.
1777 Since the balancing is done according to MAC address, in a
1778 "gatewayed" configuration (as described above), this mode will
1779 send all traffic across a single device. However, in a
1780 "local" network configuration, this mode balances multiple
1781 local network peers across devices in a vaguely intelligent
1782 manner (not a simple XOR as in balance-xor or 802.3ad mode),
1783 so that mathematically unlucky MAC addresses (i.e., ones that
1784 XOR to the same value) will not all "bunch up" on a single
1787 Unlike 802.3ad, interfaces may be of differing speeds, and no
1788 special switch configuration is required. On the down side,
1789 in this mode all incoming traffic arrives over a single
1790 interface, this mode requires certain ethtool support in the
1791 network device driver of the slave interfaces, and the ARP
1792 monitor is not available.
1794 balance-alb: This mode is everything that balance-tlb is, and more.
1795 It has all of the features (and restrictions) of balance-tlb,
1796 and will also balance incoming traffic from local network
1797 peers (as described in the Bonding Module Options section,
1800 The only additional down side to this mode is that the network
1801 device driver must support changing the hardware address while
1804 12.1.2 MT Link Monitoring for Single Switch Topology
1805 ----------------------------------------------------
1807 The choice of link monitoring may largely depend upon which
1808 mode you choose to use. The more advanced load balancing modes do not
1809 support the use of the ARP monitor, and are thus restricted to using
1810 the MII monitor (which does not provide as high a level of end to end
1811 assurance as the ARP monitor).
1813 12.2 Maximum Throughput in a Multiple Switch Topology
1814 -----------------------------------------------------
1816 Multiple switches may be utilized to optimize for throughput
1817 when they are configured in parallel as part of an isolated network
1818 between two or more systems, for example:
1824 +--------+ | +---------+
1826 +------+---+ +-----+----+ +-----+----+
1827 | Switch A | | Switch B | | Switch C |
1828 +------+---+ +-----+----+ +-----+----+
1830 +--------+ | +---------+
1836 In this configuration, the switches are isolated from one
1837 another. One reason to employ a topology such as this is for an
1838 isolated network with many hosts (a cluster configured for high
1839 performance, for example), using multiple smaller switches can be more
1840 cost effective than a single larger switch, e.g., on a network with 24
1841 hosts, three 24 port switches can be significantly less expensive than
1842 a single 72 port switch.
1844 If access beyond the network is required, an individual host
1845 can be equipped with an additional network device connected to an
1846 external network; this host then additionally acts as a gateway.
1848 12.2.1 MT Bonding Mode Selection for Multiple Switch Topology
1849 -------------------------------------------------------------
1851 In actual practice, the bonding mode typically employed in
1852 configurations of this type is balance-rr. Historically, in this
1853 network configuration, the usual caveats about out of order packet
1854 delivery are mitigated by the use of network adapters that do not do
1855 any kind of packet coalescing (via the use of NAPI, or because the
1856 device itself does not generate interrupts until some number of
1857 packets has arrived). When employed in this fashion, the balance-rr
1858 mode allows individual connections between two hosts to effectively
1859 utilize greater than one interface's bandwidth.
1861 12.2.2 MT Link Monitoring for Multiple Switch Topology
1862 ------------------------------------------------------
1864 Again, in actual practice, the MII monitor is most often used
1865 in this configuration, as performance is given preference over
1866 availability. The ARP monitor will function in this topology, but its
1867 advantages over the MII monitor are mitigated by the volume of probes
1868 needed as the number of systems involved grows (remember that each
1869 host in the network is configured with bonding).
1871 13. Switch Behavior Issues
1872 ==========================
1874 13.1 Link Establishment and Failover Delays
1875 -------------------------------------------
1877 Some switches exhibit undesirable behavior with regard to the
1878 timing of link up and down reporting by the switch.
1880 First, when a link comes up, some switches may indicate that
1881 the link is up (carrier available), but not pass traffic over the
1882 interface for some period of time. This delay is typically due to
1883 some type of autonegotiation or routing protocol, but may also occur
1884 during switch initialization (e.g., during recovery after a switch
1885 failure). If you find this to be a problem, specify an appropriate
1886 value to the updelay bonding module option to delay the use of the
1887 relevant interface(s).
1889 Second, some switches may "bounce" the link state one or more
1890 times while a link is changing state. This occurs most commonly while
1891 the switch is initializing. Again, an appropriate updelay value may
1894 Note that when a bonding interface has no active links, the
1895 driver will immediately reuse the first link that goes up, even if the
1896 updelay parameter has been specified (the updelay is ignored in this
1897 case). If there are slave interfaces waiting for the updelay timeout
1898 to expire, the interface that first went into that state will be
1899 immediately reused. This reduces down time of the network if the
1900 value of updelay has been overestimated, and since this occurs only in
1901 cases with no connectivity, there is no additional penalty for
1902 ignoring the updelay.
1904 In addition to the concerns about switch timings, if your
1905 switches take a long time to go into backup mode, it may be desirable
1906 to not activate a backup interface immediately after a link goes down.
1907 Failover may be delayed via the downdelay bonding module option.
1909 13.2 Duplicated Incoming Packets
1910 --------------------------------
1912 It is not uncommon to observe a short burst of duplicated
1913 traffic when the bonding device is first used, or after it has been
1914 idle for some period of time. This is most easily observed by issuing
1915 a "ping" to some other host on the network, and noticing that the
1916 output from ping flags duplicates (typically one per slave).
1918 For example, on a bond in active-backup mode with five slaves
1919 all connected to one switch, the output may appear as follows:
1922 PING 10.0.4.2 (10.0.4.2) from 10.0.3.10 : 56(84) bytes of data.
1923 64 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.7 ms
1924 64 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.8 ms (DUP!)
1925 64 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.8 ms (DUP!)
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=2 ttl=64 time=0.216 ms
1929 64 bytes from 10.0.4.2: icmp_seq=3 ttl=64 time=0.267 ms
1930 64 bytes from 10.0.4.2: icmp_seq=4 ttl=64 time=0.222 ms
1932 This is not due to an error in the bonding driver, rather, it
1933 is a side effect of how many switches update their MAC forwarding
1934 tables. Initially, the switch does not associate the MAC address in
1935 the packet with a particular switch port, and so it may send the
1936 traffic to all ports until its MAC forwarding table is updated. Since
1937 the interfaces attached to the bond may occupy multiple ports on a
1938 single switch, when the switch (temporarily) floods the traffic to all
1939 ports, the bond device receives multiple copies of the same packet
1940 (one per slave device).
1942 The duplicated packet behavior is switch dependent, some
1943 switches exhibit this, and some do not. On switches that display this
1944 behavior, it can be induced by clearing the MAC forwarding table (on
1945 most Cisco switches, the privileged command "clear mac address-table
1946 dynamic" will accomplish this).
1948 14. Hardware Specific Considerations
1949 ====================================
1951 This section contains additional information for configuring
1952 bonding on specific hardware platforms, or for interfacing bonding
1953 with particular switches or other devices.
1955 14.1 IBM BladeCenter
1956 --------------------
1958 This applies to the JS20 and similar systems.
1960 On the JS20 blades, the bonding driver supports only
1961 balance-rr, active-backup, balance-tlb and balance-alb modes. This is
1962 largely due to the network topology inside the BladeCenter, detailed
1965 JS20 network adapter information
1966 --------------------------------
1968 All JS20s come with two Broadcom Gigabit Ethernet ports
1969 integrated on the planar (that's "motherboard" in IBM-speak). In the
1970 BladeCenter chassis, the eth0 port of all JS20 blades is hard wired to
1971 I/O Module #1; similarly, all eth1 ports are wired to I/O Module #2.
1972 An add-on Broadcom daughter card can be installed on a JS20 to provide
1973 two more Gigabit Ethernet ports. These ports, eth2 and eth3, are
1974 wired to I/O Modules 3 and 4, respectively.
1976 Each I/O Module may contain either a switch or a passthrough
1977 module (which allows ports to be directly connected to an external
1978 switch). Some bonding modes require a specific BladeCenter internal
1979 network topology in order to function; these are detailed below.
1981 Additional BladeCenter-specific networking information can be
1982 found in two IBM Redbooks (www.ibm.com/redbooks):
1984 "IBM eServer BladeCenter Networking Options"
1985 "IBM eServer BladeCenter Layer 2-7 Network Switching"
1987 BladeCenter networking configuration
1988 ------------------------------------
1990 Because a BladeCenter can be configured in a very large number
1991 of ways, this discussion will be confined to describing basic
1994 Normally, Ethernet Switch Modules (ESMs) are used in I/O
1995 modules 1 and 2. In this configuration, the eth0 and eth1 ports of a
1996 JS20 will be connected to different internal switches (in the
1997 respective I/O modules).
1999 A passthrough module (OPM or CPM, optical or copper,
2000 passthrough module) connects the I/O module directly to an external
2001 switch. By using PMs in I/O module #1 and #2, the eth0 and eth1
2002 interfaces of a JS20 can be redirected to the outside world and
2003 connected to a common external switch.
2005 Depending upon the mix of ESMs and PMs, the network will
2006 appear to bonding as either a single switch topology (all PMs) or as a
2007 multiple switch topology (one or more ESMs, zero or more PMs). It is
2008 also possible to connect ESMs together, resulting in a configuration
2009 much like the example in "High Availability in a Multiple Switch
2012 Requirements for specific modes
2013 -------------------------------
2015 The balance-rr mode requires the use of passthrough modules
2016 for devices in the bond, all connected to an common external switch.
2017 That switch must be configured for "etherchannel" or "trunking" on the
2018 appropriate ports, as is usual for balance-rr.
2020 The balance-alb and balance-tlb modes will function with
2021 either switch modules or passthrough modules (or a mix). The only
2022 specific requirement for these modes is that all network interfaces
2023 must be able to reach all destinations for traffic sent over the
2024 bonding device (i.e., the network must converge at some point outside
2027 The active-backup mode has no additional requirements.
2029 Link monitoring issues
2030 ----------------------
2032 When an Ethernet Switch Module is in place, only the ARP
2033 monitor will reliably detect link loss to an external switch. This is
2034 nothing unusual, but examination of the BladeCenter cabinet would
2035 suggest that the "external" network ports are the ethernet ports for
2036 the system, when it fact there is a switch between these "external"
2037 ports and the devices on the JS20 system itself. The MII monitor is
2038 only able to detect link failures between the ESM and the JS20 system.
2040 When a passthrough module is in place, the MII monitor does
2041 detect failures to the "external" port, which is then directly
2042 connected to the JS20 system.
2047 The Serial Over LAN (SoL) link is established over the primary
2048 ethernet (eth0) only, therefore, any loss of link to eth0 will result
2049 in losing your SoL connection. It will not fail over with other
2050 network traffic, as the SoL system is beyond the control of the
2053 It may be desirable to disable spanning tree on the switch
2054 (either the internal Ethernet Switch Module, or an external switch) to
2055 avoid fail-over delay issues when using bonding.
2058 15. Frequently Asked Questions
2059 ==============================
2063 Yes. The old 2.0.xx channel bonding patch was not SMP safe.
2064 The new driver was designed to be SMP safe from the start.
2066 2. What type of cards will work with it?
2068 Any Ethernet type cards (you can even mix cards - a Intel
2069 EtherExpress PRO/100 and a 3com 3c905b, for example). For most modes,
2070 devices need not be of the same speed.
2072 3. How many bonding devices can I have?
2076 4. How many slaves can a bonding device have?
2078 This is limited only by the number of network interfaces Linux
2079 supports and/or the number of network cards you can place in your
2082 5. What happens when a slave link dies?
2084 If link monitoring is enabled, then the failing device will be
2085 disabled. The active-backup mode will fail over to a backup link, and
2086 other modes will ignore the failed link. The link will continue to be
2087 monitored, and should it recover, it will rejoin the bond (in whatever
2088 manner is appropriate for the mode). See the sections on High
2089 Availability and the documentation for each mode for additional
2092 Link monitoring can be enabled via either the miimon or
2093 arp_interval parameters (described in the module parameters section,
2094 above). In general, miimon monitors the carrier state as sensed by
2095 the underlying network device, and the arp monitor (arp_interval)
2096 monitors connectivity to another host on the local network.
2098 If no link monitoring is configured, the bonding driver will
2099 be unable to detect link failures, and will assume that all links are
2100 always available. This will likely result in lost packets, and a
2101 resulting degradation of performance. The precise performance loss
2102 depends upon the bonding mode and network configuration.
2104 6. Can bonding be used for High Availability?
2106 Yes. See the section on High Availability for details.
2108 7. Which switches/systems does it work with?
2110 The full answer to this depends upon the desired mode.
2112 In the basic balance modes (balance-rr and balance-xor), it
2113 works with any system that supports etherchannel (also called
2114 trunking). Most managed switches currently available have such
2115 support, and many unmanaged switches as well.
2117 The advanced balance modes (balance-tlb and balance-alb) do
2118 not have special switch requirements, but do need device drivers that
2119 support specific features (described in the appropriate section under
2120 module parameters, above).
2122 In 802.3ad mode, it works with systems that support IEEE
2123 802.3ad Dynamic Link Aggregation. Most managed and many unmanaged
2124 switches currently available support 802.3ad.
2126 The active-backup mode should work with any Layer-II switch.
2128 8. Where does a bonding device get its MAC address from?
2130 If not explicitly configured (with ifconfig or ip link), the
2131 MAC address of the bonding device is taken from its first slave
2132 device. This MAC address is then passed to all following slaves and
2133 remains persistent (even if the first slave is removed) until the
2134 bonding device is brought down or reconfigured.
2136 If you wish to change the MAC address, you can set it with
2137 ifconfig or ip link:
2139 # ifconfig bond0 hw ether 00:11:22:33:44:55
2141 # ip link set bond0 address 66:77:88:99:aa:bb
2143 The MAC address can be also changed by bringing down/up the
2144 device and then changing its slaves (or their order):
2146 # ifconfig bond0 down ; modprobe -r bonding
2147 # ifconfig bond0 .... up
2148 # ifenslave bond0 eth...
2150 This method will automatically take the address from the next
2151 slave that is added.
2153 To restore your slaves' MAC addresses, you need to detach them
2154 from the bond (`ifenslave -d bond0 eth0'). The bonding driver will
2155 then restore the MAC addresses that the slaves had before they were
2158 16. Resources and Links
2159 =======================
2161 The latest version of the bonding driver can be found in the latest
2162 version of the linux kernel, found on http://kernel.org
2164 The latest version of this document can be found in either the latest
2165 kernel source (named Documentation/networking/bonding.txt), or on the
2166 bonding sourceforge site:
2168 http://www.sourceforge.net/projects/bonding
2170 Discussions regarding the bonding driver take place primarily on the
2171 bonding-devel mailing list, hosted at sourceforge.net. If you have
2172 questions or problems, post them to the list. The list address is:
2174 bonding-devel@lists.sourceforge.net
2176 The administrative interface (to subscribe or unsubscribe) can
2179 https://lists.sourceforge.net/lists/listinfo/bonding-devel
2181 Donald Becker's Ethernet Drivers and diag programs may be found at :
2182 - http://www.scyld.com/network/
2184 You will also find a lot of information regarding Ethernet, NWay, MII,
2185 etc. at www.scyld.com.
git.samba.org / sfrench / cifs-2.6.git / blob