blob: 688dfe1e6b70f75a36d84f30dda2234c53664be0 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002 Linux Ethernet Bonding Driver HOWTO
3
Jay Vosburgh9a6c6862007-11-13 20:25:48 -08004 Latest update: 12 November 2007
Linus Torvalds1da177e2005-04-16 15:20:36 -07005
6Initial release : Thomas Davis <tadavis at lbl.gov>
7Corrections, 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>
13
14Reorganized and updated Feb 2005 by Jay Vosburgh
Auke Kok6224e012006-06-08 11:15:35 -070015Added Sysfs information: 2006/04/24
16 - Mitch Williams <mitch.a.williams at intel.com>
Linus Torvalds1da177e2005-04-16 15:20:36 -070017
Jay Vosburgh00354cf2005-07-21 12:18:02 -070018Introduction
19============
Linus Torvalds1da177e2005-04-16 15:20:36 -070020
Jay Vosburgh00354cf2005-07-21 12:18:02 -070021 The Linux bonding driver provides a method for aggregating
22multiple network interfaces into a single logical "bonded" interface.
23The behavior of the bonded interfaces depends upon the mode; generally
24speaking, modes provide either hot standby or load balancing services.
25Additionally, link integrity monitoring may be performed.
26
27 The bonding driver originally came from Donald Becker's
28beowulf patches for kernel 2.0. It has changed quite a bit since, and
29the original tools from extreme-linux and beowulf sites will not work
30with this version of the driver.
31
32 For new versions of the driver, updated userspace tools, and
33who to ask for help, please follow the links at the end of this file.
Linus Torvalds1da177e2005-04-16 15:20:36 -070034
35Table of Contents
36=================
37
381. Bonding Driver Installation
39
402. Bonding Driver Options
41
423. Configuring Bonding Devices
Auke Kok6224e012006-06-08 11:15:35 -0700433.1 Configuration with Sysconfig Support
443.1.1 Using DHCP with Sysconfig
453.1.2 Configuring Multiple Bonds with Sysconfig
463.2 Configuration with Initscripts Support
473.2.1 Using DHCP with Initscripts
483.2.2 Configuring Multiple Bonds with Initscripts
493.3 Configuring Bonding Manually with Ifenslave
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700503.3.1 Configuring Multiple Bonds Manually
Auke Kok6224e012006-06-08 11:15:35 -0700513.4 Configuring Bonding Manually via Sysfs
Linus Torvalds1da177e2005-04-16 15:20:36 -070052
Auke Kok6224e012006-06-08 11:15:35 -0700534. Querying Bonding Configuration
544.1 Bonding Configuration
554.2 Network Configuration
Linus Torvalds1da177e2005-04-16 15:20:36 -070056
Auke Kok6224e012006-06-08 11:15:35 -0700575. Switch Configuration
Linus Torvalds1da177e2005-04-16 15:20:36 -070058
Auke Kok6224e012006-06-08 11:15:35 -0700596. 802.1q VLAN Support
Linus Torvalds1da177e2005-04-16 15:20:36 -070060
Auke Kok6224e012006-06-08 11:15:35 -0700617. Link Monitoring
627.1 ARP Monitor Operation
637.2 Configuring Multiple ARP Targets
647.3 MII Monitor Operation
Linus Torvalds1da177e2005-04-16 15:20:36 -070065
Auke Kok6224e012006-06-08 11:15:35 -0700668. Potential Trouble Sources
678.1 Adventures in Routing
688.2 Ethernet Device Renaming
698.3 Painfully Slow Or No Failed Link Detection By Miimon
Linus Torvalds1da177e2005-04-16 15:20:36 -070070
Auke Kok6224e012006-06-08 11:15:35 -0700719. SNMP agents
Linus Torvalds1da177e2005-04-16 15:20:36 -070072
Auke Kok6224e012006-06-08 11:15:35 -07007310. Promiscuous mode
Linus Torvalds1da177e2005-04-16 15:20:36 -070074
Auke Kok6224e012006-06-08 11:15:35 -07007511. Configuring Bonding for High Availability
7611.1 High Availability in a Single Switch Topology
7711.2 High Availability in a Multiple Switch Topology
7811.2.1 HA Bonding Mode Selection for Multiple Switch Topology
7911.2.2 HA Link Monitoring for Multiple Switch Topology
Linus Torvalds1da177e2005-04-16 15:20:36 -070080
Auke Kok6224e012006-06-08 11:15:35 -07008112. Configuring Bonding for Maximum Throughput
8212.1 Maximum Throughput in a Single Switch Topology
8312.1.1 MT Bonding Mode Selection for Single Switch Topology
8412.1.2 MT Link Monitoring for Single Switch Topology
8512.2 Maximum Throughput in a Multiple Switch Topology
8612.2.1 MT Bonding Mode Selection for Multiple Switch Topology
8712.2.2 MT Link Monitoring for Multiple Switch Topology
Linus Torvalds1da177e2005-04-16 15:20:36 -070088
Auke Kok6224e012006-06-08 11:15:35 -07008913. Switch Behavior Issues
9013.1 Link Establishment and Failover Delays
9113.2 Duplicated Incoming Packets
Linus Torvalds1da177e2005-04-16 15:20:36 -070092
Auke Kok6224e012006-06-08 11:15:35 -07009314. Hardware Specific Considerations
9414.1 IBM BladeCenter
Jay Vosburgh00354cf2005-07-21 12:18:02 -070095
Auke Kok6224e012006-06-08 11:15:35 -07009615. Frequently Asked Questions
Jay Vosburgh00354cf2005-07-21 12:18:02 -070097
Auke Kok6224e012006-06-08 11:15:35 -07009816. Resources and Links
Linus Torvalds1da177e2005-04-16 15:20:36 -070099
100
1011. Bonding Driver Installation
102==============================
103
104 Most popular distro kernels ship with the bonding driver
105already available as a module and the ifenslave user level control
106program installed and ready for use. If your distro does not, or you
107have need to compile bonding from source (e.g., configuring and
108installing a mainline kernel from kernel.org), you'll need to perform
109the following steps:
110
1111.1 Configure and build the kernel with bonding
112-----------------------------------------------
113
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700114 The current version of the bonding driver is available in the
Linus Torvalds1da177e2005-04-16 15:20:36 -0700115drivers/net/bonding subdirectory of the most recent kernel source
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700116(which is available on http://kernel.org). Most users "rolling their
117own" will want to use the most recent kernel from kernel.org.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700118
119 Configure kernel with "make menuconfig" (or "make xconfig" or
120"make config"), then select "Bonding driver support" in the "Network
121device support" section. It is recommended that you configure the
122driver as module since it is currently the only way to pass parameters
123to the driver or configure more than one bonding device.
124
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700125 Build and install the new kernel and modules, then continue
126below to install ifenslave.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700127
1281.2 Install ifenslave Control Utility
129-------------------------------------
130
131 The ifenslave user level control program is included in the
132kernel source tree, in the file Documentation/networking/ifenslave.c.
133It is generally recommended that you use the ifenslave that
134corresponds to the kernel that you are using (either from the same
135source tree or supplied with the distro), however, ifenslave
136executables from older kernels should function (but features newer
137than the ifenslave release are not supported). Running an ifenslave
138that is newer than the kernel is not supported, and may or may not
139work.
140
141 To install ifenslave, do the following:
142
143# gcc -Wall -O -I/usr/src/linux/include ifenslave.c -o ifenslave
144# cp ifenslave /sbin/ifenslave
145
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
148source include directory.
149
150 You may wish to back up any existing /sbin/ifenslave, or, for
151testing or informal use, tag the ifenslave to the kernel version
152(e.g., name the ifenslave executable /sbin/ifenslave-2.6.10).
153
154IMPORTANT NOTE:
155
156 If you omit the "-I" or specify an incorrect directory, you
157may end up with an ifenslave that is incompatible with the kernel
158you're trying to build it for. Some distros (e.g., Red Hat from 7.1
159onwards) do not have /usr/include/linux symbolically linked to the
160default kernel source include directory.
161
Auke Kok6224e012006-06-08 11:15:35 -0700162SECOND IMPORTANT NOTE:
163 If you plan to configure bonding using sysfs, you do not need
164to use ifenslave.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700165
1662. Bonding Driver Options
167=========================
168
Jay Vosburgh9a6c6862007-11-13 20:25:48 -0800169 Options for the bonding driver are supplied as parameters to the
170bonding module at load time, or are specified via sysfs.
171
172 Module options may be given as command line arguments to the
173insmod or modprobe command, but are usually specified in either the
174/etc/modules.conf or /etc/modprobe.conf configuration file, or in a
175distro-specific configuration file (some of which are detailed in the next
176section).
177
178 Details on bonding support for sysfs is provided in the
179"Configuring Bonding Manually via Sysfs" section, below.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700180
181 The available bonding driver parameters are listed below. If a
182parameter is not specified the default value is used. When initially
183configuring a bond, it is recommended "tail -f /var/log/messages" be
184run in a separate window to watch for bonding driver error messages.
185
186 It is critical that either the miimon or arp_interval and
187arp_ip_target parameters be specified, otherwise serious network
188degradation will occur during link failures. Very few devices do not
189support at least miimon, so there is really no reason not to use it.
190
191 Options with textual values will accept either the text name
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700192or, for backwards compatibility, the option value. E.g.,
193"mode=802.3ad" and "mode=4" set the same mode.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700194
195 The parameters are as follows:
196
197arp_interval
198
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700199 Specifies the ARP link monitoring frequency in milliseconds.
Jay Vosburghf5b2b962006-09-22 21:54:53 -0700200
201 The ARP monitor works by periodically checking the slave
202 devices to determine whether they have sent or received
203 traffic recently (the precise criteria depends upon the
204 bonding mode, and the state of the slave). Regular traffic is
205 generated via ARP probes issued for the addresses specified by
206 the arp_ip_target option.
207
208 This behavior can be modified by the arp_validate option,
209 below.
210
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700211 If ARP monitoring is used in an etherchannel compatible mode
212 (modes 0 and 2), the switch should be configured in a mode
213 that evenly distributes packets across all links. If the
214 switch is configured to distribute the packets in an XOR
Linus Torvalds1da177e2005-04-16 15:20:36 -0700215 fashion, all replies from the ARP targets will be received on
216 the same link which could cause the other team members to
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700217 fail. ARP monitoring should not be used in conjunction with
218 miimon. A value of 0 disables ARP monitoring. The default
Linus Torvalds1da177e2005-04-16 15:20:36 -0700219 value is 0.
220
221arp_ip_target
222
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700223 Specifies the IP addresses to use as ARP monitoring peers when
224 arp_interval is > 0. These are the targets of the ARP request
225 sent to determine the health of the link to the targets.
226 Specify these values in ddd.ddd.ddd.ddd format. Multiple IP
227 addresses must be separated by a comma. At least one IP
228 address must be given for ARP monitoring to function. The
229 maximum number of targets that can be specified is 16. The
230 default value is no IP addresses.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700231
Jay Vosburghf5b2b962006-09-22 21:54:53 -0700232arp_validate
233
234 Specifies whether or not ARP probes and replies should be
235 validated in the active-backup mode. This causes the ARP
236 monitor to examine the incoming ARP requests and replies, and
237 only consider a slave to be up if it is receiving the
238 appropriate ARP traffic.
239
240 Possible values are:
241
242 none or 0
243
244 No validation is performed. This is the default.
245
246 active or 1
247
248 Validation is performed only for the active slave.
249
250 backup or 2
251
252 Validation is performed only for backup slaves.
253
254 all or 3
255
256 Validation is performed for all slaves.
257
258 For the active slave, the validation checks ARP replies to
259 confirm that they were generated by an arp_ip_target. Since
260 backup slaves do not typically receive these replies, the
261 validation performed for backup slaves is on the ARP request
262 sent out via the active slave. It is possible that some
263 switch or network configurations may result in situations
264 wherein the backup slaves do not receive the ARP requests; in
265 such a situation, validation of backup slaves must be
266 disabled.
267
268 This option is useful in network configurations in which
269 multiple bonding hosts are concurrently issuing ARPs to one or
270 more targets beyond a common switch. Should the link between
271 the switch and target fail (but not the switch itself), the
272 probe traffic generated by the multiple bonding instances will
273 fool the standard ARP monitor into considering the links as
274 still up. Use of the arp_validate option can resolve this, as
275 the ARP monitor will only consider ARP requests and replies
276 associated with its own instance of bonding.
277
278 This option was added in bonding version 3.1.0.
279
Linus Torvalds1da177e2005-04-16 15:20:36 -0700280downdelay
281
282 Specifies the time, in milliseconds, to wait before disabling
283 a slave after a link failure has been detected. This option
284 is only valid for the miimon link monitor. The downdelay
285 value should be a multiple of the miimon value; if not, it
286 will be rounded down to the nearest multiple. The default
287 value is 0.
288
Jay Vosburghdd957c52007-10-09 19:57:24 -0700289fail_over_mac
290
291 Specifies whether active-backup mode should set all slaves to
Jay Vosburgh3915c1e82008-05-17 21:10:14 -0700292 the same MAC address at enslavement (the traditional
293 behavior), or, when enabled, perform special handling of the
294 bond's MAC address in accordance with the selected policy.
Jay Vosburghdd957c52007-10-09 19:57:24 -0700295
Jay Vosburgh3915c1e82008-05-17 21:10:14 -0700296 Possible values are:
Jay Vosburghdd957c52007-10-09 19:57:24 -0700297
Jay Vosburgh3915c1e82008-05-17 21:10:14 -0700298 none or 0
Jay Vosburghdd957c52007-10-09 19:57:24 -0700299
Jay Vosburgh3915c1e82008-05-17 21:10:14 -0700300 This setting disables fail_over_mac, and causes
301 bonding to set all slaves of an active-backup bond to
302 the same MAC address at enslavement time. This is the
303 default.
Jay Vosburghdd957c52007-10-09 19:57:24 -0700304
Jay Vosburgh3915c1e82008-05-17 21:10:14 -0700305 active or 1
Jay Vosburghdd957c52007-10-09 19:57:24 -0700306
Jay Vosburgh3915c1e82008-05-17 21:10:14 -0700307 The "active" fail_over_mac policy indicates that the
308 MAC address of the bond should always be the MAC
309 address of the currently active slave. The MAC
310 address of the slaves is not changed; instead, the MAC
311 address of the bond changes during a failover.
312
313 This policy is useful for devices that cannot ever
314 alter their MAC address, or for devices that refuse
315 incoming broadcasts with their own source MAC (which
316 interferes with the ARP monitor).
317
318 The down side of this policy is that every device on
319 the network must be updated via gratuitous ARP,
320 vs. just updating a switch or set of switches (which
321 often takes place for any traffic, not just ARP
322 traffic, if the switch snoops incoming traffic to
323 update its tables) for the traditional method. If the
324 gratuitous ARP is lost, communication may be
325 disrupted.
326
327 When this policy is used in conjuction with the mii
328 monitor, devices which assert link up prior to being
329 able to actually transmit and receive are particularly
330 susecptible to loss of the gratuitous ARP, and an
331 appropriate updelay setting may be required.
332
333 follow or 2
334
335 The "follow" fail_over_mac policy causes the MAC
336 address of the bond to be selected normally (normally
337 the MAC address of the first slave added to the bond).
338 However, the second and subsequent slaves are not set
339 to this MAC address while they are in a backup role; a
340 slave is programmed with the bond's MAC address at
341 failover time (and the formerly active slave receives
342 the newly active slave's MAC address).
343
344 This policy is useful for multiport devices that
345 either become confused or incur a performance penalty
346 when multiple ports are programmed with the same MAC
347 address.
348
349
350 The default policy is none, unless the first slave cannot
351 change its MAC address, in which case the active policy is
352 selected by default.
353
354 This option may be modified via sysfs only when no slaves are
355 present in the bond.
356
357 This option was added in bonding version 3.2.0. The "follow"
358 policy was added in bonding version 3.3.0.
Jay Vosburghdd957c52007-10-09 19:57:24 -0700359
Linus Torvalds1da177e2005-04-16 15:20:36 -0700360lacp_rate
361
362 Option specifying the rate in which we'll ask our link partner
363 to transmit LACPDU packets in 802.3ad mode. Possible values
364 are:
365
366 slow or 0
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700367 Request partner to transmit LACPDUs every 30 seconds
Linus Torvalds1da177e2005-04-16 15:20:36 -0700368
369 fast or 1
370 Request partner to transmit LACPDUs every 1 second
371
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700372 The default is slow.
373
Linus Torvalds1da177e2005-04-16 15:20:36 -0700374max_bonds
375
376 Specifies the number of bonding devices to create for this
377 instance of the bonding driver. E.g., if max_bonds is 3, and
378 the bonding driver is not already loaded, then bond0, bond1
Jay Vosburghb8a97872008-06-13 18:12:04 -0700379 and bond2 will be created. The default value is 1. Specifying
380 a value of 0 will load bonding, but will not create any devices.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700381
382miimon
383
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700384 Specifies the MII link monitoring frequency in milliseconds.
385 This determines how often the link state of each slave is
386 inspected for link failures. A value of zero disables MII
387 link monitoring. A value of 100 is a good starting point.
388 The use_carrier option, below, affects how the link state is
Linus Torvalds1da177e2005-04-16 15:20:36 -0700389 determined. See the High Availability section for additional
390 information. The default value is 0.
391
392mode
393
394 Specifies one of the bonding policies. The default is
395 balance-rr (round robin). Possible values are:
396
397 balance-rr or 0
398
399 Round-robin policy: Transmit packets in sequential
400 order from the first available slave through the
401 last. This mode provides load balancing and fault
402 tolerance.
403
404 active-backup or 1
405
406 Active-backup policy: Only one slave in the bond is
407 active. A different slave becomes active if, and only
408 if, the active slave fails. The bond's MAC address is
409 externally visible on only one port (network adapter)
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700410 to avoid confusing the switch.
411
412 In bonding version 2.6.2 or later, when a failover
413 occurs in active-backup mode, bonding will issue one
414 or more gratuitous ARPs on the newly active slave.
Auke Kok6224e012006-06-08 11:15:35 -0700415 One gratuitous ARP is issued for the bonding master
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700416 interface and each VLAN interfaces configured above
417 it, provided that the interface has at least one IP
418 address configured. Gratuitous ARPs issued for VLAN
419 interfaces are tagged with the appropriate VLAN id.
420
421 This mode provides fault tolerance. The primary
422 option, documented below, affects the behavior of this
423 mode.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700424
425 balance-xor or 2
426
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700427 XOR policy: Transmit based on the selected transmit
428 hash policy. The default policy is a simple [(source
429 MAC address XOR'd with destination MAC address) modulo
430 slave count]. Alternate transmit policies may be
431 selected via the xmit_hash_policy option, described
432 below.
433
434 This mode provides load balancing and fault tolerance.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700435
436 broadcast or 3
437
438 Broadcast policy: transmits everything on all slave
439 interfaces. This mode provides fault tolerance.
440
441 802.3ad or 4
442
443 IEEE 802.3ad Dynamic link aggregation. Creates
444 aggregation groups that share the same speed and
445 duplex settings. Utilizes all slaves in the active
446 aggregator according to the 802.3ad specification.
447
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700448 Slave selection for outgoing traffic is done according
449 to the transmit hash policy, which may be changed from
450 the default simple XOR policy via the xmit_hash_policy
451 option, documented below. Note that not all transmit
452 policies may be 802.3ad compliant, particularly in
453 regards to the packet mis-ordering requirements of
454 section 43.2.4 of the 802.3ad standard. Differing
455 peer implementations will have varying tolerances for
456 noncompliance.
457
458 Prerequisites:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700459
460 1. Ethtool support in the base drivers for retrieving
461 the speed and duplex of each slave.
462
463 2. A switch that supports IEEE 802.3ad Dynamic link
464 aggregation.
465
466 Most switches will require some type of configuration
467 to enable 802.3ad mode.
468
469 balance-tlb or 5
470
471 Adaptive transmit load balancing: channel bonding that
472 does not require any special switch support. The
473 outgoing traffic is distributed according to the
474 current load (computed relative to the speed) on each
475 slave. Incoming traffic is received by the current
476 slave. If the receiving slave fails, another slave
477 takes over the MAC address of the failed receiving
478 slave.
479
480 Prerequisite:
481
482 Ethtool support in the base drivers for retrieving the
483 speed of each slave.
484
485 balance-alb or 6
486
487 Adaptive load balancing: includes balance-tlb plus
488 receive load balancing (rlb) for IPV4 traffic, and
489 does not require any special switch support. The
490 receive load balancing is achieved by ARP negotiation.
491 The bonding driver intercepts the ARP Replies sent by
492 the local system on their way out and overwrites the
493 source hardware address with the unique hardware
494 address of one of the slaves in the bond such that
495 different peers use different hardware addresses for
496 the server.
497
498 Receive traffic from connections created by the server
499 is also balanced. When the local system sends an ARP
500 Request the bonding driver copies and saves the peer's
501 IP information from the ARP packet. When the ARP
502 Reply arrives from the peer, its hardware address is
503 retrieved and the bonding driver initiates an ARP
504 reply to this peer assigning it to one of the slaves
505 in the bond. A problematic outcome of using ARP
506 negotiation for balancing is that each time that an
507 ARP request is broadcast it uses the hardware address
508 of the bond. Hence, peers learn the hardware address
509 of the bond and the balancing of receive traffic
510 collapses to the current slave. This is handled by
511 sending updates (ARP Replies) to all the peers with
512 their individually assigned hardware address such that
513 the traffic is redistributed. Receive traffic is also
514 redistributed when a new slave is added to the bond
515 and when an inactive slave is re-activated. The
516 receive load is distributed sequentially (round robin)
517 among the group of highest speed slaves in the bond.
518
519 When a link is reconnected or a new slave joins the
520 bond the receive traffic is redistributed among all
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700521 active slaves in the bond by initiating ARP Replies
Auke Kok6224e012006-06-08 11:15:35 -0700522 with the selected MAC address to each of the
Linus Torvalds1da177e2005-04-16 15:20:36 -0700523 clients. The updelay parameter (detailed below) must
524 be set to a value equal or greater than the switch's
525 forwarding delay so that the ARP Replies sent to the
526 peers will not be blocked by the switch.
527
528 Prerequisites:
529
530 1. Ethtool support in the base drivers for retrieving
531 the speed of each slave.
532
533 2. Base driver support for setting the hardware
534 address of a device while it is open. This is
535 required so that there will always be one slave in the
536 team using the bond hardware address (the
537 curr_active_slave) while having a unique hardware
538 address for each slave in the bond. If the
539 curr_active_slave fails its hardware address is
540 swapped with the new curr_active_slave that was
541 chosen.
542
Jay Vosburghb59f9f72008-06-13 18:12:03 -0700543num_grat_arp
544
545 Specifies the number of gratuitous ARPs to be issued after a
546 failover event. One gratuitous ARP is issued immediately after
547 the failover, subsequent ARPs are sent at a rate of one per link
548 monitor interval (arp_interval or miimon, whichever is active).
549
550 The valid range is 0 - 255; the default value is 1. This option
551 affects only the active-backup mode. This option was added for
552 bonding version 3.3.0.
553
Linus Torvalds1da177e2005-04-16 15:20:36 -0700554primary
555
556 A string (eth0, eth2, etc) specifying which slave is the
557 primary device. The specified device will always be the
558 active slave while it is available. Only when the primary is
559 off-line will alternate devices be used. This is useful when
560 one slave is preferred over another, e.g., when one slave has
561 higher throughput than another.
562
563 The primary option is only valid for active-backup mode.
564
565updelay
566
567 Specifies the time, in milliseconds, to wait before enabling a
568 slave after a link recovery has been detected. This option is
569 only valid for the miimon link monitor. The updelay value
570 should be a multiple of the miimon value; if not, it will be
571 rounded down to the nearest multiple. The default value is 0.
572
573use_carrier
574
575 Specifies whether or not miimon should use MII or ETHTOOL
576 ioctls vs. netif_carrier_ok() to determine the link
577 status. The MII or ETHTOOL ioctls are less efficient and
578 utilize a deprecated calling sequence within the kernel. The
579 netif_carrier_ok() relies on the device driver to maintain its
580 state with netif_carrier_on/off; at this writing, most, but
581 not all, device drivers support this facility.
582
583 If bonding insists that the link is up when it should not be,
584 it may be that your network device driver does not support
585 netif_carrier_on/off. The default state for netif_carrier is
586 "carrier on," so if a driver does not support netif_carrier,
587 it will appear as if the link is always up. In this case,
588 setting use_carrier to 0 will cause bonding to revert to the
589 MII / ETHTOOL ioctl method to determine the link state.
590
591 A value of 1 enables the use of netif_carrier_ok(), a value of
592 0 will use the deprecated MII / ETHTOOL ioctls. The default
593 value is 1.
594
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700595xmit_hash_policy
596
597 Selects the transmit hash policy to use for slave selection in
598 balance-xor and 802.3ad modes. Possible values are:
599
600 layer2
601
602 Uses XOR of hardware MAC addresses to generate the
603 hash. The formula is
604
605 (source MAC XOR destination MAC) modulo slave count
606
607 This algorithm will place all traffic to a particular
608 network peer on the same slave.
609
610 This algorithm is 802.3ad compliant.
611
Jay Vosburgh6f6652b2007-12-06 23:40:34 -0800612 layer2+3
613
614 This policy uses a combination of layer2 and layer3
615 protocol information to generate the hash.
616
617 Uses XOR of hardware MAC addresses and IP addresses to
618 generate the hash. The formula is
619
620 (((source IP XOR dest IP) AND 0xffff) XOR
621 ( source MAC XOR destination MAC ))
622 modulo slave count
623
624 This algorithm will place all traffic to a particular
625 network peer on the same slave. For non-IP traffic,
626 the formula is the same as for the layer2 transmit
627 hash policy.
628
629 This policy is intended to provide a more balanced
630 distribution of traffic than layer2 alone, especially
631 in environments where a layer3 gateway device is
632 required to reach most destinations.
633
Matt LaPlanted9195882008-07-25 19:45:33 -0700634 This algorithm is 802.3ad compliant.
Jay Vosburgh6f6652b2007-12-06 23:40:34 -0800635
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700636 layer3+4
637
638 This policy uses upper layer protocol information,
639 when available, to generate the hash. This allows for
640 traffic to a particular network peer to span multiple
641 slaves, although a single connection will not span
642 multiple slaves.
643
644 The formula for unfragmented TCP and UDP packets is
645
646 ((source port XOR dest port) XOR
647 ((source IP XOR dest IP) AND 0xffff)
648 modulo slave count
649
650 For fragmented TCP or UDP packets and all other IP
651 protocol traffic, the source and destination port
652 information is omitted. For non-IP traffic, the
653 formula is the same as for the layer2 transmit hash
654 policy.
655
656 This policy is intended to mimic the behavior of
657 certain switches, notably Cisco switches with PFC2 as
658 well as some Foundry and IBM products.
659
660 This algorithm is not fully 802.3ad compliant. A
661 single TCP or UDP conversation containing both
662 fragmented and unfragmented packets will see packets
663 striped across two interfaces. This may result in out
664 of order delivery. Most traffic types will not meet
665 this criteria, as TCP rarely fragments traffic, and
666 most UDP traffic is not involved in extended
667 conversations. Other implementations of 802.3ad may
668 or may not tolerate this noncompliance.
669
670 The default value is layer2. This option was added in bonding
Jay Vosburgh6f6652b2007-12-06 23:40:34 -0800671 version 2.6.3. In earlier versions of bonding, this parameter
672 does not exist, and the layer2 policy is the only policy. The
673 layer2+3 value was added for bonding version 3.2.2.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700674
675
6763. Configuring Bonding Devices
677==============================
678
Auke Kok6224e012006-06-08 11:15:35 -0700679 You can configure bonding using either your distro's network
680initialization scripts, or manually using either ifenslave or the
681sysfs interface. Distros generally use one of two packages for the
682network initialization scripts: initscripts or sysconfig. Recent
683versions of these packages have support for bonding, while older
684versions do not.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700685
686 We will first describe the options for configuring bonding for
687distros using versions of initscripts and sysconfig with full or
688partial support for bonding, then provide information on enabling
689bonding without support from the network initialization scripts (i.e.,
690older versions of initscripts or sysconfig).
691
692 If you're unsure whether your distro uses sysconfig or
693initscripts, or don't know if it's new enough, have no fear.
694Determining this is fairly straightforward.
695
696 First, issue the command:
697
698$ rpm -qf /sbin/ifup
699
700 It will respond with a line of text starting with either
701"initscripts" or "sysconfig," followed by some numbers. This is the
702package that provides your network initialization scripts.
703
704 Next, to determine if your installation supports bonding,
705issue the command:
706
707$ grep ifenslave /sbin/ifup
708
709 If this returns any matches, then your initscripts or
710sysconfig has support for bonding.
711
Auke Kok6224e012006-06-08 11:15:35 -07007123.1 Configuration with Sysconfig Support
Linus Torvalds1da177e2005-04-16 15:20:36 -0700713----------------------------------------
714
715 This section applies to distros using a version of sysconfig
716with bonding support, for example, SuSE Linux Enterprise Server 9.
717
718 SuSE SLES 9's networking configuration system does support
719bonding, however, at this writing, the YaST system configuration
Auke Kok6224e012006-06-08 11:15:35 -0700720front end does not provide any means to work with bonding devices.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700721Bonding devices can be managed by hand, however, as follows.
722
723 First, if they have not already been configured, configure the
724slave devices. On SLES 9, this is most easily done by running the
725yast2 sysconfig configuration utility. The goal is for to create an
726ifcfg-id file for each slave device. The simplest way to accomplish
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700727this is to configure the devices for DHCP (this is only to get the
728file ifcfg-id file created; see below for some issues with DHCP). The
729name of the configuration file for each device will be of the form:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700730
731ifcfg-id-xx:xx:xx:xx:xx:xx
732
733 Where the "xx" portion will be replaced with the digits from
734the device's permanent MAC address.
735
736 Once the set of ifcfg-id-xx:xx:xx:xx:xx:xx files has been
737created, it is necessary to edit the configuration files for the slave
738devices (the MAC addresses correspond to those of the slave devices).
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700739Before editing, the file will contain multiple lines, and will look
Linus Torvalds1da177e2005-04-16 15:20:36 -0700740something like this:
741
742BOOTPROTO='dhcp'
743STARTMODE='on'
744USERCTL='no'
745UNIQUE='XNzu.WeZGOGF+4wE'
746_nm_name='bus-pci-0001:61:01.0'
747
748 Change the BOOTPROTO and STARTMODE lines to the following:
749
750BOOTPROTO='none'
751STARTMODE='off'
752
753 Do not alter the UNIQUE or _nm_name lines. Remove any other
754lines (USERCTL, etc).
755
756 Once the ifcfg-id-xx:xx:xx:xx:xx:xx files have been modified,
757it's time to create the configuration file for the bonding device
758itself. This file is named ifcfg-bondX, where X is the number of the
759bonding device to create, starting at 0. The first such file is
760ifcfg-bond0, the second is ifcfg-bond1, and so on. The sysconfig
761network configuration system will correctly start multiple instances
762of bonding.
763
764 The contents of the ifcfg-bondX file is as follows:
765
766BOOTPROTO="static"
767BROADCAST="10.0.2.255"
768IPADDR="10.0.2.10"
769NETMASK="255.255.0.0"
770NETWORK="10.0.2.0"
771REMOTE_IPADDR=""
772STARTMODE="onboot"
773BONDING_MASTER="yes"
774BONDING_MODULE_OPTS="mode=active-backup miimon=100"
775BONDING_SLAVE0="eth0"
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700776BONDING_SLAVE1="bus-pci-0000:06:08.1"
Linus Torvalds1da177e2005-04-16 15:20:36 -0700777
778 Replace the sample BROADCAST, IPADDR, NETMASK and NETWORK
779values with the appropriate values for your network.
780
Linus Torvalds1da177e2005-04-16 15:20:36 -0700781 The STARTMODE specifies when the device is brought online.
782The possible values are:
783
784 onboot: The device is started at boot time. If you're not
785 sure, this is probably what you want.
786
787 manual: The device is started only when ifup is called
788 manually. Bonding devices may be configured this
789 way if you do not wish them to start automatically
790 at boot for some reason.
791
792 hotplug: The device is started by a hotplug event. This is not
793 a valid choice for a bonding device.
794
795 off or ignore: The device configuration is ignored.
796
797 The line BONDING_MASTER='yes' indicates that the device is a
798bonding master device. The only useful value is "yes."
799
800 The contents of BONDING_MODULE_OPTS are supplied to the
801instance of the bonding module for this device. Specify the options
802for the bonding mode, link monitoring, and so on here. Do not include
803the max_bonds bonding parameter; this will confuse the configuration
804system if you have multiple bonding devices.
805
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700806 Finally, supply one BONDING_SLAVEn="slave device" for each
807slave. where "n" is an increasing value, one for each slave. The
808"slave device" is either an interface name, e.g., "eth0", or a device
809specifier for the network device. The interface name is easier to
810find, but the ethN names are subject to change at boot time if, e.g.,
811a device early in the sequence has failed. The device specifiers
812(bus-pci-0000:06:08.1 in the example above) specify the physical
813network device, and will not change unless the device's bus location
814changes (for example, it is moved from one PCI slot to another). The
815example above uses one of each type for demonstration purposes; most
816configurations will choose one or the other for all slave devices.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700817
818 When all configuration files have been modified or created,
819networking must be restarted for the configuration changes to take
820effect. This can be accomplished via the following:
821
822# /etc/init.d/network restart
823
824 Note that the network control script (/sbin/ifdown) will
825remove the bonding module as part of the network shutdown processing,
826so it is not necessary to remove the module by hand if, e.g., the
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700827module parameters have changed.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700828
829 Also, at this writing, YaST/YaST2 will not manage bonding
830devices (they do not show bonding interfaces on its list of network
831devices). It is necessary to edit the configuration file by hand to
832change the bonding configuration.
833
834 Additional general options and details of the ifcfg file
835format can be found in an example ifcfg template file:
836
837/etc/sysconfig/network/ifcfg.template
838
839 Note that the template does not document the various BONDING_
840settings described above, but does describe many of the other options.
841
Auke Kok6224e012006-06-08 11:15:35 -07008423.1.1 Using DHCP with Sysconfig
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700843-------------------------------
844
845 Under sysconfig, configuring a device with BOOTPROTO='dhcp'
846will cause it to query DHCP for its IP address information. At this
847writing, this does not function for bonding devices; the scripts
848attempt to obtain the device address from DHCP prior to adding any of
849the slave devices. Without active slaves, the DHCP requests are not
850sent to the network.
851
Auke Kok6224e012006-06-08 11:15:35 -07008523.1.2 Configuring Multiple Bonds with Sysconfig
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700853-----------------------------------------------
854
855 The sysconfig network initialization system is capable of
856handling multiple bonding devices. All that is necessary is for each
857bonding instance to have an appropriately configured ifcfg-bondX file
858(as described above). Do not specify the "max_bonds" parameter to any
859instance of bonding, as this will confuse sysconfig. If you require
860multiple bonding devices with identical parameters, create multiple
861ifcfg-bondX files.
862
863 Because the sysconfig scripts supply the bonding module
864options in the ifcfg-bondX file, it is not necessary to add them to
865the system /etc/modules.conf or /etc/modprobe.conf configuration file.
866
Auke Kok6224e012006-06-08 11:15:35 -07008673.2 Configuration with Initscripts Support
Linus Torvalds1da177e2005-04-16 15:20:36 -0700868------------------------------------------
869
Jay Vosburgh9a6c6862007-11-13 20:25:48 -0800870 This section applies to distros using a recent version of
871initscripts with bonding support, for example, Red Hat Enterprise Linux
872version 3 or later, Fedora, etc. On these systems, the network
873initialization scripts have knowledge of bonding, and can be configured to
874control bonding devices. Note that older versions of the initscripts
875package have lower levels of support for bonding; this will be noted where
876applicable.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700877
878 These distros will not automatically load the network adapter
879driver unless the ethX device is configured with an IP address.
880Because of this constraint, users must manually configure a
881network-script file for all physical adapters that will be members of
882a bondX link. Network script files are located in the directory:
883
884/etc/sysconfig/network-scripts
885
886 The file name must be prefixed with "ifcfg-eth" and suffixed
887with the adapter's physical adapter number. For example, the script
888for eth0 would be named /etc/sysconfig/network-scripts/ifcfg-eth0.
889Place the following text in the file:
890
891DEVICE=eth0
892USERCTL=no
893ONBOOT=yes
894MASTER=bond0
895SLAVE=yes
896BOOTPROTO=none
897
898 The DEVICE= line will be different for every ethX device and
899must correspond with the name of the file, i.e., ifcfg-eth1 must have
900a device line of DEVICE=eth1. The setting of the MASTER= line will
901also depend on the final bonding interface name chosen for your bond.
902As with other network devices, these typically start at 0, and go up
903one for each device, i.e., the first bonding instance is bond0, the
904second is bond1, and so on.
905
906 Next, create a bond network script. The file name for this
907script will be /etc/sysconfig/network-scripts/ifcfg-bondX where X is
908the number of the bond. For bond0 the file is named "ifcfg-bond0",
909for bond1 it is named "ifcfg-bond1", and so on. Within that file,
910place the following text:
911
912DEVICE=bond0
913IPADDR=192.168.1.1
914NETMASK=255.255.255.0
915NETWORK=192.168.1.0
916BROADCAST=192.168.1.255
917ONBOOT=yes
918BOOTPROTO=none
919USERCTL=no
920
921 Be sure to change the networking specific lines (IPADDR,
922NETMASK, NETWORK and BROADCAST) to match your network configuration.
923
Jay Vosburgh9a6c6862007-11-13 20:25:48 -0800924 For later versions of initscripts, such as that found with Fedora
9257 and Red Hat Enterprise Linux version 5 (or later), it is possible, and,
926indeed, preferable, to specify the bonding options in the ifcfg-bond0
927file, e.g. a line of the format:
928
929BONDING_OPTS="mode=active-backup arp_interval=60 arp_ip_target=+192.168.1.254"
930
931 will configure the bond with the specified options. The options
932specified in BONDING_OPTS are identical to the bonding module parameters
933except for the arp_ip_target field. Each target should be included as a
934separate option and should be preceded by a '+' to indicate it should be
935added to the list of queried targets, e.g.,
936
937 arp_ip_target=+192.168.1.1 arp_ip_target=+192.168.1.2
938
939 is the proper syntax to specify multiple targets. When specifying
940options via BONDING_OPTS, it is not necessary to edit /etc/modules.conf or
941/etc/modprobe.conf.
942
943 For older versions of initscripts that do not support
944BONDING_OPTS, it is necessary to edit /etc/modules.conf (or
945/etc/modprobe.conf, depending upon your distro) to load the bonding module
946with your desired options when the bond0 interface is brought up. The
947following lines in /etc/modules.conf (or modprobe.conf) will load the
948bonding module, and select its options:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700949
950alias bond0 bonding
951options bond0 mode=balance-alb miimon=100
952
953 Replace the sample parameters with the appropriate set of
954options for your configuration.
955
956 Finally run "/etc/rc.d/init.d/network restart" as root. This
957will restart the networking subsystem and your bond link should be now
958up and running.
959
Auke Kok6224e012006-06-08 11:15:35 -07009603.2.1 Using DHCP with Initscripts
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700961---------------------------------
962
Jay Vosburgh9a6c6862007-11-13 20:25:48 -0800963 Recent versions of initscripts (the versions supplied with Fedora
964Core 3 and Red Hat Enterprise Linux 4, or later versions, are reported to
965work) have support for assigning IP information to bonding devices via
966DHCP.
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700967
968 To configure bonding for DHCP, configure it as described
969above, except replace the line "BOOTPROTO=none" with "BOOTPROTO=dhcp"
970and add a line consisting of "TYPE=Bonding". Note that the TYPE value
971is case sensitive.
972
Auke Kok6224e012006-06-08 11:15:35 -07009733.2.2 Configuring Multiple Bonds with Initscripts
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700974-------------------------------------------------
975
Jay Vosburgh9a6c6862007-11-13 20:25:48 -0800976 Initscripts packages that are included with Fedora 7 and Red Hat
977Enterprise Linux 5 support multiple bonding interfaces by simply
978specifying the appropriate BONDING_OPTS= in ifcfg-bondX where X is the
979number of the bond. This support requires sysfs support in the kernel,
980and a bonding driver of version 3.0.0 or later. Other configurations may
981not support this method for specifying multiple bonding interfaces; for
982those instances, see the "Configuring Multiple Bonds Manually" section,
983below.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700984
Auke Kok6224e012006-06-08 11:15:35 -07009853.3 Configuring Bonding Manually with Ifenslave
986-----------------------------------------------
Linus Torvalds1da177e2005-04-16 15:20:36 -0700987
988 This section applies to distros whose network initialization
989scripts (the sysconfig or initscripts package) do not have specific
990knowledge of bonding. One such distro is SuSE Linux Enterprise Server
991version 8.
992
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700993 The general method for these systems is to place the bonding
994module parameters into /etc/modules.conf or /etc/modprobe.conf (as
995appropriate for the installed distro), then add modprobe and/or
996ifenslave commands to the system's global init script. The name of
997the global init script differs; for sysconfig, it is
Linus Torvalds1da177e2005-04-16 15:20:36 -0700998/etc/init.d/boot.local and for initscripts it is /etc/rc.d/rc.local.
999
1000 For example, if you wanted to make a simple bond of two e100
1001devices (presumed to be eth0 and eth1), and have it persist across
1002reboots, edit the appropriate file (/etc/init.d/boot.local or
1003/etc/rc.d/rc.local), and add the following:
1004
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001005modprobe bonding mode=balance-alb miimon=100
Linus Torvalds1da177e2005-04-16 15:20:36 -07001006modprobe e100
1007ifconfig bond0 192.168.1.1 netmask 255.255.255.0 up
1008ifenslave bond0 eth0
1009ifenslave bond0 eth1
1010
1011 Replace the example bonding module parameters and bond0
1012network configuration (IP address, netmask, etc) with the appropriate
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001013values for your configuration.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001014
1015 Unfortunately, this method will not provide support for the
1016ifup and ifdown scripts on the bond devices. To reload the bonding
1017configuration, it is necessary to run the initialization script, e.g.,
1018
1019# /etc/init.d/boot.local
1020
1021 or
1022
1023# /etc/rc.d/rc.local
1024
1025 It may be desirable in such a case to create a separate script
1026which only initializes the bonding configuration, then call that
1027separate script from within boot.local. This allows for bonding to be
1028enabled without re-running the entire global init script.
1029
1030 To shut down the bonding devices, it is necessary to first
1031mark the bonding device itself as being down, then remove the
1032appropriate device driver modules. For our example above, you can do
1033the following:
1034
1035# ifconfig bond0 down
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001036# rmmod bonding
Linus Torvalds1da177e2005-04-16 15:20:36 -07001037# rmmod e100
1038
1039 Again, for convenience, it may be desirable to create a script
1040with these commands.
1041
1042
Jay Vosburgh00354cf2005-07-21 12:18:02 -070010433.3.1 Configuring Multiple Bonds Manually
1044-----------------------------------------
Linus Torvalds1da177e2005-04-16 15:20:36 -07001045
1046 This section contains information on configuring multiple
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001047bonding devices with differing options for those systems whose network
1048initialization scripts lack support for configuring multiple bonds.
1049
1050 If you require multiple bonding devices, but all with the same
1051options, you may wish to use the "max_bonds" module parameter,
1052documented above.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001053
Jay Vosburgh9a6c6862007-11-13 20:25:48 -08001054 To create multiple bonding devices with differing options, it is
1055preferrable to use bonding parameters exported by sysfs, documented in the
1056section below.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001057
Jay Vosburgh9a6c6862007-11-13 20:25:48 -08001058 For versions of bonding without sysfs support, the only means to
1059provide multiple instances of bonding with differing options is to load
1060the bonding driver multiple times. Note that current versions of the
1061sysconfig network initialization scripts handle this automatically; if
1062your distro uses these scripts, no special action is needed. See the
1063section Configuring Bonding Devices, above, if you're not sure about your
1064network initialization scripts.
1065
1066 To load multiple instances of the module, it is necessary to
1067specify a different name for each instance (the module loading system
1068requires that every loaded module, even multiple instances of the same
1069module, have a unique name). This is accomplished by supplying multiple
1070sets of bonding options in /etc/modprobe.conf, for example:
1071
1072alias bond0 bonding
1073options bond0 -o bond0 mode=balance-rr miimon=100
1074
1075alias bond1 bonding
1076options bond1 -o bond1 mode=balance-alb miimon=50
1077
1078 will load the bonding module two times. The first instance is
1079named "bond0" and creates the bond0 device in balance-rr mode with an
1080miimon of 100. The second instance is named "bond1" and creates the
1081bond1 device in balance-alb mode with an miimon of 50.
1082
1083 In some circumstances (typically with older distributions),
1084the above does not work, and the second bonding instance never sees
1085its options. In that case, the second options line can be substituted
1086as follows:
1087
1088install bond1 /sbin/modprobe --ignore-install bonding -o bond1 \
1089 mode=balance-alb miimon=50
1090
1091 This may be repeated any number of times, specifying a new and
1092unique name in place of bond1 for each subsequent instance.
1093
1094 It has been observed that some Red Hat supplied kernels are unable
1095to rename modules at load time (the "-o bond1" part). Attempts to pass
1096that option to modprobe will produce an "Operation not permitted" error.
1097This has been reported on some Fedora Core kernels, and has been seen on
1098RHEL 4 as well. On kernels exhibiting this problem, it will be impossible
1099to configure multiple bonds with differing parameters (as they are older
1100kernels, and also lack sysfs support).
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001101
Auke Kok6224e012006-06-08 11:15:35 -070011023.4 Configuring Bonding Manually via Sysfs
1103------------------------------------------
Linus Torvalds1da177e2005-04-16 15:20:36 -07001104
Jay Vosburgh9a6c6862007-11-13 20:25:48 -08001105 Starting with version 3.0.0, Channel Bonding may be configured
Auke Kok6224e012006-06-08 11:15:35 -07001106via the sysfs interface. This interface allows dynamic configuration
1107of all bonds in the system without unloading the module. It also
1108allows for adding and removing bonds at runtime. Ifenslave is no
1109longer required, though it is still supported.
1110
1111 Use of the sysfs interface allows you to use multiple bonds
1112with different configurations without having to reload the module.
1113It also allows you to use multiple, differently configured bonds when
1114bonding is compiled into the kernel.
1115
1116 You must have the sysfs filesystem mounted to configure
1117bonding this way. The examples in this document assume that you
1118are using the standard mount point for sysfs, e.g. /sys. If your
1119sysfs filesystem is mounted elsewhere, you will need to adjust the
1120example paths accordingly.
1121
1122Creating and Destroying Bonds
1123-----------------------------
1124To add a new bond foo:
1125# echo +foo > /sys/class/net/bonding_masters
1126
1127To remove an existing bond bar:
1128# echo -bar > /sys/class/net/bonding_masters
1129
1130To show all existing bonds:
1131# cat /sys/class/net/bonding_masters
1132
1133NOTE: due to 4K size limitation of sysfs files, this list may be
1134truncated if you have more than a few hundred bonds. This is unlikely
1135to occur under normal operating conditions.
1136
1137Adding and Removing Slaves
1138--------------------------
1139 Interfaces may be enslaved to a bond using the file
1140/sys/class/net/<bond>/bonding/slaves. The semantics for this file
1141are the same as for the bonding_masters file.
1142
1143To enslave interface eth0 to bond bond0:
1144# ifconfig bond0 up
1145# echo +eth0 > /sys/class/net/bond0/bonding/slaves
1146
1147To free slave eth0 from bond bond0:
1148# echo -eth0 > /sys/class/net/bond0/bonding/slaves
1149
Auke Kok6224e012006-06-08 11:15:35 -07001150 When an interface is enslaved to a bond, symlinks between the
1151two are created in the sysfs filesystem. In this case, you would get
1152/sys/class/net/bond0/slave_eth0 pointing to /sys/class/net/eth0, and
1153/sys/class/net/eth0/master pointing to /sys/class/net/bond0.
1154
1155 This means that you can tell quickly whether or not an
1156interface is enslaved by looking for the master symlink. Thus:
1157# echo -eth0 > /sys/class/net/eth0/master/bonding/slaves
1158will free eth0 from whatever bond it is enslaved to, regardless of
1159the name of the bond interface.
1160
1161Changing a Bond's Configuration
1162-------------------------------
1163 Each bond may be configured individually by manipulating the
1164files located in /sys/class/net/<bond name>/bonding
1165
1166 The names of these files correspond directly with the command-
Paolo Ornati670e9f32006-10-03 22:57:56 +02001167line parameters described elsewhere in this file, and, with the
Auke Kok6224e012006-06-08 11:15:35 -07001168exception of arp_ip_target, they accept the same values. To see the
1169current setting, simply cat the appropriate file.
1170
1171 A few examples will be given here; for specific usage
1172guidelines for each parameter, see the appropriate section in this
1173document.
1174
1175To configure bond0 for balance-alb mode:
1176# ifconfig bond0 down
1177# echo 6 > /sys/class/net/bond0/bonding/mode
1178 - or -
1179# echo balance-alb > /sys/class/net/bond0/bonding/mode
1180 NOTE: The bond interface must be down before the mode can be
1181changed.
1182
1183To enable MII monitoring on bond0 with a 1 second interval:
1184# echo 1000 > /sys/class/net/bond0/bonding/miimon
1185 NOTE: If ARP monitoring is enabled, it will disabled when MII
1186monitoring is enabled, and vice-versa.
1187
1188To add ARP targets:
1189# echo +192.168.0.100 > /sys/class/net/bond0/bonding/arp_ip_target
1190# echo +192.168.0.101 > /sys/class/net/bond0/bonding/arp_ip_target
1191 NOTE: up to 10 target addresses may be specified.
1192
1193To remove an ARP target:
1194# echo -192.168.0.100 > /sys/class/net/bond0/bonding/arp_ip_target
1195
1196Example Configuration
1197---------------------
1198 We begin with the same example that is shown in section 3.3,
1199executed with sysfs, and without using ifenslave.
1200
1201 To make a simple bond of two e100 devices (presumed to be eth0
1202and eth1), and have it persist across reboots, edit the appropriate
1203file (/etc/init.d/boot.local or /etc/rc.d/rc.local), and add the
1204following:
1205
1206modprobe bonding
1207modprobe e100
1208echo balance-alb > /sys/class/net/bond0/bonding/mode
1209ifconfig bond0 192.168.1.1 netmask 255.255.255.0 up
1210echo 100 > /sys/class/net/bond0/bonding/miimon
1211echo +eth0 > /sys/class/net/bond0/bonding/slaves
1212echo +eth1 > /sys/class/net/bond0/bonding/slaves
1213
1214 To add a second bond, with two e1000 interfaces in
1215active-backup mode, using ARP monitoring, add the following lines to
1216your init script:
1217
1218modprobe e1000
1219echo +bond1 > /sys/class/net/bonding_masters
1220echo active-backup > /sys/class/net/bond1/bonding/mode
1221ifconfig bond1 192.168.2.1 netmask 255.255.255.0 up
1222echo +192.168.2.100 /sys/class/net/bond1/bonding/arp_ip_target
1223echo 2000 > /sys/class/net/bond1/bonding/arp_interval
1224echo +eth2 > /sys/class/net/bond1/bonding/slaves
1225echo +eth3 > /sys/class/net/bond1/bonding/slaves
1226
1227
12284. Querying Bonding Configuration
Linus Torvalds1da177e2005-04-16 15:20:36 -07001229=================================
1230
Auke Kok6224e012006-06-08 11:15:35 -070012314.1 Bonding Configuration
Linus Torvalds1da177e2005-04-16 15:20:36 -07001232-------------------------
1233
1234 Each bonding device has a read-only file residing in the
1235/proc/net/bonding directory. The file contents include information
1236about the bonding configuration, options and state of each slave.
1237
1238 For example, the contents of /proc/net/bonding/bond0 after the
1239driver is loaded with parameters of mode=0 and miimon=1000 is
1240generally as follows:
1241
1242 Ethernet Channel Bonding Driver: 2.6.1 (October 29, 2004)
1243 Bonding Mode: load balancing (round-robin)
1244 Currently Active Slave: eth0
1245 MII Status: up
1246 MII Polling Interval (ms): 1000
1247 Up Delay (ms): 0
1248 Down Delay (ms): 0
1249
1250 Slave Interface: eth1
1251 MII Status: up
1252 Link Failure Count: 1
1253
1254 Slave Interface: eth0
1255 MII Status: up
1256 Link Failure Count: 1
1257
1258 The precise format and contents will change depending upon the
1259bonding configuration, state, and version of the bonding driver.
1260
Auke Kok6224e012006-06-08 11:15:35 -070012614.2 Network configuration
Linus Torvalds1da177e2005-04-16 15:20:36 -07001262-------------------------
1263
1264 The network configuration can be inspected using the ifconfig
1265command. Bonding devices will have the MASTER flag set; Bonding slave
1266devices will have the SLAVE flag set. The ifconfig output does not
1267contain information on which slaves are associated with which masters.
1268
1269 In the example below, the bond0 interface is the master
1270(MASTER) while eth0 and eth1 are slaves (SLAVE). Notice all slaves of
1271bond0 have the same MAC address (HWaddr) as bond0 for all modes except
1272TLB and ALB that require a unique MAC address for each slave.
1273
1274# /sbin/ifconfig
1275bond0 Link encap:Ethernet HWaddr 00:C0:F0:1F:37:B4
1276 inet addr:XXX.XXX.XXX.YYY Bcast:XXX.XXX.XXX.255 Mask:255.255.252.0
1277 UP BROADCAST RUNNING MASTER MULTICAST MTU:1500 Metric:1
1278 RX packets:7224794 errors:0 dropped:0 overruns:0 frame:0
1279 TX packets:3286647 errors:1 dropped:0 overruns:1 carrier:0
1280 collisions:0 txqueuelen:0
1281
1282eth0 Link encap:Ethernet HWaddr 00:C0:F0:1F:37:B4
Linus Torvalds1da177e2005-04-16 15:20:36 -07001283 UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
1284 RX packets:3573025 errors:0 dropped:0 overruns:0 frame:0
1285 TX packets:1643167 errors:1 dropped:0 overruns:1 carrier:0
1286 collisions:0 txqueuelen:100
1287 Interrupt:10 Base address:0x1080
1288
1289eth1 Link encap:Ethernet HWaddr 00:C0:F0:1F:37:B4
Linus Torvalds1da177e2005-04-16 15:20:36 -07001290 UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
1291 RX packets:3651769 errors:0 dropped:0 overruns:0 frame:0
1292 TX packets:1643480 errors:0 dropped:0 overruns:0 carrier:0
1293 collisions:0 txqueuelen:100
1294 Interrupt:9 Base address:0x1400
1295
Auke Kok6224e012006-06-08 11:15:35 -070012965. Switch Configuration
Linus Torvalds1da177e2005-04-16 15:20:36 -07001297=======================
1298
1299 For this section, "switch" refers to whatever system the
1300bonded devices are directly connected to (i.e., where the other end of
1301the cable plugs into). This may be an actual dedicated switch device,
1302or it may be another regular system (e.g., another computer running
1303Linux),
1304
1305 The active-backup, balance-tlb and balance-alb modes do not
1306require any specific configuration of the switch.
1307
1308 The 802.3ad mode requires that the switch have the appropriate
1309ports configured as an 802.3ad aggregation. The precise method used
1310to configure this varies from switch to switch, but, for example, a
1311Cisco 3550 series switch requires that the appropriate ports first be
1312grouped together in a single etherchannel instance, then that
1313etherchannel is set to mode "lacp" to enable 802.3ad (instead of
1314standard EtherChannel).
1315
1316 The balance-rr, balance-xor and broadcast modes generally
1317require that the switch have the appropriate ports grouped together.
1318The nomenclature for such a group differs between switches, it may be
1319called an "etherchannel" (as in the Cisco example, above), a "trunk
1320group" or some other similar variation. For these modes, each switch
1321will also have its own configuration options for the switch's transmit
1322policy to the bond. Typical choices include XOR of either the MAC or
1323IP addresses. The transmit policy of the two peers does not need to
1324match. For these three modes, the bonding mode really selects a
1325transmit policy for an EtherChannel group; all three will interoperate
1326with another EtherChannel group.
1327
1328
Auke Kok6224e012006-06-08 11:15:35 -070013296. 802.1q VLAN Support
Linus Torvalds1da177e2005-04-16 15:20:36 -07001330======================
1331
1332 It is possible to configure VLAN devices over a bond interface
1333using the 8021q driver. However, only packets coming from the 8021q
1334driver and passing through bonding will be tagged by default. Self
1335generated packets, for example, bonding's learning packets or ARP
1336packets generated by either ALB mode or the ARP monitor mechanism, are
1337tagged internally by bonding itself. As a result, bonding must
1338"learn" the VLAN IDs configured above it, and use those IDs to tag
1339self generated packets.
1340
1341 For reasons of simplicity, and to support the use of adapters
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001342that can do VLAN hardware acceleration offloading, the bonding
1343interface declares itself as fully hardware offloading capable, it gets
Linus Torvalds1da177e2005-04-16 15:20:36 -07001344the add_vid/kill_vid notifications to gather the necessary
1345information, and it propagates those actions to the slaves. In case
1346of mixed adapter types, hardware accelerated tagged packets that
1347should go through an adapter that is not offloading capable are
1348"un-accelerated" by the bonding driver so the VLAN tag sits in the
1349regular location.
1350
1351 VLAN interfaces *must* be added on top of a bonding interface
1352only after enslaving at least one slave. The bonding interface has a
1353hardware address of 00:00:00:00:00:00 until the first slave is added.
1354If the VLAN interface is created prior to the first enslavement, it
1355would pick up the all-zeroes hardware address. Once the first slave
1356is attached to the bond, the bond device itself will pick up the
1357slave's hardware address, which is then available for the VLAN device.
1358
1359 Also, be aware that a similar problem can occur if all slaves
1360are released from a bond that still has one or more VLAN interfaces on
1361top of it. When a new slave is added, the bonding interface will
1362obtain its hardware address from the first slave, which might not
1363match the hardware address of the VLAN interfaces (which was
1364ultimately copied from an earlier slave).
1365
1366 There are two methods to insure that the VLAN device operates
1367with the correct hardware address if all slaves are removed from a
1368bond interface:
1369
1370 1. Remove all VLAN interfaces then recreate them
1371
1372 2. Set the bonding interface's hardware address so that it
1373matches the hardware address of the VLAN interfaces.
1374
1375 Note that changing a VLAN interface's HW address would set the
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001376underlying device -- i.e. the bonding interface -- to promiscuous
Linus Torvalds1da177e2005-04-16 15:20:36 -07001377mode, which might not be what you want.
1378
1379
Auke Kok6224e012006-06-08 11:15:35 -070013807. Link Monitoring
Linus Torvalds1da177e2005-04-16 15:20:36 -07001381==================
1382
1383 The bonding driver at present supports two schemes for
1384monitoring a slave device's link state: the ARP monitor and the MII
1385monitor.
1386
1387 At the present time, due to implementation restrictions in the
1388bonding driver itself, it is not possible to enable both ARP and MII
1389monitoring simultaneously.
1390
Auke Kok6224e012006-06-08 11:15:35 -070013917.1 ARP Monitor Operation
Linus Torvalds1da177e2005-04-16 15:20:36 -07001392-------------------------
1393
1394 The ARP monitor operates as its name suggests: it sends ARP
1395queries to one or more designated peer systems on the network, and
1396uses the response as an indication that the link is operating. This
1397gives some assurance that traffic is actually flowing to and from one
1398or more peers on the local network.
1399
1400 The ARP monitor relies on the device driver itself to verify
1401that traffic is flowing. In particular, the driver must keep up to
1402date the last receive time, dev->last_rx, and transmit start time,
1403dev->trans_start. If these are not updated by the driver, then the
1404ARP monitor will immediately fail any slaves using that driver, and
1405those slaves will stay down. If networking monitoring (tcpdump, etc)
1406shows the ARP requests and replies on the network, then it may be that
1407your device driver is not updating last_rx and trans_start.
1408
Auke Kok6224e012006-06-08 11:15:35 -070014097.2 Configuring Multiple ARP Targets
Linus Torvalds1da177e2005-04-16 15:20:36 -07001410------------------------------------
1411
1412 While ARP monitoring can be done with just one target, it can
1413be useful in a High Availability setup to have several targets to
1414monitor. In the case of just one target, the target itself may go
1415down or have a problem making it unresponsive to ARP requests. Having
1416an additional target (or several) increases the reliability of the ARP
1417monitoring.
1418
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001419 Multiple ARP targets must be separated by commas as follows:
Linus Torvalds1da177e2005-04-16 15:20:36 -07001420
1421# example options for ARP monitoring with three targets
1422alias bond0 bonding
1423options bond0 arp_interval=60 arp_ip_target=192.168.0.1,192.168.0.3,192.168.0.9
1424
1425 For just a single target the options would resemble:
1426
1427# example options for ARP monitoring with one target
1428alias bond0 bonding
1429options bond0 arp_interval=60 arp_ip_target=192.168.0.100
1430
1431
Auke Kok6224e012006-06-08 11:15:35 -070014327.3 MII Monitor Operation
Linus Torvalds1da177e2005-04-16 15:20:36 -07001433-------------------------
1434
1435 The MII monitor monitors only the carrier state of the local
1436network interface. It accomplishes this in one of three ways: by
1437depending upon the device driver to maintain its carrier state, by
1438querying the device's MII registers, or by making an ethtool query to
1439the device.
1440
1441 If the use_carrier module parameter is 1 (the default value),
1442then the MII monitor will rely on the driver for carrier state
1443information (via the netif_carrier subsystem). As explained in the
1444use_carrier parameter information, above, if the MII monitor fails to
1445detect carrier loss on the device (e.g., when the cable is physically
1446disconnected), it may be that the driver does not support
1447netif_carrier.
1448
1449 If use_carrier is 0, then the MII monitor will first query the
1450device's (via ioctl) MII registers and check the link state. If that
1451request fails (not just that it returns carrier down), then the MII
1452monitor will make an ethtool ETHOOL_GLINK request to attempt to obtain
1453the same information. If both methods fail (i.e., the driver either
1454does not support or had some error in processing both the MII register
1455and ethtool requests), then the MII monitor will assume the link is
1456up.
1457
Auke Kok6224e012006-06-08 11:15:35 -070014588. Potential Sources of Trouble
Linus Torvalds1da177e2005-04-16 15:20:36 -07001459===============================
1460
Auke Kok6224e012006-06-08 11:15:35 -070014618.1 Adventures in Routing
Linus Torvalds1da177e2005-04-16 15:20:36 -07001462-------------------------
1463
1464 When bonding is configured, it is important that the slave
Auke Kok6224e012006-06-08 11:15:35 -07001465devices not have routes that supersede routes of the master (or,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001466generally, not have routes at all). For example, suppose the bonding
1467device bond0 has two slaves, eth0 and eth1, and the routing table is
1468as follows:
1469
1470Kernel IP routing table
1471Destination Gateway Genmask Flags MSS Window irtt Iface
147210.0.0.0 0.0.0.0 255.255.0.0 U 40 0 0 eth0
147310.0.0.0 0.0.0.0 255.255.0.0 U 40 0 0 eth1
147410.0.0.0 0.0.0.0 255.255.0.0 U 40 0 0 bond0
1475127.0.0.0 0.0.0.0 255.0.0.0 U 40 0 0 lo
1476
1477 This routing configuration will likely still update the
1478receive/transmit times in the driver (needed by the ARP monitor), but
1479may bypass the bonding driver (because outgoing traffic to, in this
1480case, another host on network 10 would use eth0 or eth1 before bond0).
1481
1482 The ARP monitor (and ARP itself) may become confused by this
1483configuration, because ARP requests (generated by the ARP monitor)
1484will be sent on one interface (bond0), but the corresponding reply
1485will arrive on a different interface (eth0). This reply looks to ARP
1486as an unsolicited ARP reply (because ARP matches replies on an
1487interface basis), and is discarded. The MII monitor is not affected
1488by the state of the routing table.
1489
1490 The solution here is simply to insure that slaves do not have
1491routes of their own, and if for some reason they must, those routes do
Auke Kok6224e012006-06-08 11:15:35 -07001492not supersede routes of their master. This should generally be the
Linus Torvalds1da177e2005-04-16 15:20:36 -07001493case, but unusual configurations or errant manual or automatic static
1494route additions may cause trouble.
1495
Auke Kok6224e012006-06-08 11:15:35 -070014968.2 Ethernet Device Renaming
Linus Torvalds1da177e2005-04-16 15:20:36 -07001497----------------------------
1498
1499 On systems with network configuration scripts that do not
1500associate physical devices directly with network interface names (so
1501that the same physical device always has the same "ethX" name), it may
1502be necessary to add some special logic to either /etc/modules.conf or
1503/etc/modprobe.conf (depending upon which is installed on the system).
1504
1505 For example, given a modules.conf containing the following:
1506
1507alias bond0 bonding
1508options bond0 mode=some-mode miimon=50
1509alias eth0 tg3
1510alias eth1 tg3
1511alias eth2 e1000
1512alias eth3 e1000
1513
1514 If neither eth0 and eth1 are slaves to bond0, then when the
1515bond0 interface comes up, the devices may end up reordered. This
1516happens because bonding is loaded first, then its slave device's
1517drivers are loaded next. Since no other drivers have been loaded,
1518when the e1000 driver loads, it will receive eth0 and eth1 for its
1519devices, but the bonding configuration tries to enslave eth2 and eth3
1520(which may later be assigned to the tg3 devices).
1521
1522 Adding the following:
1523
1524add above bonding e1000 tg3
1525
1526 causes modprobe to load e1000 then tg3, in that order, when
1527bonding is loaded. This command is fully documented in the
1528modules.conf manual page.
1529
1530 On systems utilizing modprobe.conf (or modprobe.conf.local),
1531an equivalent problem can occur. In this case, the following can be
1532added to modprobe.conf (or modprobe.conf.local, as appropriate), as
1533follows (all on one line; it has been split here for clarity):
1534
1535install bonding /sbin/modprobe tg3; /sbin/modprobe e1000;
1536 /sbin/modprobe --ignore-install bonding
1537
1538 This will, when loading the bonding module, rather than
1539performing the normal action, instead execute the provided command.
1540This command loads the device drivers in the order needed, then calls
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001541modprobe with --ignore-install to cause the normal action to then take
Linus Torvalds1da177e2005-04-16 15:20:36 -07001542place. Full documentation on this can be found in the modprobe.conf
1543and modprobe manual pages.
1544
Auke Kok6224e012006-06-08 11:15:35 -070015458.3. Painfully Slow Or No Failed Link Detection By Miimon
Linus Torvalds1da177e2005-04-16 15:20:36 -07001546---------------------------------------------------------
1547
1548 By default, bonding enables the use_carrier option, which
1549instructs bonding to trust the driver to maintain carrier state.
1550
1551 As discussed in the options section, above, some drivers do
1552not support the netif_carrier_on/_off link state tracking system.
1553With use_carrier enabled, bonding will always see these links as up,
1554regardless of their actual state.
1555
1556 Additionally, other drivers do support netif_carrier, but do
1557not maintain it in real time, e.g., only polling the link state at
1558some fixed interval. In this case, miimon will detect failures, but
1559only after some long period of time has expired. If it appears that
1560miimon is very slow in detecting link failures, try specifying
1561use_carrier=0 to see if that improves the failure detection time. If
1562it does, then it may be that the driver checks the carrier state at a
1563fixed interval, but does not cache the MII register values (so the
1564use_carrier=0 method of querying the registers directly works). If
1565use_carrier=0 does not improve the failover, then the driver may cache
1566the registers, or the problem may be elsewhere.
1567
1568 Also, remember that miimon only checks for the device's
1569carrier state. It has no way to determine the state of devices on or
1570beyond other ports of a switch, or if a switch is refusing to pass
1571traffic while still maintaining carrier on.
1572
Auke Kok6224e012006-06-08 11:15:35 -070015739. SNMP agents
Linus Torvalds1da177e2005-04-16 15:20:36 -07001574===============
1575
1576 If running SNMP agents, the bonding driver should be loaded
1577before any network drivers participating in a bond. This requirement
Tobias Klauserd533f672005-09-10 00:26:46 -07001578is due to the interface index (ipAdEntIfIndex) being associated to
Linus Torvalds1da177e2005-04-16 15:20:36 -07001579the first interface found with a given IP address. That is, there is
1580only one ipAdEntIfIndex for each IP address. For example, if eth0 and
1581eth1 are slaves of bond0 and the driver for eth0 is loaded before the
1582bonding driver, the interface for the IP address will be associated
1583with the eth0 interface. This configuration is shown below, the IP
1584address 192.168.1.1 has an interface index of 2 which indexes to eth0
1585in the ifDescr table (ifDescr.2).
1586
1587 interfaces.ifTable.ifEntry.ifDescr.1 = lo
1588 interfaces.ifTable.ifEntry.ifDescr.2 = eth0
1589 interfaces.ifTable.ifEntry.ifDescr.3 = eth1
1590 interfaces.ifTable.ifEntry.ifDescr.4 = eth2
1591 interfaces.ifTable.ifEntry.ifDescr.5 = eth3
1592 interfaces.ifTable.ifEntry.ifDescr.6 = bond0
1593 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.10.10.10 = 5
1594 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.192.168.1.1 = 2
1595 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.74.20.94 = 4
1596 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.127.0.0.1 = 1
1597
1598 This problem is avoided by loading the bonding driver before
1599any network drivers participating in a bond. Below is an example of
1600loading the bonding driver first, the IP address 192.168.1.1 is
1601correctly associated with ifDescr.2.
1602
1603 interfaces.ifTable.ifEntry.ifDescr.1 = lo
1604 interfaces.ifTable.ifEntry.ifDescr.2 = bond0
1605 interfaces.ifTable.ifEntry.ifDescr.3 = eth0
1606 interfaces.ifTable.ifEntry.ifDescr.4 = eth1
1607 interfaces.ifTable.ifEntry.ifDescr.5 = eth2
1608 interfaces.ifTable.ifEntry.ifDescr.6 = eth3
1609 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.10.10.10 = 6
1610 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.192.168.1.1 = 2
1611 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.10.74.20.94 = 5
1612 ip.ipAddrTable.ipAddrEntry.ipAdEntIfIndex.127.0.0.1 = 1
1613
1614 While some distributions may not report the interface name in
1615ifDescr, the association between the IP address and IfIndex remains
1616and SNMP functions such as Interface_Scan_Next will report that
1617association.
1618
Auke Kok6224e012006-06-08 11:15:35 -0700161910. Promiscuous mode
Linus Torvalds1da177e2005-04-16 15:20:36 -07001620====================
1621
1622 When running network monitoring tools, e.g., tcpdump, it is
1623common to enable promiscuous mode on the device, so that all traffic
1624is seen (instead of seeing only traffic destined for the local host).
1625The bonding driver handles promiscuous mode changes to the bonding
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001626master device (e.g., bond0), and propagates the setting to the slave
Linus Torvalds1da177e2005-04-16 15:20:36 -07001627devices.
1628
1629 For the balance-rr, balance-xor, broadcast, and 802.3ad modes,
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001630the promiscuous mode setting is propagated to all slaves.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001631
1632 For the active-backup, balance-tlb and balance-alb modes, the
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001633promiscuous mode setting is propagated only to the active slave.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001634
1635 For balance-tlb mode, the active slave is the slave currently
1636receiving inbound traffic.
1637
1638 For balance-alb mode, the active slave is the slave used as a
1639"primary." This slave is used for mode-specific control traffic, for
1640sending to peers that are unassigned or if the load is unbalanced.
1641
1642 For the active-backup, balance-tlb and balance-alb modes, when
1643the active slave changes (e.g., due to a link failure), the
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001644promiscuous setting will be propagated to the new active slave.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001645
Auke Kok6224e012006-06-08 11:15:35 -0700164611. Configuring Bonding for High Availability
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001647=============================================
Linus Torvalds1da177e2005-04-16 15:20:36 -07001648
1649 High Availability refers to configurations that provide
1650maximum network availability by having redundant or backup devices,
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001651links or switches between the host and the rest of the world. The
1652goal is to provide the maximum availability of network connectivity
1653(i.e., the network always works), even though other configurations
1654could provide higher throughput.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001655
Auke Kok6224e012006-06-08 11:15:35 -0700165611.1 High Availability in a Single Switch Topology
Linus Torvalds1da177e2005-04-16 15:20:36 -07001657--------------------------------------------------
1658
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001659 If two hosts (or a host and a single switch) are directly
1660connected via multiple physical links, then there is no availability
1661penalty to optimizing for maximum bandwidth. In this case, there is
1662only one switch (or peer), so if it fails, there is no alternative
1663access to fail over to. Additionally, the bonding load balance modes
1664support link monitoring of their members, so if individual links fail,
1665the load will be rebalanced across the remaining devices.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001666
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001667 See Section 13, "Configuring Bonding for Maximum Throughput"
1668for information on configuring bonding with one peer device.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001669
Auke Kok6224e012006-06-08 11:15:35 -0700167011.2 High Availability in a Multiple Switch Topology
Linus Torvalds1da177e2005-04-16 15:20:36 -07001671----------------------------------------------------
1672
1673 With multiple switches, the configuration of bonding and the
1674network changes dramatically. In multiple switch topologies, there is
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001675a trade off between network availability and usable bandwidth.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001676
1677 Below is a sample network, configured to maximize the
1678availability of the network:
1679
1680 | |
1681 |port3 port3|
1682 +-----+----+ +-----+----+
1683 | |port2 ISL port2| |
1684 | switch A +--------------------------+ switch B |
1685 | | | |
1686 +-----+----+ +-----++---+
1687 |port1 port1|
1688 | +-------+ |
1689 +-------------+ host1 +---------------+
1690 eth0 +-------+ eth1
1691
1692 In this configuration, there is a link between the two
1693switches (ISL, or inter switch link), and multiple ports connecting to
1694the outside world ("port3" on each switch). There is no technical
1695reason that this could not be extended to a third switch.
1696
Auke Kok6224e012006-06-08 11:15:35 -0700169711.2.1 HA Bonding Mode Selection for Multiple Switch Topology
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001698-------------------------------------------------------------
Linus Torvalds1da177e2005-04-16 15:20:36 -07001699
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001700 In a topology such as the example above, the active-backup and
1701broadcast modes are the only useful bonding modes when optimizing for
1702availability; the other modes require all links to terminate on the
1703same peer for them to behave rationally.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001704
1705active-backup: This is generally the preferred mode, particularly if
1706 the switches have an ISL and play together well. If the
1707 network configuration is such that one switch is specifically
1708 a backup switch (e.g., has lower capacity, higher cost, etc),
1709 then the primary option can be used to insure that the
1710 preferred link is always used when it is available.
1711
1712broadcast: This mode is really a special purpose mode, and is suitable
1713 only for very specific needs. For example, if the two
1714 switches are not connected (no ISL), and the networks beyond
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001715 them are totally independent. In this case, if it is
Linus Torvalds1da177e2005-04-16 15:20:36 -07001716 necessary for some specific one-way traffic to reach both
1717 independent networks, then the broadcast mode may be suitable.
1718
Auke Kok6224e012006-06-08 11:15:35 -0700171911.2.2 HA Link Monitoring Selection for Multiple Switch Topology
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001720----------------------------------------------------------------
Linus Torvalds1da177e2005-04-16 15:20:36 -07001721
1722 The choice of link monitoring ultimately depends upon your
1723switch. If the switch can reliably fail ports in response to other
1724failures, then either the MII or ARP monitors should work. For
1725example, in the above example, if the "port3" link fails at the remote
1726end, the MII monitor has no direct means to detect this. The ARP
1727monitor could be configured with a target at the remote end of port3,
1728thus detecting that failure without switch support.
1729
1730 In general, however, in a multiple switch topology, the ARP
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001731monitor can provide a higher level of reliability in detecting end to
1732end connectivity failures (which may be caused by the failure of any
1733individual component to pass traffic for any reason). Additionally,
1734the ARP monitor should be configured with multiple targets (at least
1735one for each switch in the network). This will insure that,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001736regardless of which switch is active, the ARP monitor has a suitable
1737target to query.
1738
Jay Vosburgh9a6c6862007-11-13 20:25:48 -08001739 Note, also, that of late many switches now support a functionality
1740generally referred to as "trunk failover." This is a feature of the
1741switch that causes the link state of a particular switch port to be set
1742down (or up) when the state of another switch port goes down (or up).
1743It's purpose is to propogate link failures from logically "exterior" ports
1744to the logically "interior" ports that bonding is able to monitor via
1745miimon. Availability and configuration for trunk failover varies by
1746switch, but this can be a viable alternative to the ARP monitor when using
1747suitable switches.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001748
Auke Kok6224e012006-06-08 11:15:35 -0700174912. Configuring Bonding for Maximum Throughput
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001750==============================================
Linus Torvalds1da177e2005-04-16 15:20:36 -07001751
Auke Kok6224e012006-06-08 11:15:35 -0700175212.1 Maximizing Throughput in a Single Switch Topology
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001753------------------------------------------------------
1754
1755 In a single switch configuration, the best method to maximize
1756throughput depends upon the application and network environment. The
1757various load balancing modes each have strengths and weaknesses in
1758different environments, as detailed below.
1759
1760 For this discussion, we will break down the topologies into
1761two categories. Depending upon the destination of most traffic, we
1762categorize them into either "gatewayed" or "local" configurations.
1763
1764 In a gatewayed configuration, the "switch" is acting primarily
1765as a router, and the majority of traffic passes through this router to
1766other networks. An example would be the following:
1767
1768
1769 +----------+ +----------+
1770 | |eth0 port1| | to other networks
1771 | Host A +---------------------+ router +------------------->
1772 | +---------------------+ | Hosts B and C are out
1773 | |eth1 port2| | here somewhere
1774 +----------+ +----------+
1775
1776 The router may be a dedicated router device, or another host
1777acting as a gateway. For our discussion, the important point is that
1778the majority of traffic from Host A will pass through the router to
1779some other network before reaching its final destination.
1780
1781 In a gatewayed network configuration, although Host A may
1782communicate with many other systems, all of its traffic will be sent
1783and received via one other peer on the local network, the router.
1784
1785 Note that the case of two systems connected directly via
1786multiple physical links is, for purposes of configuring bonding, the
1787same as a gatewayed configuration. In that case, it happens that all
1788traffic is destined for the "gateway" itself, not some other network
1789beyond the gateway.
1790
1791 In a local configuration, the "switch" is acting primarily as
1792a switch, and the majority of traffic passes through this switch to
1793reach other stations on the same network. An example would be the
1794following:
1795
1796 +----------+ +----------+ +--------+
1797 | |eth0 port1| +-------+ Host B |
1798 | Host A +------------+ switch |port3 +--------+
1799 | +------------+ | +--------+
1800 | |eth1 port2| +------------------+ Host C |
1801 +----------+ +----------+port4 +--------+
1802
1803
1804 Again, the switch may be a dedicated switch device, or another
1805host acting as a gateway. For our discussion, the important point is
1806that the majority of traffic from Host A is destined for other hosts
1807on the same local network (Hosts B and C in the above example).
1808
1809 In summary, in a gatewayed configuration, traffic to and from
1810the bonded device will be to the same MAC level peer on the network
1811(the gateway itself, i.e., the router), regardless of its final
1812destination. In a local configuration, traffic flows directly to and
1813from the final destinations, thus, each destination (Host B, Host C)
1814will be addressed directly by their individual MAC addresses.
1815
1816 This distinction between a gatewayed and a local network
1817configuration is important because many of the load balancing modes
1818available use the MAC addresses of the local network source and
1819destination to make load balancing decisions. The behavior of each
1820mode is described below.
1821
1822
Auke Kok6224e012006-06-08 11:15:35 -0700182312.1.1 MT Bonding Mode Selection for Single Switch Topology
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001824-----------------------------------------------------------
1825
1826 This configuration is the easiest to set up and to understand,
1827although you will have to decide which bonding mode best suits your
1828needs. The trade offs for each mode are detailed below:
1829
1830balance-rr: This mode is the only mode that will permit a single
1831 TCP/IP connection to stripe traffic across multiple
1832 interfaces. It is therefore the only mode that will allow a
1833 single TCP/IP stream to utilize more than one interface's
1834 worth of throughput. This comes at a cost, however: the
Jay Vosburgh9a6c6862007-11-13 20:25:48 -08001835 striping generally results in peer systems receiving packets out
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001836 of order, causing TCP/IP's congestion control system to kick
1837 in, often by retransmitting segments.
1838
1839 It is possible to adjust TCP/IP's congestion limits by
1840 altering the net.ipv4.tcp_reordering sysctl parameter. The
1841 usual default value is 3, and the maximum useful value is 127.
1842 For a four interface balance-rr bond, expect that a single
1843 TCP/IP stream will utilize no more than approximately 2.3
1844 interface's worth of throughput, even after adjusting
1845 tcp_reordering.
1846
Jay Vosburgh9a6c6862007-11-13 20:25:48 -08001847 Note that the fraction of packets that will be delivered out of
1848 order is highly variable, and is unlikely to be zero. The level
1849 of reordering depends upon a variety of factors, including the
1850 networking interfaces, the switch, and the topology of the
1851 configuration. Speaking in general terms, higher speed network
1852 cards produce more reordering (due to factors such as packet
1853 coalescing), and a "many to many" topology will reorder at a
1854 higher rate than a "many slow to one fast" configuration.
1855
1856 Many switches do not support any modes that stripe traffic
1857 (instead choosing a port based upon IP or MAC level addresses);
1858 for those devices, traffic for a particular connection flowing
1859 through the switch to a balance-rr bond will not utilize greater
1860 than one interface's worth of bandwidth.
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001861
1862 If you are utilizing protocols other than TCP/IP, UDP for
1863 example, and your application can tolerate out of order
1864 delivery, then this mode can allow for single stream datagram
1865 performance that scales near linearly as interfaces are added
1866 to the bond.
1867
1868 This mode requires the switch to have the appropriate ports
1869 configured for "etherchannel" or "trunking."
1870
1871active-backup: There is not much advantage in this network topology to
1872 the active-backup mode, as the inactive backup devices are all
1873 connected to the same peer as the primary. In this case, a
1874 load balancing mode (with link monitoring) will provide the
1875 same level of network availability, but with increased
1876 available bandwidth. On the plus side, active-backup mode
1877 does not require any configuration of the switch, so it may
1878 have value if the hardware available does not support any of
1879 the load balance modes.
1880
1881balance-xor: This mode will limit traffic such that packets destined
1882 for specific peers will always be sent over the same
1883 interface. Since the destination is determined by the MAC
1884 addresses involved, this mode works best in a "local" network
1885 configuration (as described above), with destinations all on
1886 the same local network. This mode is likely to be suboptimal
1887 if all your traffic is passed through a single router (i.e., a
1888 "gatewayed" network configuration, as described above).
1889
1890 As with balance-rr, the switch ports need to be configured for
1891 "etherchannel" or "trunking."
1892
1893broadcast: Like active-backup, there is not much advantage to this
1894 mode in this type of network topology.
1895
1896802.3ad: This mode can be a good choice for this type of network
1897 topology. The 802.3ad mode is an IEEE standard, so all peers
1898 that implement 802.3ad should interoperate well. The 802.3ad
1899 protocol includes automatic configuration of the aggregates,
1900 so minimal manual configuration of the switch is needed
1901 (typically only to designate that some set of devices is
1902 available for 802.3ad). The 802.3ad standard also mandates
1903 that frames be delivered in order (within certain limits), so
1904 in general single connections will not see misordering of
1905 packets. The 802.3ad mode does have some drawbacks: the
1906 standard mandates that all devices in the aggregate operate at
1907 the same speed and duplex. Also, as with all bonding load
1908 balance modes other than balance-rr, no single connection will
1909 be able to utilize more than a single interface's worth of
1910 bandwidth.
1911
1912 Additionally, the linux bonding 802.3ad implementation
1913 distributes traffic by peer (using an XOR of MAC addresses),
1914 so in a "gatewayed" configuration, all outgoing traffic will
1915 generally use the same device. Incoming traffic may also end
1916 up on a single device, but that is dependent upon the
1917 balancing policy of the peer's 8023.ad implementation. In a
1918 "local" configuration, traffic will be distributed across the
1919 devices in the bond.
1920
1921 Finally, the 802.3ad mode mandates the use of the MII monitor,
1922 therefore, the ARP monitor is not available in this mode.
1923
1924balance-tlb: The balance-tlb mode balances outgoing traffic by peer.
1925 Since the balancing is done according to MAC address, in a
1926 "gatewayed" configuration (as described above), this mode will
1927 send all traffic across a single device. However, in a
1928 "local" network configuration, this mode balances multiple
1929 local network peers across devices in a vaguely intelligent
1930 manner (not a simple XOR as in balance-xor or 802.3ad mode),
1931 so that mathematically unlucky MAC addresses (i.e., ones that
1932 XOR to the same value) will not all "bunch up" on a single
1933 interface.
1934
1935 Unlike 802.3ad, interfaces may be of differing speeds, and no
1936 special switch configuration is required. On the down side,
1937 in this mode all incoming traffic arrives over a single
1938 interface, this mode requires certain ethtool support in the
1939 network device driver of the slave interfaces, and the ARP
1940 monitor is not available.
1941
1942balance-alb: This mode is everything that balance-tlb is, and more.
1943 It has all of the features (and restrictions) of balance-tlb,
1944 and will also balance incoming traffic from local network
1945 peers (as described in the Bonding Module Options section,
1946 above).
1947
1948 The only additional down side to this mode is that the network
1949 device driver must support changing the hardware address while
1950 the device is open.
1951
Auke Kok6224e012006-06-08 11:15:35 -0700195212.1.2 MT Link Monitoring for Single Switch Topology
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001953----------------------------------------------------
1954
1955 The choice of link monitoring may largely depend upon which
1956mode you choose to use. The more advanced load balancing modes do not
1957support the use of the ARP monitor, and are thus restricted to using
1958the MII monitor (which does not provide as high a level of end to end
1959assurance as the ARP monitor).
1960
Auke Kok6224e012006-06-08 11:15:35 -0700196112.2 Maximum Throughput in a Multiple Switch Topology
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001962-----------------------------------------------------
1963
1964 Multiple switches may be utilized to optimize for throughput
1965when they are configured in parallel as part of an isolated network
1966between two or more systems, for example:
1967
1968 +-----------+
1969 | Host A |
1970 +-+---+---+-+
1971 | | |
1972 +--------+ | +---------+
1973 | | |
1974 +------+---+ +-----+----+ +-----+----+
1975 | Switch A | | Switch B | | Switch C |
1976 +------+---+ +-----+----+ +-----+----+
1977 | | |
1978 +--------+ | +---------+
1979 | | |
1980 +-+---+---+-+
1981 | Host B |
1982 +-----------+
1983
1984 In this configuration, the switches are isolated from one
1985another. One reason to employ a topology such as this is for an
1986isolated network with many hosts (a cluster configured for high
1987performance, for example), using multiple smaller switches can be more
1988cost effective than a single larger switch, e.g., on a network with 24
1989hosts, three 24 port switches can be significantly less expensive than
1990a single 72 port switch.
1991
1992 If access beyond the network is required, an individual host
1993can be equipped with an additional network device connected to an
1994external network; this host then additionally acts as a gateway.
1995
Auke Kok6224e012006-06-08 11:15:35 -0700199612.2.1 MT Bonding Mode Selection for Multiple Switch Topology
Jay Vosburgh00354cf2005-07-21 12:18:02 -07001997-------------------------------------------------------------
1998
1999 In actual practice, the bonding mode typically employed in
2000configurations of this type is balance-rr. Historically, in this
2001network configuration, the usual caveats about out of order packet
2002delivery are mitigated by the use of network adapters that do not do
2003any kind of packet coalescing (via the use of NAPI, or because the
2004device itself does not generate interrupts until some number of
2005packets has arrived). When employed in this fashion, the balance-rr
2006mode allows individual connections between two hosts to effectively
2007utilize greater than one interface's bandwidth.
2008
Auke Kok6224e012006-06-08 11:15:35 -0700200912.2.2 MT Link Monitoring for Multiple Switch Topology
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002010------------------------------------------------------
2011
2012 Again, in actual practice, the MII monitor is most often used
2013in this configuration, as performance is given preference over
2014availability. The ARP monitor will function in this topology, but its
2015advantages over the MII monitor are mitigated by the volume of probes
2016needed as the number of systems involved grows (remember that each
2017host in the network is configured with bonding).
2018
Auke Kok6224e012006-06-08 11:15:35 -0700201913. Switch Behavior Issues
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002020==========================
2021
Auke Kok6224e012006-06-08 11:15:35 -0700202213.1 Link Establishment and Failover Delays
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002023-------------------------------------------
2024
2025 Some switches exhibit undesirable behavior with regard to the
2026timing of link up and down reporting by the switch.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002027
2028 First, when a link comes up, some switches may indicate that
2029the link is up (carrier available), but not pass traffic over the
2030interface for some period of time. This delay is typically due to
2031some type of autonegotiation or routing protocol, but may also occur
2032during switch initialization (e.g., during recovery after a switch
2033failure). If you find this to be a problem, specify an appropriate
2034value to the updelay bonding module option to delay the use of the
2035relevant interface(s).
2036
2037 Second, some switches may "bounce" the link state one or more
2038times while a link is changing state. This occurs most commonly while
2039the switch is initializing. Again, an appropriate updelay value may
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002040help.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002041
2042 Note that when a bonding interface has no active links, the
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002043driver will immediately reuse the first link that goes up, even if the
2044updelay parameter has been specified (the updelay is ignored in this
2045case). If there are slave interfaces waiting for the updelay timeout
2046to expire, the interface that first went into that state will be
2047immediately reused. This reduces down time of the network if the
2048value of updelay has been overestimated, and since this occurs only in
2049cases with no connectivity, there is no additional penalty for
2050ignoring the updelay.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002051
2052 In addition to the concerns about switch timings, if your
2053switches take a long time to go into backup mode, it may be desirable
2054to not activate a backup interface immediately after a link goes down.
2055Failover may be delayed via the downdelay bonding module option.
2056
Auke Kok6224e012006-06-08 11:15:35 -0700205713.2 Duplicated Incoming Packets
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002058--------------------------------
2059
Jay Vosburgh9a6c6862007-11-13 20:25:48 -08002060 NOTE: Starting with version 3.0.2, the bonding driver has logic to
2061suppress duplicate packets, which should largely eliminate this problem.
2062The following description is kept for reference.
2063
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002064 It is not uncommon to observe a short burst of duplicated
2065traffic when the bonding device is first used, or after it has been
2066idle for some period of time. This is most easily observed by issuing
2067a "ping" to some other host on the network, and noticing that the
2068output from ping flags duplicates (typically one per slave).
2069
2070 For example, on a bond in active-backup mode with five slaves
2071all connected to one switch, the output may appear as follows:
2072
2073# ping -n 10.0.4.2
2074PING 10.0.4.2 (10.0.4.2) from 10.0.3.10 : 56(84) bytes of data.
207564 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.7 ms
207664 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.8 ms (DUP!)
207764 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.8 ms (DUP!)
207864 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.8 ms (DUP!)
207964 bytes from 10.0.4.2: icmp_seq=1 ttl=64 time=13.8 ms (DUP!)
208064 bytes from 10.0.4.2: icmp_seq=2 ttl=64 time=0.216 ms
208164 bytes from 10.0.4.2: icmp_seq=3 ttl=64 time=0.267 ms
208264 bytes from 10.0.4.2: icmp_seq=4 ttl=64 time=0.222 ms
2083
2084 This is not due to an error in the bonding driver, rather, it
2085is a side effect of how many switches update their MAC forwarding
2086tables. Initially, the switch does not associate the MAC address in
2087the packet with a particular switch port, and so it may send the
2088traffic to all ports until its MAC forwarding table is updated. Since
2089the interfaces attached to the bond may occupy multiple ports on a
2090single switch, when the switch (temporarily) floods the traffic to all
2091ports, the bond device receives multiple copies of the same packet
2092(one per slave device).
2093
2094 The duplicated packet behavior is switch dependent, some
2095switches exhibit this, and some do not. On switches that display this
2096behavior, it can be induced by clearing the MAC forwarding table (on
2097most Cisco switches, the privileged command "clear mac address-table
2098dynamic" will accomplish this).
2099
Auke Kok6224e012006-06-08 11:15:35 -0700210014. Hardware Specific Considerations
Linus Torvalds1da177e2005-04-16 15:20:36 -07002101====================================
2102
2103 This section contains additional information for configuring
2104bonding on specific hardware platforms, or for interfacing bonding
2105with particular switches or other devices.
2106
Auke Kok6224e012006-06-08 11:15:35 -0700210714.1 IBM BladeCenter
Linus Torvalds1da177e2005-04-16 15:20:36 -07002108--------------------
2109
2110 This applies to the JS20 and similar systems.
2111
2112 On the JS20 blades, the bonding driver supports only
2113balance-rr, active-backup, balance-tlb and balance-alb modes. This is
2114largely due to the network topology inside the BladeCenter, detailed
2115below.
2116
2117JS20 network adapter information
2118--------------------------------
2119
2120 All JS20s come with two Broadcom Gigabit Ethernet ports
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002121integrated on the planar (that's "motherboard" in IBM-speak). In the
2122BladeCenter chassis, the eth0 port of all JS20 blades is hard wired to
2123I/O Module #1; similarly, all eth1 ports are wired to I/O Module #2.
2124An add-on Broadcom daughter card can be installed on a JS20 to provide
2125two more Gigabit Ethernet ports. These ports, eth2 and eth3, are
2126wired to I/O Modules 3 and 4, respectively.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002127
2128 Each I/O Module may contain either a switch or a passthrough
2129module (which allows ports to be directly connected to an external
2130switch). Some bonding modes require a specific BladeCenter internal
2131network topology in order to function; these are detailed below.
2132
2133 Additional BladeCenter-specific networking information can be
2134found in two IBM Redbooks (www.ibm.com/redbooks):
2135
2136"IBM eServer BladeCenter Networking Options"
2137"IBM eServer BladeCenter Layer 2-7 Network Switching"
2138
2139BladeCenter networking configuration
2140------------------------------------
2141
2142 Because a BladeCenter can be configured in a very large number
2143of ways, this discussion will be confined to describing basic
2144configurations.
2145
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002146 Normally, Ethernet Switch Modules (ESMs) are used in I/O
Linus Torvalds1da177e2005-04-16 15:20:36 -07002147modules 1 and 2. In this configuration, the eth0 and eth1 ports of a
2148JS20 will be connected to different internal switches (in the
2149respective I/O modules).
2150
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002151 A passthrough module (OPM or CPM, optical or copper,
2152passthrough module) connects the I/O module directly to an external
2153switch. By using PMs in I/O module #1 and #2, the eth0 and eth1
2154interfaces of a JS20 can be redirected to the outside world and
2155connected to a common external switch.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002156
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002157 Depending upon the mix of ESMs and PMs, the network will
2158appear to bonding as either a single switch topology (all PMs) or as a
2159multiple switch topology (one or more ESMs, zero or more PMs). It is
2160also possible to connect ESMs together, resulting in a configuration
2161much like the example in "High Availability in a Multiple Switch
2162Topology," above.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002163
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002164Requirements for specific modes
2165-------------------------------
Linus Torvalds1da177e2005-04-16 15:20:36 -07002166
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002167 The balance-rr mode requires the use of passthrough modules
2168for devices in the bond, all connected to an common external switch.
2169That switch must be configured for "etherchannel" or "trunking" on the
Linus Torvalds1da177e2005-04-16 15:20:36 -07002170appropriate ports, as is usual for balance-rr.
2171
2172 The balance-alb and balance-tlb modes will function with
2173either switch modules or passthrough modules (or a mix). The only
2174specific requirement for these modes is that all network interfaces
2175must be able to reach all destinations for traffic sent over the
2176bonding device (i.e., the network must converge at some point outside
2177the BladeCenter).
2178
2179 The active-backup mode has no additional requirements.
2180
2181Link monitoring issues
2182----------------------
2183
2184 When an Ethernet Switch Module is in place, only the ARP
2185monitor will reliably detect link loss to an external switch. This is
2186nothing unusual, but examination of the BladeCenter cabinet would
2187suggest that the "external" network ports are the ethernet ports for
2188the system, when it fact there is a switch between these "external"
2189ports and the devices on the JS20 system itself. The MII monitor is
2190only able to detect link failures between the ESM and the JS20 system.
2191
2192 When a passthrough module is in place, the MII monitor does
2193detect failures to the "external" port, which is then directly
2194connected to the JS20 system.
2195
2196Other concerns
2197--------------
2198
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002199 The Serial Over LAN (SoL) link is established over the primary
Linus Torvalds1da177e2005-04-16 15:20:36 -07002200ethernet (eth0) only, therefore, any loss of link to eth0 will result
2201in losing your SoL connection. It will not fail over with other
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002202network traffic, as the SoL system is beyond the control of the
2203bonding driver.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002204
2205 It may be desirable to disable spanning tree on the switch
2206(either the internal Ethernet Switch Module, or an external switch) to
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002207avoid fail-over delay issues when using bonding.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002208
2209
Auke Kok6224e012006-06-08 11:15:35 -0700221015. Frequently Asked Questions
Linus Torvalds1da177e2005-04-16 15:20:36 -07002211==============================
2212
22131. Is it SMP safe?
2214
2215 Yes. The old 2.0.xx channel bonding patch was not SMP safe.
2216The new driver was designed to be SMP safe from the start.
2217
22182. What type of cards will work with it?
2219
2220 Any Ethernet type cards (you can even mix cards - a Intel
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002221EtherExpress PRO/100 and a 3com 3c905b, for example). For most modes,
2222devices need not be of the same speed.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002223
Jay Vosburgh9a6c6862007-11-13 20:25:48 -08002224 Starting with version 3.2.1, bonding also supports Infiniband
2225slaves in active-backup mode.
2226
Linus Torvalds1da177e2005-04-16 15:20:36 -070022273. How many bonding devices can I have?
2228
2229 There is no limit.
2230
22314. How many slaves can a bonding device have?
2232
2233 This is limited only by the number of network interfaces Linux
2234supports and/or the number of network cards you can place in your
2235system.
2236
22375. What happens when a slave link dies?
2238
2239 If link monitoring is enabled, then the failing device will be
2240disabled. The active-backup mode will fail over to a backup link, and
2241other modes will ignore the failed link. The link will continue to be
2242monitored, and should it recover, it will rejoin the bond (in whatever
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002243manner is appropriate for the mode). See the sections on High
2244Availability and the documentation for each mode for additional
2245information.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002246
2247 Link monitoring can be enabled via either the miimon or
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002248arp_interval parameters (described in the module parameters section,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002249above). In general, miimon monitors the carrier state as sensed by
2250the underlying network device, and the arp monitor (arp_interval)
2251monitors connectivity to another host on the local network.
2252
2253 If no link monitoring is configured, the bonding driver will
2254be unable to detect link failures, and will assume that all links are
2255always available. This will likely result in lost packets, and a
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002256resulting degradation of performance. The precise performance loss
Linus Torvalds1da177e2005-04-16 15:20:36 -07002257depends upon the bonding mode and network configuration.
2258
22596. Can bonding be used for High Availability?
2260
2261 Yes. See the section on High Availability for details.
2262
22637. Which switches/systems does it work with?
2264
2265 The full answer to this depends upon the desired mode.
2266
2267 In the basic balance modes (balance-rr and balance-xor), it
2268works with any system that supports etherchannel (also called
2269trunking). Most managed switches currently available have such
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002270support, and many unmanaged switches as well.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002271
2272 The advanced balance modes (balance-tlb and balance-alb) do
2273not have special switch requirements, but do need device drivers that
2274support specific features (described in the appropriate section under
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002275module parameters, above).
Linus Torvalds1da177e2005-04-16 15:20:36 -07002276
Auke Kok6224e012006-06-08 11:15:35 -07002277 In 802.3ad mode, it works with systems that support IEEE
Linus Torvalds1da177e2005-04-16 15:20:36 -07002278802.3ad Dynamic Link Aggregation. Most managed and many unmanaged
2279switches currently available support 802.3ad.
2280
2281 The active-backup mode should work with any Layer-II switch.
2282
22838. Where does a bonding device get its MAC address from?
2284
Jay Vosburgh9a6c6862007-11-13 20:25:48 -08002285 When using slave devices that have fixed MAC addresses, or when
2286the fail_over_mac option is enabled, the bonding device's MAC address is
2287the MAC address of the active slave.
2288
2289 For other configurations, if not explicitly configured (with
2290ifconfig or ip link), the MAC address of the bonding device is taken from
2291its first slave device. This MAC address is then passed to all following
2292slaves and remains persistent (even if the first slave is removed) until
2293the bonding device is brought down or reconfigured.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002294
2295 If you wish to change the MAC address, you can set it with
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002296ifconfig or ip link:
Linus Torvalds1da177e2005-04-16 15:20:36 -07002297
2298# ifconfig bond0 hw ether 00:11:22:33:44:55
2299
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002300# ip link set bond0 address 66:77:88:99:aa:bb
2301
Linus Torvalds1da177e2005-04-16 15:20:36 -07002302 The MAC address can be also changed by bringing down/up the
2303device and then changing its slaves (or their order):
2304
2305# ifconfig bond0 down ; modprobe -r bonding
2306# ifconfig bond0 .... up
2307# ifenslave bond0 eth...
2308
2309 This method will automatically take the address from the next
2310slave that is added.
2311
2312 To restore your slaves' MAC addresses, you need to detach them
2313from the bond (`ifenslave -d bond0 eth0'). The bonding driver will
2314then restore the MAC addresses that the slaves had before they were
2315enslaved.
2316
Jay Vosburgh00354cf2005-07-21 12:18:02 -0700231716. Resources and Links
Linus Torvalds1da177e2005-04-16 15:20:36 -07002318=======================
2319
2320The latest version of the bonding driver can be found in the latest
2321version of the linux kernel, found on http://kernel.org
2322
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002323The latest version of this document can be found in either the latest
2324kernel source (named Documentation/networking/bonding.txt), or on the
2325bonding sourceforge site:
2326
2327http://www.sourceforge.net/projects/bonding
2328
Linus Torvalds1da177e2005-04-16 15:20:36 -07002329Discussions regarding the bonding driver take place primarily on the
2330bonding-devel mailing list, hosted at sourceforge.net. If you have
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002331questions or problems, post them to the list. The list address is:
Linus Torvalds1da177e2005-04-16 15:20:36 -07002332
2333bonding-devel@lists.sourceforge.net
2334
Jay Vosburgh00354cf2005-07-21 12:18:02 -07002335 The administrative interface (to subscribe or unsubscribe) can
2336be found at:
2337
Linus Torvalds1da177e2005-04-16 15:20:36 -07002338https://lists.sourceforge.net/lists/listinfo/bonding-devel
2339
Linus Torvalds1da177e2005-04-16 15:20:36 -07002340Donald Becker's Ethernet Drivers and diag programs may be found at :
2341 - http://www.scyld.com/network/
2342
2343You will also find a lot of information regarding Ethernet, NWay, MII,
2344etc. at www.scyld.com.
2345
2346-- END --