1TEAMD.CONF(5) Team daemon configuration TEAMD.CONF(5)
2
3
4
6 teamd.conf — libteam daemon configuration file
7
9 teamd uses JSON format configuration.
10
12 device (string)
13 Desired name of new team device.
14
15 debug_level (int)
16 Level of debug messages. The higher it is the more debug mes‐
17 sages will be printed. It is the same as adding "-g" command
18 line options.
19
20 Default: 0 (disabled)
21
22 hwaddr (string)
23 Desired hardware address of new team device. Usual MAC address
24 format is accepted.
25
26 runner.name (string)
27 Name of team device. The following runners are available:
28
29 broadcast — Simple runner which directs the team device to
30 transmit packets via all ports.
31
32 roundrobin — Simple runner which directs the team device to
33 transmits packets in a round-robin fashion.
34
35 random — Simple runner which directs the team device to trans‐
36 mits packets on a randomly selected port.
37
38 activebackup — Watches for link changes and selects active port
39 to be used for data transfers.
40
41 loadbalance — To do passive load balancing, runner only sets up
42 BPF hash function which will determine port for packet transmit.
43 To do active load balancing, runner moves hashes among available
44 ports trying to reach perfect balance.
45
46 lacp — Implements 802.3ad LACP protocol. Can use same Tx port
47 selection possibilities as loadbalance runner.
48
49 notify_peers.count (int)
50 Number of bursts of unsolicited NAs and gratuitous ARP packets
51 sent after port is enabled or disabled.
52
53 Default: 0 (disabled)
54
55 Default for activebackup runner: 1
56
57 notify_peers.interval (int)
58 Value is positive number in milliseconds. Specifies an interval
59 between bursts of notify-peer packets.
60
61 Default: 0
62
63 mcast_rejoin.count (int)
64 Number of bursts of multicast group rejoin requests sent after
65 port is enabled or disabled.
66
67 Default: 0 (disabled)
68
69 Default for activebackup runner: 1
70
71 mcast_rejoin.interval (int)
72 Value is positive number in milliseconds. Specifies an interval
73 between bursts of multicast group rejoin requests.
74
75 Default: 0
76
77 link_watch.name | ports.PORTIFNAME.link_watch.name (string)
78 Name of link watcher to be used. The following link watchers are
79 available:
80
81 ethtool — Uses Libteam lib to get port ethtool state changes.
82
83 arp_ping — ARP requests are sent through a port. If an ARP reply
84 is received, the link is considered to be up.
85
86 nsna_ping — Similar to the previous, except that it uses IPv6
87 Neighbor Solicitation / Neighbor Advertisement mechanism. This
88 is an alternative to arp_ping and becomes handy in pure-IPv6
89 environments.
90
91 ports (object)
92 List of ports, network devices, to be used in a team device.
93
94 See examples for more information.
95
96 ports.PORTIFNAME.queue_id (int)
97 ID of queue which this port should be mapped to.
98
99 Default: None
100
102 runner.hwaddr_policy (string)
103 This defines the policy of how hardware addresses of team device
104 and port devices should be set during the team lifetime. The
105 following are available:
106
107 same_all — All ports will always have the same hardware address
108 as the associated team device.
109
110 by_active — Team device adopts the hardware address of the cur‐
111 rently active port. This is useful when the port device is not
112 able to change its hardware address.
113
114 only_active — Only the active port adopts the hardware address
115 of the team device. The others have their own.
116
117 Default: same_all
118
119 ports.PORTIFNAME.prio (int)
120 Port priority. The higher number means higher priority.
121
122 Default: 0
123
124 ports.PORTIFNAME.sticky (bool)
125 Flag which indicates if the port is sticky. If set, it means the
126 port does not get unselected if another port with higher prior‐
127 ity or better parameters becomes available.
128
129 Default: false
130
132 runner.tx_hash (array)
133 List of fragment types (strings) which should be used for packet
134 Tx hash computation. The following are available:
135
136 eth — Uses source and destination MAC addresses.
137
138 vlan — Uses VLAN id.
139
140 ipv4 — Uses source and destination IPv4 addresses.
141
142 ipv6 — Uses source and destination IPv6 addresses.
143
144 ip — Uses source and destination IPv4 and IPv6 addresses.
145
146 l3 — Uses source and destination IPv4 and IPv6 addresses.
147
148 tcp — Uses source and destination TCP ports.
149
150 udp — Uses source and destination UDP ports.
151
152 sctp — Uses source and destination SCTP ports.
153
154 l4 — Uses source and destination TCP and UDP and SCTP ports.
155
156 runner.tx_balancer.name (string)
157 Name of active Tx balancer. Active Tx balancing is disabled by
158 default. The only value available is basic.
159
160 Default: None
161
162 runner.tx_balancer.balancing_interval (int)
163 In tenths of a second. Periodic interval between rebalancing.
164
165 Default: 50
166
168 runner.active (bool)
169 If active is true LACPDU frames are sent along the configured
170 links periodically. If not, it acts as "speak when spoken to".
171
172 Default: true
173
174 runner.fast_rate (bool)
175 Option specifies the rate at which our link partner is asked to
176 transmit LACPDU packets. If this is true then packets will be
177 sent once per second. Otherwise they will be sent every 30 sec‐
178 onds.
179
180 runner.tx_hash (array)
181 Same as for load balance runner.
182
183 runner.tx_balancer.name (string)
184 Same as for load balance runner.
185
186 runner.tx_balancer.balancing_interval (int)
187 Same as for load balance runner.
188
189 runner.sys_prio (int)
190 System priority, value can be 0 – 65535.
191
192 Default: 65535
193
194 runner.min_ports (int)
195 Specifies the minimum number of ports that must be active before
196 asserting carrier in the master interface, value can be 1 – 255.
197
198 Default: 0
199
200 runner.agg_select_policy (string)
201 This selects the policy of how the aggregators will be selected.
202 The following are available:
203
204 lacp_prio — Aggregator with highest priority according to LACP
205 standard will be selected. Aggregator priority is affected by
206 per-port option lacp_prio.
207
208 lacp_prio_stable — Same as previous one, except do not replace
209 selected aggregator if it is still usable.
210
211 bandwidth — Select aggregator with highest total bandwidth.
212
213 count — Select aggregator with highest number of ports.
214
215 port_config — Aggregator with highest priority according to per-
216 port options prio and sticky will be selected. This means that
217 the aggregator containing the port with the highest priority
218 will be selected unless at least one of the ports in the cur‐
219 rently selected aggregator is sticky.
220
221 Default: lacp_prio
222
223 ports.PORTIFNAME.lacp_prio (int)
224 Port priority according to LACP standard. The lower number means
225 higher priority.
226
227 Default: 255
228
229 ports.PORTIFNAME.lacp_key (int)
230 Port key according to LACP standard. It is only possible to
231 aggregate ports with the same key.
232
233 Default: 0
234
236 link_watch.delay_up | ports.PORTIFNAME.link_watch.delay_up (int)
237 Value is a positive number in milliseconds. It is the delay
238 between the link coming up and the runner being notified about
239 it.
240
241 Default: 0
242
243 link_watch.delay_down | ports.PORTIFNAME.link_watch.delay_down (int)
244 Value is a positive number in milliseconds. It is the delay
245 between the link going down and the runner being notified about
246 it.
247
248 Default: 0
249
251 link_watch.interval | ports.PORTIFNAME.link_watch.interval (int)
252 Value is a positive number in milliseconds. It is the interval
253 between ARP requests being sent.
254
255 link_watch.init_wait | ports.PORTIFNAME.link_watch.init_wait (int)
256 Value is a positive number in milliseconds. It is the delay
257 between link watch initialization and the first ARP request
258 being sent.
259
260 Default: 0
261
262 link_watch.missed_max | ports.PORTIFNAME.link_watch.missed_max (int)
263 Maximum number of missed ARP replies. If this number is
264 exceeded, link is reported as down.
265
266 Default: 3
267
268 link_watch.source_host | ports.PORTIFNAME.link_watch.source_host (host‐
269 name)
270 Hostname to be converted to IP address which will be filled into
271 ARP request as source address.
272
273 Default: 0.0.0.0
274
275 link_watch.target_host | ports.PORTIFNAME.link_watch.target_host (host‐
276 name)
277 Hostname to be converted to IP address which will be filled into
278 ARP request as destination address.
279
280 link_watch.validate_active | ports.PORTIFNAME.link_watch.vali‐
281 date_active (bool)
282 Validate received ARP packets on active ports. If this is not
283 set, all incoming ARP packets will be considered as a good
284 reply.
285
286 Default: false
287
288 link_watch.validate_inactive | ports.PORTIFNAME.link_watch.vali‐
289 date_inactive (bool)
290 Validate received ARP packets on inactive ports. If this is not
291 set, all incoming ARP packets will be considered as a good
292 reply.
293
294 Default: false
295
296 link_watch.vlanid | ports.PORTIFNAME.link_watch.vlanid (int)
297 By default, ARP requests are sent without VLAN tags. This option
298 causes outgoing ARP requests to be sent with the specified VLAN
299 ID number.
300
301 Default: None
302
303 link_watch.send_always | ports.PORTIFNAME.link_watch.send_always (bool)
304 By default, ARP requests are sent on active ports only. This
305 option allows sending even on inactive ports.
306
307 Default: false
308
310 link_watch.interval | ports.PORTIFNAME.link_watch.interval (int)
311 Value is a positive number in milliseconds. It is the interval
312 between sending NS packets.
313
314 link_watch.init_wait | ports.PORTIFNAME.link_watch.init_wait (int)
315 Value is a positive number in milliseconds. It is the delay
316 between link watch initialization and the first NS packet being
317 sent.
318
319 link_watch.missed_max | ports.PORTIFNAME.link_watch.missed_max (int)
320 Maximum number of missed NA reply packets. If this number is
321 exceeded, link is reported as down.
322
323 Default: 3
324
325 link_watch.target_host | ports.PORTIFNAME.link_watch.target_host (host‐
326 name)
327 Hostname to be converted to IPv6 address which will be filled
328 into NS packet as target address.
329
331 {
332 "device": "team0",
333 "runner": {"name": "roundrobin"},
334 "ports": {"eth1": {}, "eth2": {}}
335 }
336
337 Very basic configuration.
338
339 {
340 "device": "team0",
341 "runner": {"name": "activebackup"},
342 "link_watch": {"name": "ethtool"},
343 "ports": {
344 "eth1": {
345 "prio": -10,
346 "sticky": true
347 },
348 "eth2": {
349 "prio": 100
350 }
351 }
352 }
353
354 This configuration uses active-backup runner with ethtool link watcher.
355 Port eth2 has higher priority, but the sticky flag ensures that if eth1
356 becomes active, it stays active while the link remains up.
357
358 {
359 "device": "team0",
360 "runner": {"name": "activebackup"},
361 "link_watch": {
362 "name": "ethtool",
363 "delay_up": 2500,
364 "delay_down": 1000
365 },
366 "ports": {
367 "eth1": {
368 "prio": -10,
369 "sticky": true
370 },
371 "eth2": {
372 "prio": 100
373 }
374 }
375 }
376
377 Similar to the previous one. Only difference is that link changes are
378 not propagated to the runner immediately, but delays are applied.
379
380 {
381 "device": "team0",
382 "runner": {"name": "activebackup"},
383 "link_watch": {
384 "name": "arp_ping",
385 "interval": 100,
386 "missed_max": 30,
387 "target_host": "192.168.23.1"
388 },
389 "ports": {
390 "eth1": {
391 "prio": -10,
392 "sticky": true
393 },
394 "eth2": {
395 "prio": 100
396 }
397 }
398 }
399
400 This configuration uses ARP ping link watch.
401
402 {
403 "device": "team0",
404 "runner": {"name": "activebackup"},
405 "link_watch": [
406 {
407 "name": "arp_ping",
408 "interval": 100,
409 "missed_max": 30,
410 "target_host": "192.168.23.1"
411 },
412 {
413 "name": "arp_ping",
414 "interval": 50,
415 "missed_max": 20,
416 "target_host": "192.168.24.1"
417 }
418 ],
419 "ports": {
420 "eth1": {
421 "prio": -10,
422 "sticky": true
423 },
424 "eth2": {
425 "prio": 100
426 }
427 }
428 }
429
430 Similar to the previous one, only this time two link watchers are used
431 at the same time.
432
433 {
434 "device": "team0",
435 "runner": {
436 "name": "loadbalance",
437 "tx_hash": ["eth", "ipv4", "ipv6"]
438 },
439 "ports": {"eth1": {}, "eth2": {}}
440 }
441
442 Configuration for hash-based passive Tx load balancing.
443
444 {
445 "device": "team0",
446 "runner": {
447 "name": "loadbalance",
448 "tx_hash": ["eth", "ipv4", "ipv6"],
449 "tx_balancer": {
450 "name": "basic"
451 }
452 },
453 "ports": {"eth1": {}, "eth2": {}}
454 }
455
456 Configuration for active Tx load balancing using basic load balancer.
457
458 {
459 "device": "team0",
460 "runner": {
461 "name": "lacp",
462 "active": true,
463 "fast_rate": true,
464 "tx_hash": ["eth", "ipv4", "ipv6"]
465 },
466 "link_watch": {"name": "ethtool"},
467 "ports": {"eth1": {}, "eth2": {}}
468 }
469
470 Configuration for connection to LACP capable counterpart.
471
473 teamd(8), teamdctl(8), teamnl(8), bond2team(1)
474
476 Jiri Pirko is the original author and current maintainer of libteam.
477
478
479
480libteam 2013-07-09 TEAMD.CONF(5)