]>
Commit | Line | Data |
---|---|---|
eeecce05 BP |
1 | Open vSwitch Advanced Features Tutorial |
2 | ======================================= | |
3 | ||
4 | Many tutorials cover the basics of OpenFlow. This is not such a | |
5 | tutorial. Rather, a knowledge of the basics of OpenFlow is a | |
6 | prerequisite. If you do not already understand how an OpenFlow flow | |
7 | table works, please go read a basic tutorial and then continue reading | |
8 | here afterward. | |
9 | ||
10 | It is also important to understand the basics of Open vSwitch before | |
11 | you begin. If you have never used ovs-vsctl or ovs-ofctl before, you | |
12 | should learn a little about them before proceeding. | |
13 | ||
14 | Most of the features covered in this tutorial are Open vSwitch | |
15 | extensions to OpenFlow. Also, most of the features in this tutorial | |
16 | are specific to the software Open vSwitch implementation. If you are | |
17 | using an Open vSwitch port to an ASIC-based hardware switch, this | |
18 | tutorial will not help you. | |
19 | ||
20 | This tutorial does not cover every aspect of the features that it | |
21 | mentions. You can find the details elsewhere in the Open vSwitch | |
22 | documentation, especially ovs-ofctl(8) and the comments in the | |
23 | include/openflow/nicira-ext.h header file. | |
24 | ||
25 | >>> In this tutorial, paragraphs set off like this designate notes | |
26 | with additional information that readers may wish to skip on a | |
27 | first read. | |
28 | ||
29 | ||
30 | Getting Started | |
31 | =============== | |
32 | ||
33 | This is a hands-on tutorial. To get the most out of it, you will need | |
34 | Open vSwitch binaries. You do not, on the other hand, need any | |
35 | physical networking hardware or even supervisor privilege on your | |
36 | system. Instead, we will use a script called "ovs-sandbox", which | |
37 | accompanies the tutorial, that constructs a software simulated network | |
38 | environment based on Open vSwitch. | |
39 | ||
40 | You can use "ovs-sandbox" three ways: | |
41 | ||
42 | * If you have already installed Open vSwitch on your system, then | |
43 | you should be able to just run "ovs-sandbox" from this directory | |
44 | without any options. | |
45 | ||
46 | * If you have not installed Open vSwitch (and you do not want to | |
47 | install it), then you can build Open vSwitch according to the | |
48 | instructions in INSTALL, without installing it. Then run | |
49 | "./ovs-sandbox -b DIRECTORY" from this directory, substituting | |
50 | the Open vSwitch build directory for DIRECTORY. | |
51 | ||
52 | * As a slight variant on the latter, you can run "make sandbox" | |
53 | from an Open vSwitch build directory. | |
54 | ||
55 | When you run ovs-sandbox, it does the following: | |
56 | ||
57 | 1. CAUTION: Deletes any subdirectory of the current directory | |
58 | named "sandbox" and any files in that directory. | |
59 | ||
60 | 2. Creates a new directory "sandbox" in the current directory. | |
61 | ||
62 | 3. Sets up special environment variables that ensure that Open | |
63 | vSwitch programs will look inside the "sandbox" directory | |
64 | instead of in the Open vSwitch installation directory. | |
65 | ||
66 | 4. If you are using a built but not installed Open vSwitch, | |
67 | installs the Open vSwitch manpages in a subdirectory of | |
68 | "sandbox" and adjusts the MANPATH environment variable to point | |
69 | to this directory. This means that you can use, for example, | |
70 | "man ovs-vsctl" to see a manpage for the ovs-vsctl program that | |
71 | you built. | |
72 | ||
73 | 5. Creates an empty Open vSwitch configuration database under | |
74 | "sandbox". | |
75 | ||
76 | 6. Starts ovsdb-server running under "sandbox". | |
77 | ||
78 | 7. Starts ovs-vswitchd running under "sandbox", passing special | |
79 | options that enable a special "dummy" mode for testing. | |
80 | ||
81 | 8. Starts a nested interactive shell inside "sandbox". | |
82 | ||
83 | At this point, you can run all the usual Open vSwitch utilities from | |
84 | the nested shell environment. You can, for example, use ovs-vsctl to | |
85 | create a bridge: | |
86 | ||
87 | ovs-vsctl add-br br0 | |
88 | ||
89 | From Open vSwitch's perspective, the bridge that you create this way | |
90 | is as real as any other. You can, for example, connect it to an | |
91 | OpenFlow controller or use "ovs-ofctl" to examine and modify it and | |
92 | its OpenFlow flow table. On the other hand, the bridge is not visible | |
93 | to the operating system's network stack, so "ifconfig" or "ip" cannot | |
94 | see it or affect it, which means that utilities like "ping" and | |
95 | "tcpdump" will not work either. (That has its good side, too: you | |
96 | can't screw up your computer's network stack by manipulating a | |
97 | sandboxed OVS.) | |
98 | ||
99 | When you're done using OVS from the sandbox, exit the nested shell (by | |
100 | entering the "exit" shell command or pressing Control+D). This will | |
101 | kill the daemons that ovs-sandbox started, but it leaves the "sandbox" | |
102 | directory and its contents in place. | |
103 | ||
104 | The sandbox directory contains log files for the Open vSwitch dameons. | |
105 | You can examine them while you're running in the sandboxed environment | |
106 | or after you exit. | |
107 | ||
108 | ||
109 | Motivation | |
110 | ========== | |
111 | ||
112 | The goal of this tutorial is to demonstrate the power of Open vSwitch | |
113 | flow tables. The tutorial works through the implementation of a | |
114 | MAC-learning switch with VLAN trunk and access ports. Outside of the | |
115 | Open vSwitch features that we will discuss, OpenFlow provides at least | |
116 | two ways to implement such a switch: | |
117 | ||
118 | 1. An OpenFlow controller to implement MAC learning in a | |
119 | "reactive" fashion. Whenever a new MAC appears on the switch, | |
120 | or a MAC moves from one switch port to another, the controller | |
121 | adjusts the OpenFlow flow table to match. | |
122 | ||
123 | 2. The "normal" action. OpenFlow defines this action to submit a | |
124 | packet to "the traditional non-OpenFlow pipeline of the | |
125 | switch". That is, if a flow uses this action, then the packets | |
126 | in the flow go through the switch in the same way that they | |
127 | would if OpenFlow was not configured on the switch. | |
128 | ||
129 | Each of these approaches has unfortunate pitfalls. In the first | |
130 | approach, using an OpenFlow controller to implement MAC learning, has | |
131 | a significant cost in terms of network bandwidth and latency. It also | |
132 | makes the controller more difficult to scale to large numbers of | |
133 | switches, which is especially important in environments with thousands | |
134 | of hypervisors (each of which contains a virtual OpenFlow switch). | |
135 | MAC learning at an OpenFlow controller also behaves poorly if the | |
136 | OpenFlow controller fails, slows down, or becomes unavailable due to | |
137 | network problems. | |
138 | ||
139 | The second approach, using the "normal" action, has different | |
140 | problems. First, little about the "normal" action is standardized, so | |
141 | it behaves differently on switches from different vendors, and the | |
142 | available features and how those features are configured (usually not | |
143 | through OpenFlow) varies widely. Second, "normal" does not work well | |
144 | with other OpenFlow actions. It is "all-or-nothing", with little | |
145 | potential to adjust its behavior slightly or to compose it with other | |
146 | features. | |
147 | ||
148 | ||
149 | Scenario | |
150 | ======== | |
151 | ||
152 | We will construct Open vSwitch flow tables for a VLAN-capable, | |
153 | MAC-learning switch that has four ports: | |
154 | ||
155 | * p1, a trunk port that carries all VLANs, on OpenFlow port 1. | |
156 | ||
157 | * p2, an access port for VLAN 20, on OpenFlow port 2. | |
158 | ||
159 | * p3 and p4, both access ports for VLAN 30, on OpenFlow ports 3 | |
160 | and 4, respectively. | |
161 | ||
162 | >>> The ports' names are not significant. You could call them eth1 | |
163 | through eth4, or any other names you like. | |
164 | ||
165 | >>> An OpenFlow switch always has a "local" port as well. This | |
166 | scenario won't use the local port. | |
167 | ||
168 | Our switch design will consist of five main flow tables, each of which | |
169 | implements one stage in the switch pipeline: | |
170 | ||
171 | Table 0: Admission control. | |
172 | ||
173 | Table 1: VLAN input processing. | |
174 | ||
175 | Table 2: Learn source MAC and VLAN for ingress port. | |
176 | ||
177 | Table 3: Look up learned port for destination MAC and VLAN. | |
178 | ||
179 | Table 4: Output processing. | |
180 | ||
181 | The section below describes how to set up the scenario, followed by a | |
182 | section for each OpenFlow table. | |
183 | ||
184 | You can cut and paste the "ovs-vsctl" and "ovs-ofctl" commands in each | |
185 | of the sections below into your "ovs-sandbox" shell. They are also | |
186 | available as shell scripts in this directory, named t-setup, t-stage0, | |
187 | t-stage1, ..., t-stage4. The "ovs-appctl" test commands are intended | |
188 | for cutting and pasting and are not supplied separately. | |
189 | ||
190 | ||
191 | Setup | |
192 | ===== | |
193 | ||
194 | To get started, start "ovs-sandbox". Inside the interactive shell | |
195 | that it starts, run this command: | |
196 | ||
197 | ovs-vsctl add-br br0 -- set Bridge br0 fail-mode=secure | |
198 | ||
199 | This command creates a new bridge "br0" and puts "br0" into so-called | |
200 | "fail-secure" mode. For our purpose, this just means that the | |
201 | OpenFlow flow table starts out empty. | |
202 | ||
203 | >>> If we did not do this, then the flow table would start out with a | |
204 | single flow that executes the "normal" action. We could use that | |
205 | feature to yield a switch that behaves the same as the switch we | |
206 | are currently building, but with the caveats described under | |
207 | "Motivation" above.) | |
208 | ||
209 | The new bridge has only one port on it so far, the "local port" br0. | |
210 | We need to add p1, p2, p3, and p4. A shell "for" loop is one way to | |
211 | do it: | |
212 | ||
213 | for i in 1 2 3 4; do | |
214 | ovs-vsctl add-port br0 p$i -- set Interface p$i ofport_request=$i | |
215 | ovs-ofctl mod-port br0 p$i up | |
216 | done | |
217 | ||
218 | In addition to adding a port, the ovs-vsctl command above sets its | |
219 | "ofport_request" column to ensure that port p1 is assigned OpenFlow | |
220 | port 1, p2 is assigned OpenFlow port 2, and so on. | |
221 | ||
222 | >>> We could omit setting the ofport_request and let Open vSwitch | |
223 | choose port numbers for us, but it's convenient for the purposes | |
224 | of this tutorial because we can talk about OpenFlow port 1 and | |
225 | know that it corresponds to p1. | |
226 | ||
227 | The ovs-ofctl command above brings up the simulated interfaces, which | |
228 | are down initially, using an OpenFlow request. The effect is similar | |
229 | to "ifconfig up", but the sandbox's interfaces are not visible to the | |
230 | operating system and therefore "ifconfig" would not affect them. | |
231 | ||
232 | We have not configured anything related to VLANs or MAC learning. | |
233 | That's because we're going to implement those features in the flow | |
234 | table. | |
235 | ||
236 | To see what we've done so far to set up the scenario, you can run a | |
237 | command like "ovs-vsctl show" or "ovs-ofctl show br0". | |
238 | ||
239 | ||
240 | Implementing Table 0: Admission control | |
241 | ======================================= | |
242 | ||
243 | Table 0 is where packets enter the switch. We use this stage to | |
244 | discard packets that for one reason or another are invalid. For | |
245 | example, packets with a multicast source address are not valid, so we | |
246 | can add a flow to drop them at ingress to the switch with: | |
247 | ||
248 | ovs-ofctl add-flow br0 \ | |
249 | "table=0, dl_src=01:00:00:00:00:00/01:00:00:00:00:00, actions=drop" | |
250 | ||
251 | A switch should also not forward IEEE 802.1D Spanning Tree Protocol | |
252 | (STP) packets, so we can also add a flow to drop those and other | |
253 | packets with reserved multicast protocols: | |
254 | ||
255 | ovs-ofctl add-flow br0 \ | |
f0ac9da9 | 256 | "table=0, dl_dst=01:80:c2:00:00:00/ff:ff:ff:ff:ff:f0, actions=drop" |
eeecce05 BP |
257 | |
258 | We could add flows to drop other protocols, but these demonstrate the | |
259 | pattern. | |
260 | ||
261 | We need one more flow, with a priority lower than the default, so that | |
262 | flows that don't match either of the "drop" flows we added above go on | |
263 | to pipeline stage 1 in OpenFlow table 1: | |
264 | ||
265 | ovs-ofctl add-flow br0 "table=0, priority=0, actions=resubmit(,1)" | |
266 | ||
267 | (The "resubmit" action is an Open vSwitch extension to OpenFlow.) | |
268 | ||
269 | ||
270 | Testing Table 0 | |
271 | --------------- | |
272 | ||
273 | If we were using Open vSwitch to set up a physical or a virtual | |
274 | switch, then we would naturally test it by sending packets through it | |
275 | one way or another, perhaps with common network testing tools like | |
276 | "ping" and "tcpdump" or more specialized tools like Scapy. That's | |
277 | difficult with our simulated switch, since it's not visible to the | |
278 | operating system. | |
279 | ||
4ff8998c | 280 | But our simulated switch has a few specialized testing tools. The |
eeecce05 BP |
281 | most powerful of these tools is "ofproto/trace". Given a switch and |
282 | the specification of a flow, "ofproto/trace" shows, step-by-step, how | |
283 | such a flow would be treated as it goes through the switch. | |
284 | ||
285 | ||
286 | == EXAMPLE 1 == | |
287 | ||
288 | Try this command: | |
289 | ||
f0ac9da9 | 290 | ovs-appctl ofproto/trace br0 in_port=1,dl_dst=01:80:c2:00:00:05 |
eeecce05 BP |
291 | |
292 | The output should look something like this: | |
293 | ||
f0ac9da9 BP |
294 | Flow: metadata=0,in_port=1,vlan_tci=0x0000,dl_src=00:00:00:00:00:00,dl_dst=01:80:c2:00:00:05,dl_type=0x0000 |
295 | Rule: table=0 cookie=0 dl_dst=01:80:c2:00:00:00/ff:ff:ff:ff:ff:f0 | |
eeecce05 BP |
296 | OpenFlow actions=drop |
297 | ||
298 | Final flow: unchanged | |
299 | Datapath actions: drop | |
300 | ||
301 | The first block of lines describes an OpenFlow table lookup. The | |
302 | first line shows the fields used for the table lookup (which is mostly | |
303 | zeros because that's the default if we don't specify everything). The | |
304 | second line gives the OpenFlow flow that the fields matched (called a | |
305 | "rule" because that is the name used inside Open vSwitch for an | |
306 | OpenFlow flow). In this case, we see that this packet that has a | |
307 | reserved multicast destination address matches the rule that drops | |
308 | those packets. The third line gives the rule's OpenFlow actions. | |
309 | ||
310 | The second block of lines summarizes the results, which are not very | |
311 | interesting here. | |
312 | ||
313 | ||
314 | == EXAMPLE 2 == | |
315 | ||
316 | Try another command: | |
317 | ||
f0ac9da9 | 318 | ovs-appctl ofproto/trace br0 in_port=1,dl_dst=01:80:c2:00:00:10 |
eeecce05 BP |
319 | |
320 | The output should be: | |
321 | ||
f0ac9da9 | 322 | Flow: metadata=0,in_port=1,vlan_tci=0x0000,dl_src=00:00:00:00:00:00,dl_dst=01:80:c2:00:00:10,dl_type=0x0000 |
eeecce05 BP |
323 | Rule: table=0 cookie=0 priority=0 |
324 | OpenFlow actions=resubmit(,1) | |
325 | ||
326 | Resubmitted flow: unchanged | |
327 | Resubmitted regs: reg0=0x0 reg1=0x0 reg2=0x0 reg3=0x0 reg4=0x0 reg5=0x0 reg6=0x0 reg7=0x0 | |
328 | Resubmitted odp: drop | |
329 | No match | |
330 | ||
331 | Final flow: unchanged | |
332 | Datapath actions: drop | |
333 | ||
334 | This time the flow we handed to "ofproto/trace" doesn't match any of | |
335 | our "drop" rules, so it falls through to the low-priority "resubmit" | |
336 | rule, which we see in the rule and the actions selected in the first | |
337 | block. The "resubmit" causes a second lookup in OpenFlow table 1, | |
338 | described by the additional block of indented text in the output. We | |
339 | haven't yet added any flows to OpenFlow table 1, so no flow actually | |
340 | matches in the second lookup. Therefore, the packet is still actually | |
341 | dropped, which means that the externally observable results would be | |
342 | identical to our first example. | |
343 | ||
344 | ||
345 | Implementing Table 1: VLAN Input Processing | |
346 | =========================================== | |
347 | ||
348 | A packet that enters table 1 has already passed basic validation in | |
349 | table 0. The purpose of table 1 is validate the packet's VLAN, based | |
350 | on the VLAN configuration of the switch port through which the packet | |
351 | entered the switch. We will also use it to attach a VLAN header to | |
352 | packets that arrive on an access port, which allows later processing | |
353 | stages to rely on the packet's VLAN always being part of the VLAN | |
354 | header, reducing special cases. | |
355 | ||
356 | Let's start by adding a low-priority flow that drops all packets, | |
357 | before we add flows that pass through acceptable packets. You can | |
358 | think of this as a "default drop" rule: | |
359 | ||
360 | ovs-ofctl add-flow br0 "table=1, priority=0, actions=drop" | |
361 | ||
362 | Our trunk port p1, on OpenFlow port 1, is an easy case. p1 accepts | |
363 | any packet regardless of whether it has a VLAN header or what the VLAN | |
364 | was, so we can add a flow that resubmits everything on input port 1 to | |
365 | the next table: | |
366 | ||
367 | ovs-ofctl add-flow br0 \ | |
368 | "table=1, priority=99, in_port=1, actions=resubmit(,2)" | |
369 | ||
370 | On the access ports, we want to accept any packet that has no VLAN | |
371 | header, tag it with the access port's VLAN number, and then pass it | |
372 | along to the next stage: | |
373 | ||
374 | ovs-ofctl add-flows br0 - <<'EOF' | |
375 | table=1, priority=99, in_port=2, vlan_tci=0, actions=mod_vlan_vid:20, resubmit(,2) | |
376 | table=1, priority=99, in_port=3, vlan_tci=0, actions=mod_vlan_vid:30, resubmit(,2) | |
377 | table=1, priority=99, in_port=4, vlan_tci=0, actions=mod_vlan_vid:30, resubmit(,2) | |
378 | EOF | |
379 | ||
380 | We don't write any rules that match packets with 802.1Q that enter | |
381 | this stage on any of the access ports, so the "default drop" rule we | |
382 | added earlier causes them to be dropped, which is ordinarily what we | |
383 | want for access ports. | |
384 | ||
385 | >>> Another variation of access ports allows ingress of packets tagged | |
386 | with VLAN 0 (aka 802.1p priority tagged packets). To allow such | |
387 | packets, replace "vlan_tci=0" by "vlan_tci=0/0xfff" above. | |
388 | ||
389 | ||
390 | Testing Table 1 | |
391 | --------------- | |
392 | ||
393 | "ofproto/trace" allows us to test the ingress VLAN rules that we added | |
394 | above. | |
395 | ||
396 | ||
397 | == EXAMPLE 1: Packet on Trunk Port == | |
398 | ||
399 | Here's a test of a packet coming in on the trunk port: | |
400 | ||
401 | ovs-appctl ofproto/trace br0 in_port=1,vlan_tci=5 | |
402 | ||
403 | The output shows the lookup in table 0, the resubmit to table 1, and | |
404 | the resubmit to table 2 (which does nothing because we haven't put | |
405 | anything there yet): | |
406 | ||
407 | Flow: metadata=0,in_port=1,vlan_tci=0x0005,dl_src=00:00:00:00:00:00,dl_dst=00:00:00:00:00:00,dl_type=0x0000 | |
408 | Rule: table=0 cookie=0 priority=0 | |
409 | OpenFlow actions=resubmit(,1) | |
410 | ||
411 | Resubmitted flow: unchanged | |
412 | Resubmitted regs: reg0=0x0 reg1=0x0 reg2=0x0 reg3=0x0 reg4=0x0 reg5=0x0 reg6=0x0 reg7=0x0 | |
413 | Resubmitted odp: drop | |
414 | Rule: table=1 cookie=0 priority=99,in_port=1 | |
415 | OpenFlow actions=resubmit(,2) | |
416 | ||
417 | Resubmitted flow: unchanged | |
418 | Resubmitted regs: reg0=0x0 reg1=0x0 reg2=0x0 reg3=0x0 reg4=0x0 reg5=0x0 reg6=0x0 reg7=0x0 | |
419 | Resubmitted odp: drop | |
420 | No match | |
421 | ||
422 | Final flow: unchanged | |
423 | Datapath actions: drop | |
424 | ||
425 | ||
426 | == EXAMPLE 2: Valid Packet on Access Port == | |
427 | ||
428 | Here's a test of a valid packet (a packet without an 802.1Q header) | |
429 | coming in on access port p2: | |
430 | ||
431 | ovs-appctl ofproto/trace br0 in_port=2 | |
432 | ||
433 | The output is similar to that for the previous case, except that it | |
434 | additionally tags the packet with p2's VLAN 20 before it passes it | |
435 | along to table 2: | |
436 | ||
437 | Flow: metadata=0,in_port=2,vlan_tci=0x0000,dl_src=00:00:00:00:00:00,dl_dst=00:00:00:00:00:00,dl_type=0x0000 | |
438 | Rule: table=0 cookie=0 priority=0 | |
439 | OpenFlow actions=resubmit(,1) | |
440 | ||
441 | Resubmitted flow: unchanged | |
442 | Resubmitted regs: reg0=0x0 reg1=0x0 reg2=0x0 reg3=0x0 reg4=0x0 reg5=0x0 reg6=0x0 reg7=0x0 | |
443 | Resubmitted odp: drop | |
444 | Rule: table=1 cookie=0 priority=99,in_port=2,vlan_tci=0x0000 | |
445 | OpenFlow actions=mod_vlan_vid:20,resubmit(,2) | |
446 | ||
447 | Resubmitted flow: metadata=0,in_port=2,dl_vlan=20,dl_vlan_pcp=0,dl_src=00:00:00:00:00:00,dl_dst=00:00:00:00:00:00,dl_type=0x0000 | |
448 | Resubmitted regs: reg0=0x0 reg1=0x0 reg2=0x0 reg3=0x0 reg4=0x0 reg5=0x0 reg6=0x0 reg7=0x0 | |
449 | Resubmitted odp: drop | |
450 | No match | |
451 | ||
452 | Final flow: unchanged | |
453 | Datapath actions: drop | |
454 | ||
455 | ||
456 | == EXAMPLE 3: Invalid Packet on Access Port == | |
457 | ||
458 | This tests an invalid packet (one that includes an 802.1Q header) | |
459 | coming in on access port p2: | |
460 | ||
461 | ovs-appctl ofproto/trace br0 in_port=2,vlan_tci=5 | |
462 | ||
463 | The output shows the packet matching the default drop rule: | |
464 | ||
465 | Flow: metadata=0,in_port=2,vlan_tci=0x0005,dl_src=00:00:00:00:00:00,dl_dst=00:00:00:00:00:00,dl_type=0x0000 | |
466 | Rule: table=0 cookie=0 priority=0 | |
467 | OpenFlow actions=resubmit(,1) | |
468 | ||
469 | Resubmitted flow: unchanged | |
470 | Resubmitted regs: reg0=0x0 reg1=0x0 reg2=0x0 reg3=0x0 reg4=0x0 reg5=0x0 reg6=0x0 reg7=0x0 | |
471 | Resubmitted odp: drop | |
472 | Rule: table=1 cookie=0 priority=0 | |
473 | OpenFlow actions=drop | |
474 | ||
475 | Final flow: unchanged | |
476 | Datapath actions: drop | |
477 | ||
478 | ||
479 | Implementing Table 2: MAC+VLAN Learning for Ingress Port | |
480 | ======================================================== | |
481 | ||
482 | This table allows the switch we're implementing to learn that the | |
483 | packet's source MAC is located on the packet's ingress port in the | |
484 | packet's VLAN. | |
485 | ||
486 | >>> This table is a good example why table 1 added a VLAN tag to | |
487 | packets that entered the switch through an access port. We want | |
488 | to associate a MAC+VLAN with a port regardless of whether the VLAN | |
489 | in question was originally part of the packet or whether it was an | |
490 | assumed VLAN associated with an access port. | |
491 | ||
492 | It only takes a single flow to do this. The following command adds | |
493 | it: | |
494 | ||
495 | ovs-ofctl add-flow br0 \ | |
496 | "table=2 actions=learn(table=10, NXM_OF_VLAN_TCI[0..11], \ | |
497 | NXM_OF_ETH_DST[]=NXM_OF_ETH_SRC[], \ | |
498 | load:NXM_OF_IN_PORT[]->NXM_NX_REG0[0..15]), \ | |
499 | resubmit(,3)" | |
500 | ||
501 | The "learn" action (an Open vSwitch extension to OpenFlow) modifies a | |
502 | flow table based on the content of the flow currently being processed. | |
503 | Here's how you can interpret each part of the "learn" action above: | |
504 | ||
505 | table=10 | |
506 | ||
507 | Modify flow table 10. This will be the MAC learning table. | |
508 | ||
509 | NXM_OF_VLAN_TCI[0..11] | |
510 | ||
511 | Make the flow that we add to flow table 10 match the same VLAN | |
512 | ID that the packet we're currently processing contains. This | |
513 | effectively scopes the MAC learning entry to a single VLAN, | |
4ff8998c | 514 | which is the ordinary behavior for a VLAN-aware switch. |
eeecce05 BP |
515 | |
516 | NXM_OF_ETH_DST[]=NXM_OF_ETH_SRC[] | |
517 | ||
518 | Make the flow that we add to flow table 10 match, as Ethernet | |
519 | destination, the Ethernet source address of the packet we're | |
520 | currently processing. | |
521 | ||
522 | load:NXM_OF_IN_PORT[]->NXM_NX_REG0[0..15] | |
523 | ||
524 | Whereas the preceding parts specify fields for the new flow to | |
525 | match, this specifies an action for the flow to take when it | |
526 | matches. The action is for the flow to load the ingress port | |
527 | number of the current packet into register 0 (a special field | |
528 | that is an Open vSwitch extension to OpenFlow). | |
529 | ||
530 | >>> A real use of "learn" for MAC learning would probably involve two | |
531 | additional elements. First, the "learn" action would specify a | |
532 | hard_timeout for the new flow, to enable a learned MAC to | |
533 | eventually expire if no new packets were seen from a given source | |
534 | within a reasonable interval. Second, one would usually want to | |
535 | limit resource consumption by using the Flow_Table table in the | |
536 | Open vSwitch configuration database to specify a maximum number of | |
537 | flows in table 10. | |
538 | ||
539 | This definitely calls for examples. | |
540 | ||
541 | ||
542 | Testing Table 2 | |
543 | --------------- | |
544 | ||
545 | == EXAMPLE 1 == | |
546 | ||
547 | Try the following test command: | |
548 | ||
549 | ovs-appctl ofproto/trace br0 in_port=1,vlan_tci=20,dl_src=50:00:00:00:00:01 -generate | |
550 | ||
551 | The output shows that "learn" was executed, but it isn't otherwise | |
552 | informative, so we won't include it here. | |
553 | ||
554 | The "-generate" keyword is new. Ordinarily, "ofproto/trace" has no | |
555 | side effects: "output" actions do not actually output packets, "learn" | |
556 | actions do not actually modify the flow table, and so on. With | |
557 | "-generate", though, "ofproto/trace" does execute "learn" actions. | |
558 | That's important now, because we want to see the effect of the "learn" | |
559 | action on table 10. You can see that by running: | |
560 | ||
561 | ovs-ofctl dump-flows br0 table=10 | |
562 | ||
563 | which (omitting the "duration" and "idle_age" fields, which will vary | |
564 | based on how soon you ran this command after the previous one, as well | |
565 | as some other uninteresting fields) prints something like: | |
566 | ||
567 | NXST_FLOW reply (xid=0x4): | |
568 | table=10, vlan_tci=0x0014/0x0fff,dl_dst=50:00:00:00:00:01 actions=load:0x1->NXM_NX_REG0[0..15] | |
569 | ||
570 | You can see that the packet coming in on VLAN 20 with source MAC | |
571 | 50:00:00:00:00:01 became a flow that matches VLAN 20 (written in | |
572 | hexadecimal) and destination MAC 50:00:00:00:00:01. The flow loads | |
573 | port number 1, the input port for the flow we tested, into register 0. | |
574 | ||
575 | ||
576 | == EXAMPLE 2 == | |
577 | ||
578 | Here's a second test command: | |
579 | ||
580 | ovs-appctl ofproto/trace br0 in_port=2,dl_src=50:00:00:00:00:01 -generate | |
581 | ||
582 | The flow that this command tests has the same source MAC and VLAN as | |
583 | example 1, although the VLAN comes from an access port VLAN rather | |
584 | than an 802.1Q header. If we again dump the flows for table 10 with: | |
585 | ||
586 | ovs-ofctl dump-flows br0 table=10 | |
587 | ||
588 | then we see that the flow we saw previously has changed to indicate | |
589 | that the learned port is port 2, as we would expect: | |
590 | ||
591 | NXST_FLOW reply (xid=0x4): | |
592 | table=10, vlan_tci=0x0014/0x0fff,dl_dst=50:00:00:00:00:01 actions=load:0x2->NXM_NX_REG0[0..15] | |
593 | ||
594 | ||
595 | Implementing Table 3: Look Up Destination Port | |
596 | ============================================== | |
597 | ||
598 | This table figures out what port we should send the packet to based on | |
599 | the destination MAC and VLAN. That is, if we've learned the location | |
600 | of the destination (from table 2 processing some previous packet with | |
601 | that destination as its source), then we want to send the packet | |
602 | there. | |
603 | ||
604 | We need only one flow to do the lookup: | |
605 | ||
606 | ovs-ofctl add-flow br0 \ | |
607 | "table=3 priority=50 actions=resubmit(,10), resubmit(,4)" | |
608 | ||
609 | The flow's first action resubmits to table 10, the table that the | |
610 | "learn" action modifies. As you saw previously, the learned flows in | |
611 | this table write the learned port into register 0. If the destination | |
612 | for our packet hasn't been learned, then there will be no matching | |
613 | flow, and so the "resubmit" turns into a no-op. Because registers are | |
614 | initialized to 0, we can use a register 0 value of 0 in our next | |
615 | pipeline stage as a signal to flood the packet. | |
616 | ||
617 | The second action resubmits to table 4, continuing to the next | |
618 | pipeline stage. | |
619 | ||
620 | We can add another flow to skip the learning table lookup for | |
621 | multicast and broadcast packets, since those should always be flooded: | |
622 | ||
623 | ovs-ofctl add-flow br0 \ | |
624 | "table=3 priority=99 dl_dst=01:00:00:00:00:00/01:00:00:00:00:00 \ | |
625 | actions=resubmit(,4)" | |
626 | ||
627 | >>> We don't strictly need to add this flow, because multicast | |
628 | addresses will never show up in our learning table. (In turn, | |
629 | that's because we put a flow into table 0 to drop packets that | |
630 | have a multicast source address.) | |
631 | ||
632 | ||
633 | Testing Table 3 | |
634 | --------------- | |
635 | ||
636 | == EXAMPLE == | |
637 | ||
638 | Here's a command that should cause OVS to learn that f0:00:00:00:00:01 | |
639 | is on p1 in VLAN 20: | |
640 | ||
641 | ovs-appctl ofproto/trace br0 in_port=1,dl_vlan=20,dl_src=f0:00:00:00:00:01,dl_dst=90:00:00:00:00:01 -generate | |
642 | ||
643 | Here's an excerpt from the output that shows (from the "no match" | |
644 | looking up the resubmit to table 10) that the flow's destination was | |
645 | unknown: | |
646 | ||
647 | Resubmitted flow: unchanged | |
648 | Resubmitted regs: reg0=0x0 reg1=0x0 reg2=0x0 reg3=0x0 reg4=0x0 reg5=0x0 reg6=0x0 reg7=0x0 | |
649 | Resubmitted odp: drop | |
650 | Rule: table=3 cookie=0 priority=50 | |
651 | OpenFlow actions=resubmit(,10),resubmit(,4) | |
652 | ||
653 | Resubmitted flow: unchanged | |
654 | Resubmitted regs: reg0=0x0 reg1=0x0 reg2=0x0 reg3=0x0 reg4=0x0 reg5=0x0 reg6=0x0 reg7=0x0 | |
655 | Resubmitted odp: drop | |
656 | No match | |
657 | ||
658 | You can verify that the packet's source was learned two ways. The | |
659 | most direct way is to dump the learning table with: | |
660 | ||
661 | ovs-ofctl dump-flows br0 table=10 | |
662 | ||
663 | which ought to show roughly the following, with extraneous details | |
664 | removed: | |
665 | ||
666 | table=10, vlan_tci=0x0014/0x0fff,dl_dst=f0:00:00:00:00:01 actions=load:0x1->NXM_NX_REG0[0..15] | |
667 | ||
668 | >>> If you tried the examples for the previous step, or if you did | |
669 | some of your own experiments, then you might see additional flows | |
670 | there. These additional flows are harmless. If they bother you, | |
671 | then you can remove them with "ovs-ofctl del-flows br0 table=10". | |
672 | ||
673 | The other way is to inject a packet to take advantage of the learning | |
674 | entry. For example, we can inject a packet on p2 whose destination is | |
675 | the MAC address that we just learned on p1: | |
676 | ||
677 | ovs-appctl ofproto/trace br0 in_port=2,dl_src=90:00:00:00:00:01,dl_dst=f0:00:00:00:00:01 -generate | |
678 | ||
679 | Here's an interesting excerpt from that command's output. This group | |
680 | of lines traces the "resubmit(,10)", showing that the packet matched | |
681 | the learned flow for the first MAC we used, loading the OpenFlow port | |
682 | number for the learned port p1 into register 0: | |
683 | ||
684 | Resubmitted flow: unchanged | |
685 | Resubmitted regs: reg0=0x0 reg1=0x0 reg2=0x0 reg3=0x0 reg4=0x0 reg5=0x0 reg6=0x0 reg7=0x0 | |
686 | Resubmitted odp: drop | |
687 | Rule: table=10 cookie=0 vlan_tci=0x0014/0x0fff,dl_dst=f0:00:00:00:00:01 | |
688 | OpenFlow actions=load:0x1->NXM_NX_REG0[0..15] | |
689 | ||
690 | ||
691 | If you read the commands above carefully, then you might have noticed | |
692 | that they simply have the Ethernet source and destination addresses | |
693 | exchanged. That means that if we now rerun the first ovs-appctl | |
694 | command above, e.g.: | |
695 | ||
696 | ovs-appctl ofproto/trace br0 in_port=1,dl_vlan=20,dl_src=f0:00:00:00:00:01,dl_dst=90:00:00:00:00:01 -generate | |
697 | ||
698 | then we see in the output that the destination has now been learned: | |
699 | ||
700 | Resubmitted flow: unchanged | |
701 | Resubmitted regs: reg0=0x0 reg1=0x0 reg2=0x0 reg3=0x0 reg4=0x0 reg5=0x0 reg6=0x0 reg7=0x0 | |
702 | Resubmitted odp: drop | |
703 | Rule: table=10 cookie=0 vlan_tci=0x0014/0x0fff,dl_dst=90:00:00:00:00:01 | |
704 | OpenFlow actions=load:0x2->NXM_NX_REG0[0..15] | |
705 | ||
706 | ||
707 | Implementing Table 4: Output Processing | |
708 | ======================================= | |
709 | ||
710 | At entry to stage 4, we know that register 0 contains either the | |
711 | desired output port or is zero if the packet should be flooded. We | |
712 | also know that the packet's VLAN is in its 802.1Q header, even if the | |
713 | VLAN was implicit because the packet came in on an access port. | |
714 | ||
715 | The job of the final pipeline stage is to actually output packets. | |
716 | The job is trivial for output to our trunk port p1: | |
717 | ||
718 | ovs-ofctl add-flow br0 "table=4 reg0=1 actions=1" | |
719 | ||
720 | For output to the access ports, we just have to strip the VLAN header | |
721 | before outputting the packet: | |
722 | ||
723 | ovs-ofctl add-flows br0 - <<'EOF' | |
724 | table=4 reg0=2 actions=strip_vlan,2 | |
725 | table=4 reg0=3 actions=strip_vlan,3 | |
726 | table=4 reg0=4 actions=strip_vlan,4 | |
727 | EOF | |
728 | ||
729 | The only slightly tricky part is flooding multicast and broadcast | |
730 | packets and unicast packets with unlearned destinations. For those, | |
731 | we need to make sure that we only output the packets to the ports that | |
732 | carry our packet's VLAN, and that we include the 802.1Q header in the | |
733 | copy output to the trunk port but not in copies output to access | |
734 | ports: | |
735 | ||
736 | ovs-ofctl add-flows br0 - <<'EOF' | |
737 | table=4 reg0=0 priority=99 dl_vlan=20 actions=1,strip_vlan,2 | |
738 | table=4 reg0=0 priority=99 dl_vlan=30 actions=1,strip_vlan,3,4 | |
739 | table=4 reg0=0 priority=50 actions=1 | |
740 | EOF | |
741 | ||
742 | >>> Our rules rely on the standard OpenFlow behavior that an output | |
743 | action will not forward a packet back out the port it came in on. | |
744 | That is, if a packet comes in on p1, and we've learned that the | |
745 | packet's destination MAC is also on p1, so that we end up with | |
746 | "actions=1" as our actions, the switch will not forward the packet | |
747 | back out its input port. The multicast/broadcast/unknown | |
748 | destination cases above also rely on this behavior. | |
749 | ||
750 | ||
751 | Testing Table 4 | |
752 | --------------- | |
753 | ||
754 | == EXAMPLE 1: Broadcast, Multicast, and Unknown Destination == | |
755 | ||
756 | Try tracing a broadcast packet arriving on p1 in VLAN 30: | |
757 | ||
758 | ovs-appctl ofproto/trace br0 in_port=1,dl_dst=ff:ff:ff:ff:ff:ff,dl_vlan=30 | |
759 | ||
760 | The interesting part of the output is the final line, which shows that | |
761 | the switch would remove the 802.1Q header and then output the packet to | |
762 | p3 and p4, which are access ports for VLAN 30: | |
763 | ||
764 | Datapath actions: pop_vlan,3,4 | |
765 | ||
766 | Similarly, if we trace a broadcast packet arriving on p3: | |
767 | ||
768 | ovs-appctl ofproto/trace br0 in_port=3,dl_dst=ff:ff:ff:ff:ff:ff | |
769 | ||
770 | then we see that it is output to p1 with an 802.1Q tag and then to p4 | |
771 | without one: | |
772 | ||
773 | Datapath actions: push_vlan(vid=30,pcp=0),1,pop_vlan,4 | |
774 | ||
775 | >>> Open vSwitch could simplify the datapath actions here to just | |
776 | "4,push_vlan(vid=30,pcp=0),1" but it is not smart enough to do so. | |
777 | ||
778 | The following are also broadcasts, but the result is to drop the | |
779 | packets because the VLAN only belongs to the input port: | |
780 | ||
781 | ovs-appctl ofproto/trace br0 in_port=1,dl_dst=ff:ff:ff:ff:ff:ff | |
782 | ovs-appctl ofproto/trace br0 in_port=1,dl_dst=ff:ff:ff:ff:ff:ff,dl_vlan=55 | |
783 | ||
784 | Try some other broadcast cases on your own: | |
785 | ||
786 | ovs-appctl ofproto/trace br0 in_port=1,dl_dst=ff:ff:ff:ff:ff:ff,dl_vlan=20 | |
787 | ovs-appctl ofproto/trace br0 in_port=2,dl_dst=ff:ff:ff:ff:ff:ff | |
788 | ovs-appctl ofproto/trace br0 in_port=4,dl_dst=ff:ff:ff:ff:ff:ff | |
789 | ||
790 | You can see the same behavior with multicast packets and with unicast | |
791 | packets whose destination has not been learned, e.g.: | |
792 | ||
793 | ovs-appctl ofproto/trace br0 in_port=4,dl_dst=01:00:00:00:00:00 | |
794 | ovs-appctl ofproto/trace br0 in_port=1,dl_dst=90:12:34:56:78:90,dl_vlan=20 | |
795 | ovs-appctl ofproto/trace br0 in_port=1,dl_dst=90:12:34:56:78:90,dl_vlan=30 | |
796 | ||
797 | ||
798 | == EXAMPLE 2: MAC Learning == | |
799 | ||
800 | Let's follow the same pattern as we did for table 3. First learn a | |
801 | MAC on port p1 in VLAN 30: | |
802 | ||
803 | ovs-appctl ofproto/trace br0 in_port=1,dl_vlan=30,dl_src=10:00:00:00:00:01,dl_dst=20:00:00:00:00:01 -generate | |
804 | ||
805 | You can see from the last line of output that the packet's destination | |
806 | is unknown, so it gets flooded to both p3 and p4, the other ports in | |
807 | VLAN 30: | |
808 | ||
809 | Datapath actions: pop_vlan,3,4 | |
810 | ||
811 | Then reverse the MACs and learn the first flow's destination on port | |
812 | p4: | |
813 | ||
814 | ovs-appctl ofproto/trace br0 in_port=4,dl_src=20:00:00:00:00:01,dl_dst=10:00:00:00:00:01 -generate | |
815 | ||
816 | The last line of output shows that the this packet's destination is | |
817 | known to be p1, as learned from our previous command: | |
818 | ||
819 | Datapath actions: push_vlan(vid=30,pcp=0),1 | |
820 | ||
821 | Now, if we rerun our first command: | |
822 | ||
823 | ovs-appctl ofproto/trace br0 in_port=1,dl_vlan=30,dl_src=10:00:00:00:00:01,dl_dst=20:00:00:00:00:01 -generate | |
824 | ||
825 | we can see that the result is no longer a flood but to the specified | |
826 | learned destination port p4: | |
827 | ||
828 | Datapath actions: pop_vlan,4 | |
829 | ||
830 | ||
831 | Contact | |
832 | ======= | |
833 | ||
834 | bugs@openvswitch.org | |
835 | http://openvswitch.org/ |