4 * Experimental support for OpenFlow 1.1 (in addition to 1.2 and
5 1.3, which had experimental support in 1.10).
6 * New support for matching outer source and destination IP address
7 of tunneled packets, for tunnel ports configured with the newly
8 added "remote_ip=flow" and "local_ip=flow" options.
9 * Support for matching on metadata 'pkt_mark' for interacting with
10 other system components. On Linux this corresponds to the skb
12 * Support matching, rewriting SCTP ports
13 - The Interface table in the database has a new "ifindex" column to
14 report the interface's OS-assigned ifindex.
15 - New "check-oftest" Makefile target for running OFTest against Open
16 vSwitch. See README-OFTest for details.
17 - The flow eviction threshold has been moved to the Open_vSwitch table.
18 - Database names are now mandatory when specifying ovsdb-server options
19 through database paths (e.g. Private key option with the database name
20 should look like "--private-key=db:Open_vSwitch,SSL,private_key").
21 - Added ovs-dev.py, a utility script helpful for Open vSwitch developers.
22 - Support for Linux kernels up to 3.10
24 * New "ofp-parse" for printing OpenFlow messages read from a file.
25 - Added configurable flow caching support to IPFIX exporter.
26 - Dropped support for Linux pre-2.6.32.
27 - Log file timestamps and ovsdb commit timestamps are now reported
28 with millisecond resolution. (Previous versions only reported
34 - Support for megaflows, which allows wildcarding in the kernel (and
35 any dpif implementation that supports wildcards). Depending on
36 the flow table and switch configuration, flow set up rates are
37 close to the Linux bridge.
38 - The "tutorial" directory contains a new tutorial for some advanced
39 Open vSwitch features.
40 - Stable bond mode has been removed.
41 - The autopath action has been removed.
42 - New support for the data encapsulation format of the LISP tunnel
43 protocol (RFC 6830). An external control plane or manual flow
44 setup is required for EID-to-RLOC mapping.
46 * The "dec_mpls_ttl" and "set_mpls_ttl" actions from OpenFlow
47 1.1 and later are now implemented.
48 * New "stack" extension for use in actions, to push and pop from
50 * The "load" and "set_field" actions can now modify the "in_port". (This
51 allows one to enable output to a flow's input port by setting the
52 in_port to some unused value, such as OFPP_NONE.)
54 * New debugging commands "add-flow", "mod-flow", "del-flow".
55 * "dump-flows" now has a -m option to increase output verbosity.
56 - In dpif-based bridges, cache action translations, which can improve
57 flow set up performance by 80% with a complicated flow table.
58 - New syslog format, prefixed with "ovs|", to be easier to filter.
59 - RHEL: Removes the default firewall rule that allowed GRE traffic to
60 pass through. Any users that relied on this automatic firewall hole
61 will have to manually configure it. The ovs-ctl(8) manpage documents
62 the "enable-protocol" command that can be used as an alternative.
63 - New CFM demand mode which uses data traffic to indicate interface
68 - Bridge compatibility support has been removed. Any uses that
69 rely on ovs-brcompatd will have to stick with Open vSwitch 1.9.x
70 or adapt to native Open vSwitch support (e.g. use ovs-vsctl instead
72 - The maximum size of the MAC learning table is now configurable.
73 - With the Linux datapath, packets for new flows are now queued
74 separately on a per-port basis, so it should no longer be
75 possible for a large number of new flows arriving on one port to
76 prevent new flows from being processed on other ports.
78 * Previously ovs-vsctl would retry connecting to the database forever,
79 causing it to hang if ovsdb-server was not running. Now, ovs-vsctl
80 only tries once by default (use --retry to try forever). This change
81 means that you may want to remove uses of --timeout to avoid hangs
83 * Many "ovs-vsctl" database commands now accept an --if-exists option.
84 Please refer to the ovs-vsctl manpage for details.
86 - Experimental support for newer versions of OpenFlow. See
87 the "What versions of OpenFlow does Open vSwitch support?"
88 question in the FAQ for more details.
89 - The OpenFlow "dp_desc" may now be configured by setting the
90 value of other-config:dp-desc in the Bridge table.
91 - It is possible to request the OpenFlow port number with the
92 "ofport_request" column in the Interface table.
93 - The NXM flow_removed message now reports the OpenFlow table ID
94 from which the flow was removed.
96 - New support for the VXLAN tunnel protocol (see the IETF draft here:
97 http://tools.ietf.org/html/draft-mahalingam-dutt-dcops-vxlan-03).
98 - Tunneling requires the version of the kernel module paired with
99 Open vSwitch 1.9.0 or later.
100 - Inheritance of the Don't Fragment bit in IP tunnels (df_inherit)
101 is no longer supported.
102 - Path MTU discovery is no longer supported.
103 - CAPWAP tunneling support removed.
104 - Tunnels with multicast destination ports are no longer supported.
106 - The "dump-flows" and "del-flows" no longer require an argument
107 if only one datapath exists.
109 - New "vlog/disable-rate-limit" and "vlog/enable-rate-limit"
110 commands available allow control over logging rate limits.
111 - New "dpif/dump-dps", "dpif/show", and "dpif/dump-flows" command
112 that mimic the equivalent ovs-dpctl commands.
113 - The ofproto library is now responsible for assigning OpenFlow port
114 numbers. An ofproto implementation should assign them when
115 port_construct() is called.
116 - All dpif-based bridges of a particular type share a common
117 datapath called "ovs-<type>", e.g. "ovs-system". The ovs-dpctl
118 commands will now return information on that shared datapath. To
119 get the equivalent bridge-specific information, use the new
120 "ovs-appctl dpif/*" commands.
121 - Backward-incompatible changes:
122 - Earlier Open vSwitch versions treated ANY as a wildcard in flow
123 syntax. OpenFlow 1.1 adds a port named ANY, which introduces a
124 conflict. ANY was rarely used in flow syntax, so we chose to
125 retire that meaning of ANY in favor of the OpenFlow 1.1 meaning.
126 - Patch ports no longer require kernel support, so they now work
127 with FreeBSD and the kernel module built into Linux 3.3 and later.
128 - New "sample" action.
132 ------------------------
134 - Support for ipv6 set action.
135 - SKB mark matching and setting.
136 - support for Linux kernels up to 3.8
137 - FreeBSD is now a supported platform, thanks to code contributions from
138 Gaetano Catalli, Ed Maste, and Giuseppe Lettieri.
139 - ovs-bugtool: New --ovs option to report only OVS related information.
140 - New %t and %T log escapes to identify the subprogram within a
141 cooperating group of processes or threads that emitted a log message.
142 The default log patterns now include this information.
144 - Allow bitwise masking for SHA and THA fields in ARP, SLL and TLL
145 fields in IPv6 neighbor discovery messages, and IPv6 flow label.
146 - Adds support for writing to the metadata field for a flow.
148 - The tunneling code no longer assumes input and output keys are
149 symmetric. If they are not, PMTUD needs to be disabled for
150 tunneling to work. Note this only applies to flow-based keys.
151 - New support for a nonstandard form of GRE that supports a 64-bit key.
152 - Tunnel Path MTU Discovery default value was set to 'disabled'.
153 This feature is deprecated and will be removed soon.
154 - Tunnel header caching removed.
156 - Commands and actions that accept port numbers now also accept keywords
157 that represent those ports (such as LOCAL, NONE, and ALL). This is
158 also the recommended way to specify these ports, for compatibility
159 with OpenFlow 1.1 and later (which use the OpenFlow 1.0 numbers
160 for these ports for different purposes).
162 - Support requesting the port number with the "port_no" option in
163 the "add-if" command.
164 - ovs-pki: The "online PKI" features have been removed, along with
165 the ovs-pki-cgi program that facilitated it, because of some
166 alarmist insecurity claims. We do not believe that these claims
167 are true, but because we do not know of any users for this
168 feature it seems better on balance to remove it. (The ovs-pki-cgi
169 program was not included in distribution packaging.)
170 - ovsdb-server now enforces the immutability of immutable columns. This
171 was not enforced in earlier versions due to an oversight.
172 - The following features are now deprecated. They will be removed no
173 earlier than February 2013. Please email dev@openvswitch.org with
175 - Bridge compatibility.
177 - The autopath action.
178 - Interface type "null".
179 - Numeric values for reserved ports (see "ovs-ofctl" note above).
180 - Tunnel Path MTU Discovery.
181 - CAPWAP tunnel support.
182 - The data in the RARP packets can now be matched in the same way as the
187 ------------------------
188 *** Internal only release ***
189 - New FAQ. Please send updates and additions!
190 - Authors of controllers, please read the new section titled "Action
191 Reproduction" in DESIGN, which describes an Open vSwitch change in
192 behavior in corner cases that may affect some controllers.
194 - A new test utility that can create L3 tunnel between two Open
195 vSwitches and detect connectivity issues.
197 - New --sort and --rsort options for "dump-flows" command.
198 - "mod-port" command can now control all OpenFlow config flags.
200 - Allow general bitwise masking for IPv4 and IPv6 addresses in
201 IPv4, IPv6, and ARP packets. (Previously, only CIDR masks
203 - Allow support for arbitrary Ethernet masks. (Previously, only
204 the multicast bit in the destination address could be individually
206 - New field OXM_OF_METADATA, to align with OpenFlow 1.1.
207 - The OFPST_QUEUE request now reports an error if a specified port or
208 queue does not exist, or for requests for a specific queue on all
209 ports, if the specified queue does not exist on any port. (Previous
210 versions generally reported an empty set of results.)
211 - New "flow monitor" feature to allow controllers to be notified of
212 flow table changes as they happen.
213 - Additional protocols are not mirrored and dropped when forward-bpdu is
214 false. For a full list, see the ovs-vswitchd.conf.db man page.
215 - Open vSwitch now sends RARP packets in situations where it previously
216 sent a custom protocol, making it consistent with behavior of QEMU and
218 - All Open vSwitch programs and log files now show timestamps in UTC,
219 instead the local timezone, by default.
223 ------------------------
224 - kernel modules are renamed. openvswitch_mod.ko is now
225 openvswitch.ko and brcompat_mod.ko is now brcompat.ko.
226 - Increased the number of NXM registers to 8.
227 - Added ability to configure DSCP setting for manager and controller
228 connections. By default, these connections have a DSCP value of
229 Internetwork Control (0xc0).
230 - Added the granular link health statistics, 'cfm_health', to an
233 - Added support to mask nd_target for ICMPv6 neighbor discovery flows.
234 - Added support for OpenFlow 1.3 port description (OFPMP_PORT_DESC)
237 - Added the "dump-ports-desc" command to retrieve port
238 information using the new port description multipart messages.
240 - Added support for spawning ovs-test server from the client.
241 - Now ovs-test is able to automatically create test bridges and ports.
242 - "ovs-dpctl dump-flows" now prints observed TCP flags in TCP flows.
243 - Tripled flow setup performance.
244 - The "coverage/log" command previously available through ovs-appctl
245 has been replaced by "coverage/show". The new command replies with
246 coverage counter values, instead of logging them.
250 ------------------------
251 - Allow OFPP_CONTROLLER as the in_port for packet-out messages.
255 ------------------------
256 *** Internal only release ***
258 - LACP bonds no longer fall back to balance-slb when negotiations fail.
259 Instead they drop traffic.
260 - The default bond_mode changed from SLB to active-backup, to protect
261 unsuspecting users from the significant risks of SLB bonds (which are
262 documented in vswitchd/INTERNALS).
263 - Load balancing can be disabled by setting the bond-rebalance-interval
266 - Added support for bitwise matching on TCP and UDP ports.
267 See ovs-ofctl(8) for more information.
268 - NXM flow dumps now include times elapsed toward idle and hard
270 - Added an OpenFlow extension NXT_SET_ASYNC_CONFIG that allows
271 controllers more precise control over which OpenFlow messages they
272 receive asynchronously.
273 - New "fin_timeout" action.
274 - Added "fin_timeout" support to "learn" action.
275 - New Nicira action NXAST_CONTROLLER that offers additional features
276 over output to OFPP_CONTROLLER.
277 - When QoS settings for an interface do not configure a default queue
278 (queue 0), Open vSwitch now uses a default configuration for that
279 queue, instead of dropping all packets as in previous versions.
281 - Logging to console and file will have UTC timestamp as a default for
282 all the daemons. An example of the default format is
283 2012-01-27T16:35:17Z. ovs-appctl can be used to change the default
285 - The syntax of commands and options to set log levels was simplified,
286 to make it easier to remember.
287 - New support for limiting the number of flows in an OpenFlow flow
288 table, with configurable policy for evicting flows upon
289 overflow. See the Flow_Table table in ovs-vswitch.conf.db(5)
290 for more information.
291 - New "enable-async-messages" column in the Controller table. If set to
292 false, OpenFlow connections to the controller will initially have all
293 asynchronous messages disabled, overriding normal OpenFlow behavior.
294 - ofproto-provider interface:
295 - "struct rule" has a new member "used" that ofproto implementations
296 should maintain by updating with ofproto_rule_update_used().
298 - The new option --timestamp causes the "monitor" command to print
299 a timestamp with every update.
300 - CFM module CCM broadcasts can now be tagged with an 802.1p priority.
304 ------------------------
306 - Added support for querying, modifying, and deleting flows
307 based on flow cookie when using NXM.
308 - Added new NXM_PACKET_IN format.
309 - Added new NXAST_DEC_TTL action.
311 - Added daemonization support to the monitor and snoop commands.
313 - The "find" command supports new set relational operators
314 {=}, {!=}, {<}, {>}, {<=}, and {>=}.
315 - ovsdb-tool now uses the typical database and schema installation
316 directories as defaults.
317 - The default MAC learning timeout has been increased from 60 seconds
318 to 300 seconds. The MAC learning timeout is now configurable.
322 ------------------------
323 - Compatible with Open vSwitch kernel module included in Linux 3.3.
324 - New "VLAN splinters" feature to work around buggy device drivers
325 in old Linux versions. (This feature is deprecated. When
326 broken device drivers are no longer in widespread use, we will
327 delete this feature.) See ovs-vswitchd.conf.db(5) for more
330 - Added ability to match on IPv6 flow label through NXM.
331 - Added ability to match on ECN bits in IPv4 and IPv6 through NXM.
332 - Added ability to match on TTL in IPv4 and IPv6 through NXM.
333 - Added ability to modify ECN bits in IPv4.
334 - Added ability to modify TTL in IPv4.
336 - Don't require the "normal" action to use mirrors. Traffic will
337 now be properly mirrored for any flows, regardless of their
339 - Track packet and byte statistics sent on mirrors.
340 - The sFlow implementation can now usually infer the correct agent
341 device instead of having to be told explicitly.
343 - New "fdb/flush" command to flush bridge's MAC learning table.
345 - A new distributed testing tool that allows one to diagnose performance
346 and connectivity issues. This tool currently is not included in RH or
348 - RHEL packaging now supports integration with Red Hat network scripts.
350 - Post 1.4.*, OVS will be changing the default bond mode from balance-slb
351 to active-backup. SLB bonds carry significant risks with them
352 (documented vswitchd/INTERNALS) which we want to prevent unsuspecting
353 users from running into. Users are advised to update any scripts or
354 configuration which may be negatively impacted by explicitly setting
355 the bond mode which they want to use.
359 ------------------------
361 - Added an OpenFlow extension which allows the "output" action to accept
363 - Added an OpenFlow extension for flexible learning.
364 - Bumped number of NXM registers from four to five.
366 - New "version" command to determine version of running daemon.
367 - If no argument is provided for "cfm/show", displays detailed
368 information about all interfaces with CFM enabled.
369 - If no argument is provided for "lacp/show", displays detailed
370 information about all ports with LACP enabled.
372 - New "set-if" command to modify a datapath port's configuration.
374 - The software switch now supports 255 OpenFlow tables, instead
375 of just one. By default, only table 0 is consulted, but the
376 new NXAST_RESUBMIT_TABLE action can look up in additional
377 tables. Tables 128 and above are reserved for use by the
378 switch itself; please use only tables 0 through 127.
379 - Add support for 802.1D spanning tree (STP).
380 - Fragment handling extensions:
381 - New OFPC_FRAG_NX_MATCH fragment handling mode, in which L4
382 fields are made available for matching in fragments with
384 - New NXM_NX_IP_FRAG match field for matching IP fragments (usable
385 via "ip_frag" in ovs-ofctl).
386 - New ovs-ofctl "get-frags" and "set-frags" commands to get and set
387 fragment handling policy.
388 - CAPWAP tunneling now supports an extension to transport a 64-bit key.
389 By default it remains compatible with the old version and other
390 standards-based implementations.
391 - Flow setups are now processed in a round-robin manner across ports
392 to prevent any single client from monopolizing the CPU and conducting
393 a denial of service attack.
394 - Added support for native VLAN tagging. A new "vlan_mode"
395 parameter can be set for "port". Possible values: "access",
396 "trunk", "native-tagged" and "native-untagged".
397 - test-openflowd has been removed. Please use ovs-vswitchd instead.
400 ------------------------
401 - New "ofproto" abstraction layer to ease porting to hardware
403 - Packaging for Red Hat Enterprise Linux 5.6 and 6.0.
404 - Datapath support for Linux kernels up to 3.0.
406 - New "bundle" and "bundle_load" action extensions.
408 - Implement table unique constraints.
409 - Support cooperative locking between callers.
411 - New "-s" option for "show" command prints packet and byte
412 counters for each port.
414 - New "--readd" option for "replace-flows".
416 - New "show" command to print an overview of configuration.
417 - New "comment" command to add remark that explains intentions.
418 - ovs-brcompatd has been rewritten to fix long-standing bugs.
419 - ovs-openflowd has been renamed test-openflowd and moved into the
420 tests directory. Its presence confused too many users. Please
421 use ovs-vswitchd instead.
422 - New ovs-benchmark utility to test flow setup performance.
423 - A new log level "off" has been added. Configuring a log facility
424 "off" prevents any messages from being logged to it. Previously,
425 "emer" was effectively "off" because no messages were ever logged at
426 level "emer". Now, errors that cause a process to exit are logged
428 - "configure" option --with-l26 has been renamed --with-linux, and
429 --with-l26-source has been renamed --with-linux-source. The old
430 names will be removed after the next release, so please update
432 - The "-2.6" suffix has been dropped from the datapath/linux-2.6 and
433 datapath/linux-2.6/compat-2.6 directories.
435 - Dropped support for "tun_id_from_cookie" OpenFlow extension.
436 Please use the extensible match extensions instead.
437 - Removed the Maintenance_Point and Monitor tables in an effort
438 to simplify 802.1ag configuration.
439 - Performance and scalability improvements
443 ------------------------
444 - Ability to define policies over IPv6
447 - Support for extensible match extensions to OpenFlow
449 - Support for HFSC qdisc.
450 - Queue used by in-band control can now be configured.
452 - Kernel<->userspace interface has been reworked and should be
453 close to a stable ABI now.
454 - "Port group" concept has been dropped.
455 - GRE over IPSEC tunnels
457 - New active backup bonding mode.
458 - New L4 hashing support when LACP is enabled.
459 - Source MAC hash now includes VLAN field also.
461 - Greatly improved handling of large flow tables
463 - "show" command now prints full vport configuration.
464 - "dump-groups" command removed since kernel support for
465 port groups was dropped.
467 - New commands for working with the new Managers table.
468 - "list" command enhanced with new formatting options and --columns
470 - "get" command now accepts new --id option.
471 - New "find" command.
473 - New "queue-stats" command for printing queue stats.
474 - New commands "replace-flows" and "diff-flows".
475 - Commands to add and remove flows can now read from files.
476 - New --flow-format option to enable or disable NXM.
477 - New --more option to increase OpenFlow message verbosity.
478 - Removed "tun-cookie" command, which is no longer useful.
479 - ovs-controller enhancements for testing various features.
480 - New ovs-vlan-test command for testing for Linux kernel driver VLAN
481 bugs. New ovs-vlan-bug-workaround command for enabling and
482 disabling a workaround for these driver bugs.
484 - "Resubmit" actions now update flow statistics.
485 - New "register" extension for use in matching and actions, via NXM.
486 - New "multipath" experimental action extension.
487 - New support for matching multicast Ethernet frames, via NXM.
488 - New extension for OpenFlow vendor error codes.
489 - New extension to set the QoS output queue without actually
490 sending to an output port.
491 - Open vSwitch now reports a single flow table, instead of
492 separate hash and wildcard tables. This better models the
493 current implementation.
494 - New experimental "note" action.
495 - New "ofproto/trace" ovs-appctl command and associated utilities
496 to ease debugging complex flow tables.
498 - Schema documentation now includes an entity-relationship diagram.
499 - The database is now garbage collected. In most tables,
500 unreferenced rows will be deleted automatically.
501 - Many tables now include statistics updated periodically by
502 ovs-vswitchd or ovsdb-server.
503 - Every table now has an "external-ids" column for use by OVS
505 - There is no default controller anymore. Each bridge must have its
506 controller individually specified.
507 - The "fail-mode" is now a property of a Bridge instead of a Controller.
508 - New versioning and checksum features.
509 - New Managers table and manager_options column in Open_vSwitch table
510 for specifying managers. The old "managers" column in the
511 Open_vSwitch table has been removed.
512 - Many "name" columns are now immutable.
514 - Dropped support for XenServer pre-5.6.100.
515 - Dropped support for Linux pre-2.6.18.
516 - Dropped controller discovery support.
517 - Dropped "ovs-ofctl status" and the OpenFlow extension that it used.
518 Statistics reporting in the database is a rough equivalent.
519 - Dropped the "corekeeper" package (now separate, at
520 http://openvswitch.org/cgi-bin/gitweb.cgi?p=corekeeper).
521 - Performance and scalability improvements
524 v1.1.0pre2 - 13 Sep 2010
525 ------------------------
528 v1.1.0pre1 - 31 Aug 2010
529 ------------------------
530 - OpenFlow 1.0 slicing (QoS) functionality
531 - Python bindings for configuration database (no write support)
532 - Performance and scalability improvements
537 - New "patch" interface type
542 - Configuration database with remote management
545 - Support for XenServer 5.5 and 5.6
546 - Performance and scalability improvements
549 v0.99.2 - 18 Feb 2010
550 ---------------------
553 v0.99.1 - 25 Jan 2010
554 ---------------------
555 - Add support for sFlow(R)
556 - Make headers compatible with C++
559 v0.99.0 - 14 Jan 2010
560 ---------------------
561 - User-space forwarding engine
564 v0.90.7 - 29 Nov 2009
565 ---------------------
566 - Add support for NetFlow active timeouts
573 v0.90.5 - 21 Sep 2009
574 ---------------------
575 - Generalize in-band control to more diverse network setups