1OVSDB(7) Open vSwitch OVSDB(7)
2
6 ovsdb - Open vSwitch Database (Overview)
7
9 OVSDB, the Open vSwitch Database, is a network-accessible database sys‐
10 tem. Schemas in OVSDB specify the tables in a database and their col‐
11 umns’ types and can include data, uniqueness, and referential integrity
12 constraints. OVSDB offers atomic, consistent, isolated, durable trans‐
13 actions. RFC 7047 specifies the JSON-RPC based protocol that OVSDB
14 clients and servers use to communicate.
15 The OVSDB protocol is well suited for state synchronization because it
17 allows each client to monitor the contents of a whole database or a
18 subset of it. Whenever a monitored portion of the database changes,
19 the server tells the client what rows were added or modified (including
20 the new contents) or deleted. Thus, OVSDB clients can easily keep
21 track of the newest contents of any part of the database.
22 While OVSDB is general-purpose and not particularly specialized for use
24 with Open vSwitch, Open vSwitch does use it for multiple purposes. The
25 leading use of OVSDB is for configuring and monitoring ovs-vswitchd(8),
26 the Open vSwitch switch daemon, using the schema documented in
27 ovs-vswitchd.conf.db(5). The Open Virtual Network (OVN) project uses
28 two OVSDB schemas, documented as part of that project. Finally, Open
29 vSwitch includes the “VTEP” schema, documented in vtep(5) that many
30 third-party hardware switches support for configuring VXLAN, although
31 OVS itself does not directly use this schema.
32 The OVSDB protocol specification allows independent, interoperable im‐
34 plementations of OVSDB to be developed. Open vSwitch includes an OVSDB
35 server implementation named ovsdb-server(1), which supports several
36 protocol extensions documented in its manpage, and a basic command-line
37 OVSDB client named ovsdb-client(1), as well as OVSDB client libraries
38 for C and for Python. Open vSwitch documentation often speaks of these
39 OVSDB implementations in Open vSwitch as simply “OVSDB,” even though
40 that is distinct from the OVSDB protocol; we make the distinction ex‐
41 plicit only when it might otherwise be unclear from the context.
42 In addition to these generic OVSDB server and client tools, Open
44 vSwitch includes tools for working with databases that have specific
45 schemas: ovs-vsctl works with the ovs-vswitchd configuration database
46 and vtep-ctl works with the VTEP database.
47 RFC 7047 specifies the OVSDB protocol but it does not specify an
49 on-disk storage format. Open vSwitch includes ovsdb-tool(1) for work‐
50 ing with its own on-disk database formats. The most notable feature of
51 this format is that ovsdb-tool(1) makes it easy for users to print the
52 transactions that have changed a database since the last time it was
53 compacted. This feature is often useful for troubleshooting.
54
56 Schemas in OVSDB have a JSON format that is specified in RFC 7047.
57 They are often stored in files with an extension .ovsschema. An
58 on-disk database in OVSDB includes a schema and data, embedding both
59 into a single file. The Open vSwitch utility ovsdb-tool has commands
60 that work with schema files and with the schemas embedded in database
61 files.
62 An Open vSwitch schema has three important identifiers. The first is
64 its name, which is also the name used in JSON-RPC calls to identify a
65 database based on that schema. For example, the schema used to config‐
66 ure Open vSwitch has the name Open_vSwitch. Schema names begin with a
67 letter or an underscore, followed by any number of letters, under‐
68 scores, or digits. The ovsdb-tool commands schema-name and db-name ex‐
69 tract the schema name from a schema or database file, respectively.
70 An OVSDB schema also has a version of the form x.y.z e.g. 1.2.3.
72 Schemas managed within the Open vSwitch project manage version number‐
73 ing in the following way (but OVSDB does not mandate this approach).
74 Whenever we change the database schema in a non-backward compatible way
75 (e.g. when we delete a column or a table), we increment <x> and set <y>
76 and <z> to 0. When we change the database schema in a backward compat‐
77 ible way (e.g. when we add a new column), we increment <y> and set <z>
78 to 0. When we change the database schema cosmetically (e.g. we rein‐
79 dent its syntax), we increment <z>. The ovsdb-tool commands
80 schema-version and db-version extract the schema version from a schema
81 or database file, respectively.
82 Very old OVSDB schemas do not have a version, but RFC 7047 mandates it.
84 An OVSDB schema optionally has a “checksum.” RFC 7047 does not specify
86 the use of the checksum and recommends that clients ignore it. Open
87 vSwitch uses the checksum to remind developers to update the version:
88 at build time, if the schema’s embedded checksum, ignoring the checksum
89 field itself, does not match the schema’s content, then it fails the
90 build with a recommendation to update the version and the checksum.
91 Thus, a developer who changes the schema, but does not update the ver‐
92 sion, receives an automatic reminder. In practice this has been an ef‐
93 fective way to ensure compliance with the version number policy. The
94 ovsdb-tool commands schema-cksum and db-cksum extract the schema check‐
95 sum from a schema or database file, respectively.
96
98 OVSDB supports three service models for databases: standalone, ac‐
99 tive-backup, and clustered. The service models provide different com‐
100 promises among consistency, availability, and partition tolerance.
101 They also differ in the number of servers required and in terms of per‐
102 formance. The standalone and active-backup database service models
103 share one on-disk format, and clustered databases use a different for‐
104 mat, but the OVSDB programs work with both formats. ovsdb(5) documents
105 these file formats.
106 RFC 7047, which specifies the OVSDB protocol, does not mandate or spec‐
108 ify any particular service model.
109 The following sections describe the individual service models.
111 Standalone Database Service Model
113 A standalone database runs a single server. If the server stops run‐
114 ning, the database becomes inaccessible, and if the server’s storage is
115 lost or corrupted, the database’s content is lost. This service model
116 is appropriate when the database controls a process or activity to
117 which it is linked via “fate-sharing.” For example, an OVSDB instance
118 that controls an Open vSwitch virtual switch daemon, ovs-vswitchd, is a
119 standalone database because a server failure would take out both the
120 database and the virtual switch.
121 To set up a standalone database, use ovsdb-tool create to create a
123 database file, then run ovsdb-server to start the database service.
124 To configure a client, such as ovs-vswitchd or ovs-vsctl, to use a
126 standalone database, configure the server to listen on a “connection
127 method” that the client can reach, then point the client to that con‐
128 nection method. See Connection Methods below for information about
129 connection methods.
130 Active-Backup Database Service Model
132 An active-backup database runs two servers (on different hosts). At
133 any given time, one of the servers is designated with the active role
134 and the other the backup role. An active server behaves just like a
135 standalone server. A backup server makes an OVSDB connection to the
136 active server and uses it to continuously replicate its content as it
137 changes in real time. OVSDB clients can connect to either server but
138 only the active server allows data modification or lock transactions.
139 Setup for an active-backup database starts from a working standalone
141 database service, which is initially the active server. On another
142 node, to set up a backup server, create a database file with the same
143 schema as the active server. The initial contents of the database file
144 do not matter, as long as the schema is correct, so ovsdb-tool create
145 will work, as will copying the database file from the active server.
146 Then use ovsdb-server --sync-from=<active> to start the backup server,
147 where <active> is an OVSDB connection method (see Connection Methods
148 below) that connects to the active server. At that point, the backup
149 server will fetch a copy of the active database and keep it up-to-date
150 until it is killed.
151 When the active server in an active-backup server pair fails, an admin‐
153 istrator can switch the backup server to an active role with the
154 ovs-appctl command ovsdb-server/disconnect-active-ovsdb-server.
155 Clients then have read/write access to the now-active server. Of
156 course, administrators are slow to respond compared to software, so in
157 practice external management software detects the active server’s fail‐
158 ure and changes the backup server’s role. For example, the “Integra‐
159 tion Guide for Centralized Control” in the OVN documentation describes
160 how to use Pacemaker for this purpose in OVN.
161 Suppose an active server fails and its backup is promoted to active.
163 If the failed server is revived, it must be started as a backup server.
164 Otherwise, if both servers are active, then they may start out of sync,
165 if the database changed while the server was down, and they will con‐
166 tinue to diverge over time. This also happens if the software managing
167 the database servers cannot reach the active server and therefore
168 switches the backup to active, but other hosts can reach both servers.
169 These “split-brain” problems are unsolvable in general for server
170 pairs.
171 Compared to a standalone server, the active-backup service model some‐
173 what increases availability, at a risk of split-brain. It adds gener‐
174 ally insignificant performance overhead. On the other hand, the clus‐
175 tered service model, discussed below, requires at least 3 servers and
176 has greater performance overhead, but it avoids the need for external
177 management software and eliminates the possibility of split-brain.
178 Open vSwitch 2.6 introduced support for the active-backup service
180 model.
181 IMPORTANT:
183 There was a change of a database file format in version 2.15. To
184 upgrade/downgrade the ovsdb-server processes across this version
185 follow the instructions described under Upgrading from version 2.14
186 and earlier to 2.15 and later and Downgrading from version 2.15 and
187 later to 2.14 and earlier.
188 Clustered Database Service Model
190 A clustered database runs across 3 or 5 or more database servers (the
191 cluster) on different hosts. Servers in a cluster automatically syn‐
192 chronize writes within the cluster. A 3-server cluster can remain
193 available in the face of at most 1 server failure; a 5-server cluster
194 tolerates up to 2 failures. Clusters larger than 5 servers will also
195 work, with every 2 added servers allowing the cluster to tolerate 1
196 more failure, but write performance decreases. The number of servers
197 should be odd: a 4- or 6-server cluster cannot tolerate more failures
198 than a 3- or 5-server cluster, respectively.
199 To set up a clustered database, first initialize it on a single node by
201 running ovsdb-tool create-cluster, then start ovsdb-server. Depending
202 on its arguments, the create-cluster command can create an empty data‐
203 base or copy a standalone database’s contents into the new database.
204 To configure a client to use a clustered database, first configure all
206 of the servers to listen on a connection method that the client can
207 reach, then point the client to all of the servers’ connection methods,
208 comma-separated. See Connection Methods, below, for more detail.
209 Open vSwitch 2.9 introduced support for the clustered service model.
211 How to Maintain a Clustered Database
213 To add a server to a cluster, run ovsdb-tool join-cluster on the new
214 server and start ovsdb-server. To remove a running server from a clus‐
215 ter, use ovs-appctl to invoke the cluster/leave command. When a server
216 fails and cannot be recovered, e.g. because its hard disk crashed, or
217 to otherwise remove a server that is down from a cluster, use ovs-ap‐
218 pctl to invoke cluster/kick to make the remaining servers kick it out
219 of the cluster.
220 The above methods for adding and removing servers only work for healthy
222 clusters, that is, for clusters with no more failures than their maxi‐
223 mum tolerance. For example, in a 3-server cluster, the failure of 2
224 servers prevents servers joining or leaving the cluster (as well as
225 database access). To prevent data loss or inconsistency, the preferred
226 solution to this problem is to bring up enough of the failed servers to
227 make the cluster healthy again, then if necessary remove any remaining
228 failed servers and add new ones. If this cannot be done, though, use
229 ovs-appctl to invoke cluster/leave --force on a running server. This
230 command forces the server to which it is directed to leave its cluster
231 and form a new single-node cluster that contains only itself. The data
232 in the new cluster may be inconsistent with the former cluster: trans‐
233 actions not yet replicated to the server will be lost, and transactions
234 not yet applied to the cluster may be committed. Afterward, any
235 servers in its former cluster will regard the server to have failed.
236 Once a server leaves a cluster, it may never rejoin it. Instead, cre‐
238 ate a new server and join it to the cluster.
239 The servers in a cluster synchronize data over a cluster management
241 protocol that is specific to Open vSwitch; it is not the same as the
242 OVSDB protocol specified in RFC 7047. For this purpose, a server in a
243 cluster is tied to a particular IP address and TCP port, which is spec‐
244 ified in the ovsdb-tool command that creates or joins the cluster. The
245 TCP port used for clustering must be different from that used for OVSDB
246 clients. To change the port or address of a server in a cluster, first
247 remove it from the cluster, then add it back with the new address.
248 To upgrade the ovsdb-server processes in a cluster from one version of
250 Open vSwitch to another, upgrading them one at a time will keep the
251 cluster healthy during the upgrade process. (This is different from
252 upgrading a database schema, which is covered later under Upgrading or
253 Downgrading a Database.)
254 IMPORTANT:
256 There was a change of a database file format in version 2.15. To
257 upgrade/downgrade the ovsdb-server processes across this version
258 follow the instructions described under Upgrading from version 2.14
259 and earlier to 2.15 and later and Downgrading from version 2.15 and
260 later to 2.14 and earlier.
261 Clustered OVSDB does not support the OVSDB “ephemeral columns” feature.
263 ovsdb-tool and ovsdb-client change ephemeral columns into persistent
264 ones when they work with schemas for clustered databases. Future ver‐
265 sions of OVSDB might add support for this feature.
266 Upgrading from version 2.14 and earlier to 2.15 and later
268 There is a change of a database file format in version 2.15 that
269 doesn’t allow older versions of ovsdb-server to read the database file
270 modified by the ovsdb-server version 2.15 or later. This also affects
271 runtime communications between servers in active-backup and cluster
272 service models. To upgrade the ovsdb-server processes from one version
273 of Open vSwitch (2.14 or earlier) to another (2.15 or higher) instruc‐
274 tions below should be followed. (This is different from upgrading a
275 database schema, which is covered later under Upgrading or Downgrading
276 a Database.)
277 In case of standalone service model no special handling during upgrade
279 is required.
280 For the active-backup service model, administrator needs to update
282 backup ovsdb-server first and the active one after that, or shut down
283 both servers and upgrade at the same time.
284 For the cluster service model recommended upgrade strategy is follow‐
286 ing:
287 1. Upgrade processes one at a time. Each ovsdb-server process after
289 upgrade should be started with --disable-file-column-diff command
290 line argument.
291 2. When all ovsdb-server processes upgraded, use ovs-appctl to invoke
293 ovsdb/file/column-diff-enable command on each of them or restart all
294 ovsdb-server processes one at a time without --disable-file-col‐
295 umn-diff command line option.
296 Downgrading from version 2.15 and later to 2.14 and earlier
298 Similar to upgrading covered under Upgrading from version 2.14 and ear‐
299 lier to 2.15 and later, downgrading from the ovsdb-server version 2.15
300 and later to 2.14 and earlier requires additional steps. (This is dif‐
301 ferent from upgrading a database schema, which is covered later under
302 Upgrading or Downgrading a Database.)
303 For all service models it’s required to:
305 1. Stop all ovsdb-server processes (single process for standalone ser‐
307 vice model, all involved processes for active-backup and cluster
308 service models).
309 2. Compact all database files with ovsdb-tool compact command.
311 3. Downgrade and restart ovsdb-server processes.
313 Understanding Cluster Consistency
315 To ensure consistency, clustered OVSDB uses the Raft algorithm de‐
316 scribed in Diego Ongaro’s Ph.D. thesis, “Consensus: Bridging Theory and
317 Practice”. In an operational Raft cluster, at any given time a single
318 server is the “leader” and the other nodes are “followers”. Only the
319 leader processes transactions, but a transaction is only committed when
320 a majority of the servers confirm to the leader that they have written
321 it to persistent storage.
322 In most database systems, read and write access to the database happens
324 through transactions. In such a system, Raft allows a cluster to
325 present a strongly consistent transactional interface. OVSDB uses con‐
326 ventional transactions for writes, but clients often effectively do
327 reads a different way, by asking the server to “monitor” a database or
328 a subset of one on the client’s behalf. Whenever monitored data
329 changes, the server automatically tells the client what changed, which
330 allows the client to maintain an accurate snapshot of the database in
331 its memory. Of course, at any given time, the snapshot may be somewhat
332 dated since some of it could have changed without the change notifica‐
333 tion yet being received and processed by the client.
334 Given this unconventional usage model, OVSDB also adopts an unconven‐
336 tional clustering model. Each server in a cluster acts independently
337 for the purpose of monitors and read-only transactions, without verify‐
338 ing that data is up-to-date with the leader. Servers forward transac‐
339 tions that write to the database to the leader for execution, ensuring
340 consistency. This has the following consequences:
341 • Transactions that involve writes, against any server in the cluster,
343 are linearizable if clients take care to use correct prerequisites,
344 which is the same condition required for linearizability in a stand‐
345 alone OVSDB. (Actually, “at-least-once” consistency, because OVSDB
346 does not have a session mechanism to drop duplicate transactions if a
347 connection drops after the server commits it but before the client
348 receives the result.)
349 • Read-only transactions can yield results based on a stale version of
351 the database, if they are executed against a follower. Transactions
352 on the leader always yield fresh results. (With monitors, as ex‐
353 plained above, a client can always see stale data even without clus‐
354 tering, so clustering does not change the consistency model for moni‐
355 tors.)
356 • Monitor-based (or read-heavy) workloads scale well across a cluster,
358 because clustering OVSDB adds no additional work or communication for
359 reads and monitors.
360 • A write-heavy client should connect to the leader, to avoid the over‐
362 head of followers forwarding transactions to the leader.
363 • When a client conducts a mix of read and write transactions across
365 more than one server in a cluster, it can see inconsistent results
366 because a read transaction might read stale data whose updates have
367 not yet propagated from the leader. By default, utilities such as
368 ovn-sbctl (in OVN) connect to the cluster leader to avoid this issue.
369 The same might occur for transactions against a single follower ex‐
371 cept that the OVSDB server ensures that the results of a write for‐
372 warded to the leader by a given server are visible at that server be‐
373 fore it replies to the requesting client.
374 • If a client uses a database on one server in a cluster, then another
376 server in the cluster (perhaps because the first server failed), the
377 client could observe stale data. Clustered OVSDB clients, however,
378 can use a column in the _Server database to detect that data on a
379 server is older than data that the client previously read. The OVSDB
380 client library in Open vSwitch uses this feature to avoid servers
381 with stale data.
382
384 OVSDB can layer replication on top of any of its service models.
385 Replication, in this context, means to make, and keep up-to-date, a
386 read-only copy of the contents of a database (the replica). One use of
387 replication is to keep an up-to-date backup of a database. A replica
388 used solely for backup would not need to support clients of its own. A
389 set of replicas that do serve clients could be used to scale out read
390 access to the primary database.
391 A database replica is set up in the same way as a backup server in an
393 active-backup pair, with the difference that the replica is never pro‐
394 moted to an active role.
395 A database can have multiple replicas.
397 Open vSwitch 2.6 introduced support for database replication.
399
401 An OVSDB connection method is a string that specifies how to make a
402 JSON-RPC connection between an OVSDB client and server. Connection
403 methods are part of the Open vSwitch implementation of OVSDB and not
404 specified by RFC 7047. ovsdb-server uses connection methods to specify
405 how it should listen for connections from clients and ovsdb-client uses
406 them to specify how it should connect to a server. Connections in the
407 opposite direction, where ovsdb-server connects to a client that is
408 configured to listen for an incoming connection, are also possible.
409 Connection methods are classified as active or passive. An active con‐
411 nection method makes an outgoing connection to a remote host; a passive
412 connection method listens for connections from remote hosts. The most
413 common arrangement is to configure an OVSDB server with passive connec‐
414 tion methods and clients with active ones, but the OVSDB implementation
415 in Open vSwitch supports the opposite arrangement as well.
416 OVSDB supports the following active connection methods:
418 ssl:<host>:<port>
420 The specified SSL or TLS <port> on the given <host>.
421 tcp:<host>:<port>
423 The specified TCP <port> on the given <host>.
424 unix:<file>
426 On Unix-like systems, connect to the Unix domain server socket
427 named <file>.
428 On Windows, connect to a local named pipe that is represented by
430 a file created in the path <file> to mimic the behavior of a
431 Unix domain socket.
432 <method1>,<method2>,…,<methodN>
434 For a clustered database service to be highly available, a
435 client must be able to connect to any of the servers in the
436 cluster. To do so, specify connection methods for each of the
437 servers separated by commas (and optional spaces).
438 In theory, if machines go up and down and IP addresses change in
440 the right way, a client could talk to the wrong instance of a
441 database. To avoid this possibility, add cid:<uuid> to the list
442 of methods, where <uuid> is the cluster ID of the desired data‐
443 base cluster, as printed by ovsdb-tool db-cid. This feature is
444 optional.
445 OVSDB supports the following passive connection methods:
447 pssl:<port>[:<ip>]
449 Listen on the given TCP <port> for SSL or TLS connections. By
450 default, connections are not bound to a particular local IP ad‐
451 dress. Specifying <ip> limits connections to those from the
452 given IP.
453 ptcp:<port>[:<ip>]
455 Listen on the given TCP <port>. By default, connections are not
456 bound to a particular local IP address. Specifying <ip> limits
457 connections to those from the given IP.
458 punix:<file>
460 On Unix-like systems, listens for connections on the Unix domain
461 socket named <file>.
462 On Windows, listens on a local named pipe, creating a named pipe
464 <file> to mimic the behavior of a Unix domain socket. The ACLs
465 of the named pipe include LocalSystem, Administrators, and Cre‐
466 ator Owner.
467 All IP-based connection methods accept IPv4 and IPv6 addresses. To
469 specify an IPv6 address, wrap it in square brackets, e.g.
470 ssl:[::1]:6640. Passive IP-based connection methods by default listen
471 for IPv4 connections only; use [::] as the address to accept both IPv4
472 and IPv6 connections, e.g. pssl:6640:[::]. DNS names are also accepted
473 if built with unbound library. On Linux, use %<device> to designate a
474 scope for IPv6 link-level addresses, e.g. ssl:[fe80::1234%eth0]:6653.
475 The <port> may be omitted from connection methods that use a port num‐
477 ber. The default <port> for TCP-based connection methods is 6640, e.g.
478 pssl: is equivalent to pssl:6640. In Open vSwitch prior to version
479 2.4.0, the default port was 6632. To avoid incompatibility between
480 older and newer versions, we encourage users to specify a port number.
481 The ssl and pssl connection methods requires additional configuration
483 through --private-key, --certificate, and --ca-cert command line op‐
484 tions. Open vSwitch can be built without SSL support, in which case
485 these connection methods are not supported.
486
488 This section describes how to handle various events in the life cycle
489 of a database using the Open vSwitch implementation of OVSDB.
490 Creating a Database
492 Creating and starting up the service for a new database was covered
493 separately for each database service model in the Service Models sec‐
494 tion, above.
495 Backing Up and Restoring a Database
497 OVSDB is often used in contexts where the database contents are not
498 particularly valuable. For example, in many systems, the database for
499 configuring ovs-vswitchd is essentially rebuilt from scratch at boot
500 time. It is not worthwhile to back up these databases.
501 When OVSDB is used for valuable data, a backup strategy is worth con‐
503 sidering. One way is to use database replication, discussed above in
504 Database Replication which keeps an online, up-to-date copy of a data‐
505 base, possibly on a remote system. This works with all OVSDB service
506 models.
507 A more common backup strategy is to periodically take and store a snap‐
509 shot. For the standalone and active-backup service models, making a
510 copy of the database file, e.g. using cp, effectively makes a snapshot,
511 and because OVSDB database files are append-only, it works even if the
512 database is being modified when the snapshot takes place. This ap‐
513 proach does not work for clustered databases.
514 Another way to make a backup, which works with all OVSDB service mod‐
516 els, is to use ovsdb-client backup, which connects to a running data‐
517 base server and outputs an atomic snapshot of its schema and content,
518 in the same format used for standalone and active-backup databases.
519 Multiple options are also available when the time comes to restore a
521 database from a backup. For the standalone and active-backup service
522 models, one option is to stop the database server or servers, overwrite
523 the database file with the backup (e.g. with cp), and then restart the
524 servers. Another way, which works with any service model, is to use
525 ovsdb-client restore, which connects to a running database server and
526 replaces the data in one of its databases by a provided snapshot. The
527 advantage of ovsdb-client restore is that it causes zero downtime for
528 the database and its server. It has the downside that UUIDs of rows in
529 the restored database will differ from those in the snapshot, because
530 the OVSDB protocol does not allow clients to specify row UUIDs.
531 None of these approaches saves and restores data in columns that the
533 schema designates as ephemeral. This is by design: the designer of a
534 schema only marks a column as ephemeral if it is acceptable for its
535 data to be lost when a database server restarts.
536 Clustering and backup serve different purposes. Clustering increases
538 availability, but it does not protect against data loss if, for exam‐
539 ple, a malicious or malfunctioning OVSDB client deletes or tampers with
540 data.
541 Changing Database Service Model
543 Use ovsdb-tool create-cluster to create a clustered database from the
544 contents of a standalone database. Use ovsdb-client backup to create a
545 standalone database from the contents of a running clustered database.
546 When the cluster is down and cannot be revived, ovsdb-client backup
547 will not work.
548 Use ovsdb-tool cluster-to-standalone to convert clustered database to
550 standalone database when the cluster is down and cannot be revived.
551 Upgrading or Downgrading a Database
553 The evolution of a piece of software can require changes to the schemas
554 of the databases that it uses. For example, new features might require
555 new tables or new columns in existing tables, or conceptual changes
556 might require a database to be reorganized in other ways. In some
557 cases, the easiest way to deal with a change in a database schema is to
558 delete the existing database and start fresh with the new schema, espe‐
559 cially if the data in the database is easy to reconstruct. But in many
560 other cases, it is better to convert the database from one schema to
561 another.
562 The OVSDB implementation in Open vSwitch has built-in support for some
564 simple cases of converting a database from one schema to another. This
565 support can handle changes that add or remove database columns or ta‐
566 bles or that eliminate constraints (for example, changing a column that
567 must have exactly one value into one that has one or more values). It
568 can also handle changes that add constraints or make them stricter, but
569 only if the existing data in the database satisfies the new constraints
570 (for example, changing a column that has one or more values into a col‐
571 umn with exactly one value, if every row in the column has exactly one
572 value). The built-in conversion can cause data loss in obvious ways,
573 for example if the new schema removes tables or columns, or indirectly,
574 for example by deleting unreferenced rows in tables that the new schema
575 marks for garbage collection.
576 Converting a database can lose data, so it is wise to make a backup be‐
578 forehand.
579 To use OVSDB’s built-in support for schema conversion with a standalone
581 or active-backup database, first stop the database server or servers,
582 then use ovsdb-tool convert to convert it to the new schema, and then
583 restart the database server.
584 OVSDB also supports online database schema conversion for any of its
586 database service models. To convert a database online, use
587 ovsdb-client convert. The conversion is atomic, consistent, isolated,
588 and durable. ovsdb-server disconnects any clients connected when the
589 conversion takes place (except clients that use the set_db_change_aware
590 Open vSwitch extension RPC). Upon reconnection, clients will discover
591 that the schema has changed.
592 Schema versions and checksums (see Schemas above) can give hints about
594 whether a database needs to be converted to a new schema. If there is
595 any question, though, the needs-conversion command on ovsdb-tool and
596 ovsdb-client can provide a definitive answer.
597 Working with Database History
599 Both on-disk database formats that OVSDB supports are organized as a
600 stream of transaction records. Each record describes a change to the
601 database as a list of rows that were inserted or deleted or modified,
602 along with the details. Therefore, in normal operation, a database
603 file only grows, as each change causes another record to be appended at
604 the end. Usually, a user has no need to understand this file struc‐
605 ture. This section covers some exceptions.
606 Compacting Databases
608 If OVSDB database files were truly append-only, then over time they
609 would grow without bound. To avoid this problem, OVSDB can compact a
610 database file, that is, replace it by a new version that contains only
611 the current database contents, as if it had been inserted by a single
612 transaction. From time to time, ovsdb-server automatically compacts a
613 database that grows much larger than its minimum size.
614 Because ovsdb-server automatically compacts databases, it is usually
616 not necessary to compact them manually, but OVSDB still offers a few
617 ways to do it. First, ovsdb-tool compact can compact a standalone or
618 active-backup database that is not currently being served by
619 ovsdb-server (or otherwise locked for writing by another process). To
620 compact any database that is currently being served by ovsdb-server,
621 use ovs-appctl to send the ovsdb-server/compact command. Each server
622 in an active-backup or clustered database maintains its database file
623 independently, so to compact all of them, issue this command separately
624 on each server.
625 Viewing History
627 The ovsdb-tool utility’s show-log command displays the transaction
628 records in an OVSDB database file in a human-readable format. By de‐
629 fault, it shows minimal detail, but adding the option -m once or twice
630 increases the level of detail. In addition to the transaction data, it
631 shows the time and date of each transaction and any “comment” added to
632 the transaction by the client. The comments can be helpful for quickly
633 understanding a transaction; for example, ovs-vsctl adds its command
634 line to the transactions that it makes.
635 The show-log command works with both OVSDB file formats, but the de‐
637 tails of the output format differ. For active-backup and clustered
638 databases, the sequence of transactions in each server’s log will dif‐
639 fer, even at points when they reflect the same data.
640 Truncating History
642 It may occasionally be useful to “roll back” a database file to an ear‐
643 lier point. Because of the organization of OVSDB records, this is easy
644 to do. Start by noting the record number <i> of the first record to
645 delete in ovsdb-tool show-log output. Each record is two lines of
646 plain text, so trimming the log is as simple as running head -n <j>,
647 where <j> = 2 * <i>.
648 Corruption
650 When ovsdb-server opens an OVSDB database file, of any kind, it reads
651 as many transaction records as it can from the file until it reaches
652 the end of the file or it encounters a corrupted record. At that point
653 it stops reading and regards the data that it has read to this point as
654 the full contents of the database file, effectively rolling the data‐
655 base back to an earlier point.
656 Each transaction record contains an embedded SHA-1 checksum, which the
658 server verifies as it reads a database file. It detects corruption
659 when a checksum fails to verify. Even though SHA-1 is no longer con‐
660 sidered secure for use in cryptography, it is acceptable for this pur‐
661 pose because it is not used to defend against malicious attackers.
662 The first record in a standalone or active-backup database file speci‐
664 fies the schema. ovsdb-server will refuse to work with a database
665 where this record is corrupted, or with a clustered database file with
666 corruption in the first few records. Delete and recreate such a data‐
667 base, or restore it from a backup.
668 When ovsdb-server adds records to a database file in which it detected
670 corruption, it first truncates the file just after the last good
671 record.
672
674 RFC 7047, “The Open vSwitch Database Management Protocol.”
675 Open vSwitch implementations of generic OVSDB functionality:
677 ovsdb-server(1), ovsdb-client(1), ovsdb-tool(1).
678 Tools for working with databases that have specific OVSDB schemas:
680 ovs-vsctl(8), vtep-ctl(8), and (in OVN) ovn-nbctl(8), ovn-sbctl(8).
681 OVSDB schemas for Open vSwitch and related functionality:
683 ovs-vswitchd.conf.db(5), vtep(5), and (in OVN) ovn-nb(5), ovn-sb(5).
684
686 The Open vSwitch Development Community
687
689 2021, The Open vSwitch Development Community
6902.15 Jul 22, 2021 OVSDB(7)