2 doc/src/sgml/ref/lock.sgml
3 PostgreSQL documentation
6 <refentry id=
"sql-lock">
7 <indexterm zone=
"sql-lock">
8 <primary>LOCK
</primary>
12 <refentrytitle>LOCK
</refentrytitle>
13 <manvolnum>7</manvolnum>
14 <refmiscinfo>SQL - Language Statements
</refmiscinfo>
18 <refname>LOCK
</refname>
19 <refpurpose>lock a table
</refpurpose>
24 LOCK [ TABLE ] [ ONLY ]
<replaceable class=
"parameter">name
</replaceable> [ * ] [, ...] [ IN
<replaceable class=
"parameter">lockmode
</replaceable> MODE ] [ NOWAIT ]
26 <phrase>where
<replaceable class=
"parameter">lockmode
</replaceable> is one of:
</phrase>
28 ACCESS SHARE | ROW SHARE | ROW EXCLUSIVE | SHARE UPDATE EXCLUSIVE
29 | SHARE | SHARE ROW EXCLUSIVE | EXCLUSIVE | ACCESS EXCLUSIVE
34 <title>Description
</title>
37 <command>LOCK TABLE
</command> obtains a table-level lock, waiting
38 if necessary for any conflicting locks to be released. If
39 <literal>NOWAIT
</literal> is specified,
<command>LOCK
40 TABLE
</command> does not wait to acquire the desired lock: if it
41 cannot be acquired immediately, the command is aborted and an
42 error is emitted. Once obtained, the lock is held for the
43 remainder of the current transaction. (There is no
<command>UNLOCK
44 TABLE
</command> command; locks are always released at transaction
49 When a view is locked, all relations appearing in the view definition
50 query are also locked recursively with the same lock mode.
54 When acquiring locks automatically for commands that reference
55 tables,
<productname>PostgreSQL
</productname> always uses the least
56 restrictive lock mode possible.
<command>LOCK TABLE
</command>
57 provides for cases when you might need more restrictive locking.
58 For example, suppose an application runs a transaction at the
59 <literal>READ COMMITTED
</literal> isolation level and needs to ensure that
60 data in a table remains stable for the duration of the transaction.
61 To achieve this you could obtain
<literal>SHARE
</literal> lock mode over the
62 table before querying. This will prevent concurrent data changes
63 and ensure subsequent reads of the table see a stable view of
64 committed data, because
<literal>SHARE
</literal> lock mode conflicts with
65 the
<literal>ROW EXCLUSIVE
</literal> lock acquired by writers, and your
66 <command>LOCK TABLE
<replaceable
67 class=
"parameter">name
</replaceable> IN SHARE MODE
</command>
68 statement will wait until any concurrent holders of
<literal>ROW
69 EXCLUSIVE
</literal> mode locks commit or roll back. Thus, once you
70 obtain the lock, there are no uncommitted writes outstanding;
71 furthermore none can begin until you release the lock.
75 To achieve a similar effect when running a transaction at the
76 <literal>REPEATABLE READ
</literal> or
<literal>SERIALIZABLE
</literal>
77 isolation level, you have to execute the
<command>LOCK TABLE
</command> statement
78 before executing any
<command>SELECT
</command> or data modification statement.
79 A
<literal>REPEATABLE READ
</literal> or
<literal>SERIALIZABLE
</literal> transaction's
80 view of data will be frozen when its first
81 <command>SELECT
</command> or data modification statement begins. A
<command>LOCK
82 TABLE
</command> later in the transaction will still prevent concurrent writes
83 — but it won't ensure that what the transaction reads corresponds to
84 the latest committed values.
88 If a transaction of this sort is going to change the data in the
89 table, then it should use
<literal>SHARE ROW EXCLUSIVE
</literal> lock mode
90 instead of
<literal>SHARE
</literal> mode. This ensures that only one
91 transaction of this type runs at a time. Without this, a deadlock
92 is possible: two transactions might both acquire
<literal>SHARE
</literal>
93 mode, and then be unable to also acquire
<literal>ROW EXCLUSIVE
</literal>
94 mode to actually perform their updates. (Note that a transaction's
95 own locks never conflict, so a transaction can acquire
<literal>ROW
96 EXCLUSIVE
</literal> mode when it holds
<literal>SHARE
</literal> mode
— but not
97 if anyone else holds
<literal>SHARE
</literal> mode.) To avoid deadlocks,
98 make sure all transactions acquire locks on the same objects in the
99 same order, and if multiple lock modes are involved for a single
100 object, then transactions should always acquire the most
101 restrictive mode first.
105 More information about the lock modes and locking strategies can be
106 found in
<xref linkend=
"explicit-locking"/>.
111 <title>Parameters
</title>
115 <term><replaceable class=
"parameter">name
</replaceable></term>
118 The name (optionally schema-qualified) of an existing table to
119 lock. If
<literal>ONLY
</literal> is specified before the table name, only that
120 table is locked. If
<literal>ONLY
</literal> is not specified, the table and all
121 its descendant tables (if any) are locked. Optionally,
<literal>*
</literal>
122 can be specified after the table name to explicitly indicate that
123 descendant tables are included.
127 The command
<literal>LOCK TABLE a, b;
</literal> is equivalent to
128 <literal>LOCK TABLE a; LOCK TABLE b;
</literal>. The tables are locked
129 one-by-one in the order specified in the
<command>LOCK
130 TABLE
</command> command.
136 <term><replaceable class=
"parameter">lockmode
</replaceable></term>
139 The lock mode specifies which locks this lock conflicts with.
140 Lock modes are described in
<xref linkend=
"explicit-locking"/>.
144 If no lock mode is specified, then
<literal>ACCESS
145 EXCLUSIVE
</literal>, the most restrictive mode, is used.
151 <term><literal>NOWAIT
</literal></term>
154 Specifies that
<command>LOCK TABLE
</command> should not wait for
155 any conflicting locks to be released: if the specified lock(s)
156 cannot be acquired immediately without waiting, the transaction
168 To lock a table, the user must have the right privilege for the specified
169 <replaceable class=
"parameter">lockmode
</replaceable>, or be the table's
170 owner or a superuser. If the user has
171 <literal>UPDATE
</literal>,
<literal>DELETE
</literal>, or
172 <literal>TRUNCATE
</literal> privileges on the table, any
<replaceable
173 class=
"parameter">lockmode
</replaceable> is permitted. If the user has
174 <literal>INSERT
</literal> privileges on the table,
<literal>ROW EXCLUSIVE
175 MODE
</literal> (or a less-conflicting mode as described in
<xref
176 linkend=
"explicit-locking"/>) is permitted. If a user has
177 <literal>SELECT
</literal> privileges on the table,
<literal>ACCESS SHARE
178 MODE
</literal> is permitted.
182 The user performing the lock on the view must have the corresponding
183 privilege on the view. In addition, by default, the view's owner must
184 have the relevant privileges on the underlying base relations, whereas the
185 user performing the lock does not need any permissions on the underlying
186 base relations. However, if the view has
187 <literal>security_invoker
</literal> set to
<literal>true
</literal>
188 (see
<link linkend=
"sql-createview"><command>CREATE VIEW
</command></link>),
189 the user performing the lock, rather than the view owner, must have the
190 relevant privileges on the underlying base relations.
194 <command>LOCK TABLE
</command> is useless outside a transaction block: the lock
195 would remain held only to the completion of the statement. Therefore
196 <productname>PostgreSQL
</productname> reports an error if
<command>LOCK
</command>
197 is used outside a transaction block.
199 <link linkend=
"sql-begin"><command>BEGIN
</command></link> and
200 <link linkend=
"sql-commit"><command>COMMIT
</command></link>
201 (or
<link linkend=
"sql-rollback"><command>ROLLBACK
</command></link>)
202 to define a transaction block.
206 <command>LOCK TABLE
</command> only deals with table-level locks, and so
207 the mode names involving
<literal>ROW
</literal> are all misnomers. These
208 mode names should generally be read as indicating the intention of
209 the user to acquire row-level locks within the locked table. Also,
210 <literal>ROW EXCLUSIVE
</literal> mode is a shareable table lock. Keep in
211 mind that all the lock modes have identical semantics so far as
212 <command>LOCK TABLE
</command> is concerned, differing only in the rules
213 about which modes conflict with which. For information on how to
214 acquire an actual row-level lock, see
<xref linkend=
"locking-rows"/>
215 and
<xref linkend=
"sql-for-update-share"/>
216 in the
<xref linkend=
"sql-select"/> documentation.
221 <title>Examples
</title>
224 Obtain a
<literal>SHARE
</literal> lock on a primary key table when going to perform
225 inserts into a foreign key table:
229 LOCK TABLE films IN SHARE MODE;
231 WHERE name = 'Star Wars: Episode I - The Phantom Menace';
232 -- Do ROLLBACK if record was not returned
233 INSERT INTO films_user_comments VALUES
234 (_id_, 'GREAT! I was waiting for it for so long!');
240 Take a
<literal>SHARE ROW EXCLUSIVE
</literal> lock on a primary key table when going to perform
245 LOCK TABLE films IN SHARE ROW EXCLUSIVE MODE;
246 DELETE FROM films_user_comments WHERE id IN
247 (SELECT id FROM films WHERE rating
< 5);
248 DELETE FROM films WHERE rating
< 5;
250 </programlisting></para>
254 <title>Compatibility
</title>
257 There is no
<command>LOCK TABLE
</command> in the SQL standard,
258 which instead uses
<command>SET TRANSACTION
</command> to specify
259 concurrency levels on transactions.
<productname>PostgreSQL
</productname> supports that too;
260 see
<xref linkend=
"sql-set-transaction"/> for details.
264 Except for
<literal>ACCESS SHARE
</literal>,
<literal>ACCESS EXCLUSIVE
</literal>,
265 and
<literal>SHARE UPDATE EXCLUSIVE
</literal> lock modes, the
266 <productname>PostgreSQL
</productname> lock modes and the
267 <command>LOCK TABLE
</command> syntax are compatible with those
268 present in
<productname>Oracle
</productname>.