4 VACUUM — garbage-collect and optionally analyze a database
8 VACUUM [ ( option [, ...] ) ] [ table_and_columns [, ...] ]
10 where option can be one of:
16 DISABLE_PAGE_SKIPPING [ boolean ]
17 SKIP_LOCKED [ boolean ]
18 INDEX_CLEANUP { AUTO | ON | OFF }
19 PROCESS_MAIN [ boolean ]
20 PROCESS_TOAST [ boolean ]
23 SKIP_DATABASE_STATS [ boolean ]
24 ONLY_DATABASE_STATS [ boolean ]
25 BUFFER_USAGE_LIMIT size
27 and table_and_columns is:
29 [ ONLY ] table_name [ * ] [ ( column_name [, ...] ) ]
33 VACUUM reclaims storage occupied by dead tuples. In normal PostgreSQL
34 operation, tuples that are deleted or obsoleted by an update are not
35 physically removed from their table; they remain present until a VACUUM
36 is done. Therefore it's necessary to do VACUUM periodically, especially
37 on frequently-updated tables.
39 Without a table_and_columns list, VACUUM processes every table and
40 materialized view in the current database that the current user has
41 permission to vacuum. With a list, VACUUM processes only those
44 VACUUM ANALYZE performs a VACUUM and then an ANALYZE for each selected
45 table. This is a handy combination form for routine maintenance
46 scripts. See ANALYZE for more details about its processing.
48 Plain VACUUM (without FULL) simply reclaims space and makes it
49 available for re-use. This form of the command can operate in parallel
50 with normal reading and writing of the table, as an exclusive lock is
51 not obtained. However, extra space is not returned to the operating
52 system (in most cases); it's just kept available for re-use within the
53 same table. It also allows us to leverage multiple CPUs in order to
54 process indexes. This feature is known as parallel vacuum. To disable
55 this feature, one can use PARALLEL option and specify parallel workers
56 as zero. VACUUM FULL rewrites the entire contents of the table into a
57 new disk file with no extra space, allowing unused space to be returned
58 to the operating system. This form is much slower and requires an
59 ACCESS EXCLUSIVE lock on each table while it is being processed.
64 Selects “full” vacuum, which can reclaim more space, but takes
65 much longer and exclusively locks the table. This method also
66 requires extra disk space, since it writes a new copy of the
67 table and doesn't release the old copy until the operation is
68 complete. Usually this should only be used when a significant
69 amount of space needs to be reclaimed from within the table.
72 Selects aggressive “freezing” of tuples. Specifying FREEZE is
73 equivalent to performing VACUUM with the vacuum_freeze_min_age
74 and vacuum_freeze_table_age parameters set to zero. Aggressive
75 freezing is always performed when the table is rewritten, so
76 this option is redundant when FULL is specified.
79 Prints a detailed vacuum activity report for each table at INFO
83 Updates statistics used by the planner to determine the most
84 efficient way to execute a query.
87 Normally, VACUUM will skip pages based on the visibility map.
88 Pages where all tuples are known to be frozen can always be
89 skipped, and those where all tuples are known to be visible to
90 all transactions may be skipped except when performing an
91 aggressive vacuum. Furthermore, except when performing an
92 aggressive vacuum, some pages may be skipped in order to avoid
93 waiting for other sessions to finish using them. This option
94 disables all page-skipping behavior, and is intended to be used
95 only when the contents of the visibility map are suspect, which
96 should happen only if there is a hardware or software issue
97 causing database corruption.
100 Specifies that VACUUM should not wait for any conflicting locks
101 to be released when beginning work on a relation: if a relation
102 cannot be locked immediately without waiting, the relation is
103 skipped. Note that even with this option, VACUUM may still block
104 when opening the relation's indexes. Additionally, VACUUM
105 ANALYZE may still block when acquiring sample rows from
106 partitions, table inheritance children, and some types of
107 foreign tables. Also, while VACUUM ordinarily processes all
108 partitions of specified partitioned tables, this option will
109 cause VACUUM to skip all partitions if there is a conflicting
110 lock on the partitioned table.
113 Normally, VACUUM will skip index vacuuming when there are very
114 few dead tuples in the table. The cost of processing all of the
115 table's indexes is expected to greatly exceed the benefit of
116 removing dead index tuples when this happens. This option can be
117 used to force VACUUM to process indexes when there are more than
118 zero dead tuples. The default is AUTO, which allows VACUUM to
119 skip index vacuuming when appropriate. If INDEX_CLEANUP is set
120 to ON, VACUUM will conservatively remove all dead tuples from
121 indexes. This may be useful for backwards compatibility with
122 earlier releases of PostgreSQL where this was the standard
125 INDEX_CLEANUP can also be set to OFF to force VACUUM to always
126 skip index vacuuming, even when there are many dead tuples in
127 the table. This may be useful when it is necessary to make
128 VACUUM run as quickly as possible to avoid imminent transaction
129 ID wraparound (see Section 24.1.5). However, the wraparound
130 failsafe mechanism controlled by vacuum_failsafe_age will
131 generally trigger automatically to avoid transaction ID
132 wraparound failure, and should be preferred. If index cleanup is
133 not performed regularly, performance may suffer, because as the
134 table is modified indexes will accumulate dead tuples and the
135 table itself will accumulate dead line pointers that cannot be
136 removed until index cleanup is completed.
138 This option has no effect for tables that have no index and is
139 ignored if the FULL option is used. It also has no effect on the
140 transaction ID wraparound failsafe mechanism. When triggered it
141 will skip index vacuuming, even when INDEX_CLEANUP is set to ON.
144 Specifies that VACUUM should attempt to process the main
145 relation. This is usually the desired behavior and is the
146 default. Setting this option to false may be useful when it is
147 only necessary to vacuum a relation's corresponding TOAST table.
150 Specifies that VACUUM should attempt to process the
151 corresponding TOAST table for each relation, if one exists. This
152 is usually the desired behavior and is the default. Setting this
153 option to false may be useful when it is only necessary to
154 vacuum the main relation. This option is required when the FULL
158 Specifies that VACUUM should attempt to truncate off any empty
159 pages at the end of the table and allow the disk space for the
160 truncated pages to be returned to the operating system. This is
161 normally the desired behavior and is the default unless
162 vacuum_truncate is set to false or the vacuum_truncate option
163 has been set to false for the table to be vacuumed. Setting this
164 option to false may be useful to avoid ACCESS EXCLUSIVE lock on
165 the table that the truncation requires. This option is ignored
166 if the FULL option is used.
169 Perform index vacuum and index cleanup phases of VACUUM in
170 parallel using integer background workers (for the details of
171 each vacuum phase, please refer to Table 27.46). The number of
172 workers used to perform the operation is equal to the number of
173 indexes on the relation that support parallel vacuum which is
174 limited by the number of workers specified with PARALLEL option
175 if any which is further limited by
176 max_parallel_maintenance_workers. An index can participate in
177 parallel vacuum if and only if the size of the index is more
178 than min_parallel_index_scan_size. Please note that it is not
179 guaranteed that the number of parallel workers specified in
180 integer will be used during execution. It is possible for a
181 vacuum to run with fewer workers than specified, or even with no
182 workers at all. Only one worker can be used per index. So
183 parallel workers are launched only when there are at least 2
184 indexes in the table. Workers for vacuum are launched before the
185 start of each phase and exit at the end of the phase. These
186 behaviors might change in a future release. This option can't be
187 used with the FULL option.
190 Specifies that VACUUM should skip updating the database-wide
191 statistics about oldest unfrozen XIDs. Normally VACUUM will
192 update these statistics once at the end of the command. However,
193 this can take awhile in a database with a very large number of
194 tables, and it will accomplish nothing unless the table that had
195 contained the oldest unfrozen XID was among those vacuumed.
196 Moreover, if multiple VACUUM commands are issued in parallel,
197 only one of them can update the database-wide statistics at a
198 time. Therefore, if an application intends to issue a series of
199 many VACUUM commands, it can be helpful to set this option in
200 all but the last such command; or set it in all the commands and
201 separately issue VACUUM (ONLY_DATABASE_STATS) afterwards.
204 Specifies that VACUUM should do nothing except update the
205 database-wide statistics about oldest unfrozen XIDs. When this
206 option is specified, the table_and_columns list must be empty,
207 and no other option may be enabled except VERBOSE.
210 Specifies the Buffer Access Strategy ring buffer size for
211 VACUUM. This size is used to calculate the number of shared
212 buffers which will be reused as part of this strategy. 0
213 disables use of a Buffer Access Strategy. If ANALYZE is also
214 specified, the BUFFER_USAGE_LIMIT value is used for both the
215 vacuum and analyze stages. This option can't be used with the
216 FULL option except if ANALYZE is also specified. When this
217 option is not specified, VACUUM uses the value from
218 vacuum_buffer_usage_limit. Higher settings can allow VACUUM to
219 run more quickly, but having too large a setting may cause too
220 many other useful pages to be evicted from shared buffers. The
221 minimum value is 128 kB and the maximum value is 16 GB.
224 Specifies whether the selected option should be turned on or
225 off. You can write TRUE, ON, or 1 to enable the option, and
226 FALSE, OFF, or 0 to disable it. The boolean value can also be
227 omitted, in which case TRUE is assumed.
230 Specifies a non-negative integer value passed to the selected
234 Specifies an amount of memory in kilobytes. Sizes may also be
235 specified as a string containing the numerical size followed by
236 any one of the following memory units: B (bytes), kB
237 (kilobytes), MB (megabytes), GB (gigabytes), or TB (terabytes).
240 The name (optionally schema-qualified) of a specific table or
241 materialized view to vacuum. If ONLY is specified before the
242 table name, only that table is vacuumed. If ONLY is not
243 specified, the table and all its inheritance child tables or
244 partitions (if any) are also vacuumed. Optionally, * can be
245 specified after the table name to explicitly indicate that
246 inheritance child tables (or partitions) are to be vacuumed.
249 The name of a specific column to analyze. Defaults to all
250 columns. If a column list is specified, ANALYZE must also be
255 When VERBOSE is specified, VACUUM emits progress messages to indicate
256 which table is currently being processed. Various statistics about the
257 tables are printed as well.
261 To vacuum a table, one must ordinarily have the MAINTAIN privilege on
262 the table. However, database owners are allowed to vacuum all tables in
263 their databases, except shared catalogs. VACUUM will skip over any
264 tables that the calling user does not have permission to vacuum.
266 While VACUUM is running, the search_path is temporarily changed to
269 VACUUM cannot be executed inside a transaction block.
271 For tables with GIN indexes, VACUUM (in any form) also completes any
272 pending index insertions, by moving pending index entries to the
273 appropriate places in the main GIN index structure. See
274 Section 65.4.4.1 for details.
276 We recommend that all databases be vacuumed regularly in order to
277 remove dead rows. PostgreSQL includes an “autovacuum” facility which
278 can automate routine vacuum maintenance. For more information about
279 automatic and manual vacuuming, see Section 24.1.
281 The FULL option is not recommended for routine use, but might be useful
282 in special cases. An example is when you have deleted or updated most
283 of the rows in a table and would like the table to physically shrink to
284 occupy less disk space and allow faster table scans. VACUUM FULL will
285 usually shrink the table more than a plain VACUUM would.
287 The PARALLEL option is used only for vacuum purposes. If this option is
288 specified with the ANALYZE option, it does not affect ANALYZE.
290 VACUUM causes a substantial increase in I/O traffic, which might cause
291 poor performance for other active sessions. Therefore, it is sometimes
292 advisable to use the cost-based vacuum delay feature. For parallel
293 vacuum, each worker sleeps in proportion to the work done by that
294 worker. See Section 19.10.2 for details.
296 Each backend running VACUUM without the FULL option will report its
297 progress in the pg_stat_progress_vacuum view. Backends running VACUUM
298 FULL will instead report their progress in the pg_stat_progress_cluster
299 view. See Section 27.4.5 and Section 27.4.2 for details.
303 To clean a single table onek, analyze it for the optimizer and print a
304 detailed vacuum activity report:
305 VACUUM (VERBOSE, ANALYZE) onek;
309 There is no VACUUM statement in the SQL standard.
311 The following syntax was used before PostgreSQL version 9.0 and is
313 VACUUM [ FULL ] [ FREEZE ] [ VERBOSE ] [ ANALYZE ] [ table_and_columns [, ...] ]
315 Note that in this syntax, the options must be specified in exactly the
320 vacuumdb, Section 19.10.2, Section 24.1.6, Section 27.4.5,