1 <?xml version="1.0" encoding="UTF-8" standalone="no"?>
2 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>pg_basebackup</title><link rel="stylesheet" type="text/css" href="stylesheet.css" /><link rev="made" href="pgsql-docs@lists.postgresql.org" /><meta name="generator" content="DocBook XSL Stylesheets Vsnapshot" /><link rel="prev" href="app-pgamcheck.html" title="pg_amcheck" /><link rel="next" href="pgbench.html" title="pgbench" /></head><body id="docContent" class="container-fluid col-10"><div class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="5" align="center"><span class="application">pg_basebackup</span></th></tr><tr><td width="10%" align="left"><a accesskey="p" href="app-pgamcheck.html" title="pg_amcheck">Prev</a> </td><td width="10%" align="left"><a accesskey="u" href="reference-client.html" title="PostgreSQL Client Applications">Up</a></td><th width="60%" align="center">PostgreSQL Client Applications</th><td width="10%" align="right"><a accesskey="h" href="index.html" title="PostgreSQL 18.0 Documentation">Home</a></td><td width="10%" align="right"> <a accesskey="n" href="pgbench.html" title="pgbench">Next</a></td></tr></table><hr /></div><div class="refentry" id="APP-PGBASEBACKUP"><div class="titlepage"></div><a id="id-1.9.4.10.1" class="indexterm"></a><div class="refnamediv"><h2><span class="refentrytitle"><span class="application">pg_basebackup</span></span></h2><p>pg_basebackup — take a base backup of a <span class="productname">PostgreSQL</span> cluster</p></div><div class="refsynopsisdiv"><h2>Synopsis</h2><div class="cmdsynopsis"><p id="id-1.9.4.10.4.1"><code class="command">pg_basebackup</code> [<em class="replaceable"><code>option</code></em>...]</p></div></div><div class="refsect1" id="id-1.9.4.10.5"><h2>Description</h2><p>
3 <span class="application">pg_basebackup</span> is used to take a base backup of
4 a running <span class="productname">PostgreSQL</span> database cluster. The backup
5 is taken without affecting other clients of the database, and can be used
6 both for point-in-time recovery (see <a class="xref" href="continuous-archiving.html" title="25.3. Continuous Archiving and Point-in-Time Recovery (PITR)">Section 25.3</a>)
7 and as the starting point for a log-shipping or streaming-replication standby
8 server (see <a class="xref" href="warm-standby.html" title="26.2. Log-Shipping Standby Servers">Section 26.2</a>).
10 <span class="application">pg_basebackup</span> can take a full or incremental
11 base backup of the database. When used to take a full backup, it makes an
12 exact copy of the database cluster's files. When used to take an incremental
13 backup, some files that would have been part of a full backup may be
14 replaced with incremental versions of the same files, containing only those
15 blocks that have been modified since the reference backup. An incremental
16 backup cannot be used directly; instead,
17 <a class="xref" href="app-pgcombinebackup.html" title="pg_combinebackup"><span class="refentrytitle"><span class="application">pg_combinebackup</span></span></a> must first
18 be used to combine it with the previous backups upon which it depends.
19 See <a class="xref" href="continuous-archiving.html#BACKUP-INCREMENTAL-BACKUP" title="25.3.3. Making an Incremental Backup">Section 25.3.3</a> for more information
20 about incremental backups, and <a class="xref" href="continuous-archiving.html#BACKUP-PITR-RECOVERY" title="25.3.5. Recovering Using a Continuous Archive Backup">Section 25.3.5</a>
21 for steps to recover from a backup.
23 In any mode, <span class="application">pg_basebackup</span> makes sure the server
24 is put into and out of backup mode automatically. Backups are always taken of
25 the entire database cluster; it is not possible to back up individual
26 databases or database objects. For selective backups, another tool such as
27 <a class="xref" href="app-pgdump.html" title="pg_dump"><span class="refentrytitle"><span class="application">pg_dump</span></span></a> must be used.
29 The backup is made over a regular <span class="productname">PostgreSQL</span>
30 connection that uses the replication protocol. The connection must be made
31 with a user ID that has <code class="literal">REPLICATION</code> permissions
32 (see <a class="xref" href="role-attributes.html" title="21.2. Role Attributes">Section 21.2</a>) or is a superuser,
33 and <a class="link" href="auth-pg-hba-conf.html" title="20.1. The pg_hba.conf File"><code class="filename">pg_hba.conf</code></a>
34 must permit the replication connection. The server must also be configured
35 with <a class="xref" href="runtime-config-replication.html#GUC-MAX-WAL-SENDERS">max_wal_senders</a> set high enough to provide at
36 least one walsender for the backup plus one for WAL streaming (if used).
38 There can be multiple <code class="command">pg_basebackup</code>s running at the same time, but it is usually
39 better from a performance point of view to take only one backup, and copy
42 <span class="application">pg_basebackup</span> can make a base backup from
43 not only a primary server but also a standby. To take a backup from a standby,
44 set up the standby so that it can accept replication connections (that is, set
45 <code class="varname">max_wal_senders</code> and <a class="xref" href="runtime-config-replication.html#GUC-HOT-STANDBY">hot_standby</a>,
46 and configure its <code class="filename">pg_hba.conf</code> appropriately).
47 You will also need to enable <a class="xref" href="runtime-config-wal.html#GUC-FULL-PAGE-WRITES">full_page_writes</a> on the primary.
49 Note that there are some limitations in taking a backup from a standby:
51 </p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
52 The backup history file is not created in the database cluster backed up.
53 </p></li><li class="listitem"><p>
54 <span class="application">pg_basebackup</span> cannot force the standby
55 to switch to a new WAL file at the end of backup.
56 When you are using <code class="literal">-X none</code>, if write activity on
57 the primary is low, <span class="application">pg_basebackup</span> may
58 need to wait a long time for the last WAL file required for the backup
59 to be switched and archived. In this case, it may be useful to run
60 <code class="function">pg_switch_wal</code> on the primary in order to
61 trigger an immediate WAL file switch.
62 </p></li><li class="listitem"><p>
63 If the standby is promoted to be primary during backup, the backup fails.
64 </p></li><li class="listitem"><p>
65 All WAL records required for the backup must contain sufficient full-page writes,
66 which requires you to enable <code class="varname">full_page_writes</code> on the primary.
67 </p></li></ul></div><p>
69 Whenever <span class="application">pg_basebackup</span> is taking a base
70 backup, the server's <code class="structname">pg_stat_progress_basebackup</code>
71 view will report the progress of the backup.
72 See <a class="xref" href="progress-reporting.html#BASEBACKUP-PROGRESS-REPORTING" title="27.4.6. Base Backup Progress Reporting">Section 27.4.6</a> for details.
73 </p></div><div class="refsect1" id="id-1.9.4.10.6"><h2>Options</h2><p>
74 The following command-line options control the location and format of the
77 </p><div class="variablelist"><dl class="variablelist"><dt><span class="term"><code class="option">-D <em class="replaceable"><code>directory</code></em></code><br /></span><span class="term"><code class="option">--pgdata=<em class="replaceable"><code>directory</code></em></code></span></dt><dd><p>
78 Sets the target directory to write the output to.
79 <span class="application">pg_basebackup</span> will create this directory
80 (and any missing parent directories) if it does not exist. If it
81 already exists, it must be empty.
83 When the backup is in tar format, the target directory may be
84 specified as <code class="literal">-</code> (dash), causing the tar file to be
85 written to <code class="literal">stdout</code>.
87 This option is required.
88 </p></dd><dt><span class="term"><code class="option">-F <em class="replaceable"><code>format</code></em></code><br /></span><span class="term"><code class="option">--format=<em class="replaceable"><code>format</code></em></code></span></dt><dd><p>
89 Selects the format for the output. <em class="replaceable"><code>format</code></em>
90 can be one of the following:
92 </p><div class="variablelist"><dl class="variablelist"><dt><span class="term"><code class="literal">p</code><br /></span><span class="term"><code class="literal">plain</code></span></dt><dd><p>
93 Write the output as plain files, with the same layout as the
94 source server's data directory and tablespaces. When the cluster has
95 no additional tablespaces, the whole database will be placed in
96 the target directory. If the cluster contains additional
97 tablespaces, the main data directory will be placed in the
98 target directory, but all other tablespaces will be placed
99 in the same absolute path as they have on the source server.
100 (See <code class="option">--tablespace-mapping</code> to change that.)
102 This is the default format.
103 </p></dd><dt><span class="term"><code class="literal">t</code><br /></span><span class="term"><code class="literal">tar</code></span></dt><dd><p>
104 Write the output as tar files in the target directory. The main
105 data directory's contents will be written to a file named
106 <code class="filename">base.tar</code>, and each other tablespace will be
107 written to a separate tar file named after that tablespace's OID.
109 If the target directory is specified as <code class="literal">-</code>
110 (dash), the tar contents will be written to
111 standard output, suitable for piping to (for example)
112 <span class="productname">gzip</span>. This is only allowed if
113 the cluster has no additional tablespaces and WAL
114 streaming is not used.
115 </p></dd></dl></div></dd><dt><span class="term"><code class="option">-i <em class="replaceable"><code>old_manifest_file</code></em></code><br /></span><span class="term"><code class="option">--incremental=<em class="replaceable"><code>old_manifest_file</code></em></code></span></dt><dd><p>
116 Performs an <a class="link" href="continuous-archiving.html#BACKUP-INCREMENTAL-BACKUP" title="25.3.3. Making an Incremental Backup">incremental
117 backup</a>. The backup manifest for the reference
118 backup must be provided, and will be uploaded to the server, which will
119 respond by sending the requested incremental backup.
120 </p></dd><dt><span class="term"><code class="option">-R</code><br /></span><span class="term"><code class="option">--write-recovery-conf</code></span></dt><dd><p>
122 <a class="link" href="warm-standby.html#FILE-STANDBY-SIGNAL"><code class="filename">standby.signal</code></a>
123 <a id="id-1.9.4.10.6.2.1.4.3.1.2" class="indexterm"></a>
125 connection settings to the <code class="filename">postgresql.auto.conf</code>
126 file in the target directory (or within the base archive file when
127 using tar format). This eases setting up a standby server using the
128 results of the backup.
130 The <code class="filename">postgresql.auto.conf</code> file will record the connection
131 settings and, if specified, the replication slot
132 that <span class="application">pg_basebackup</span> is using, so that
133 streaming replication and <a class="link" href="logicaldecoding-explanation.html#LOGICALDECODING-REPLICATION-SLOTS-SYNCHRONIZATION" title="47.2.3. Replication Slot Synchronization">
134 logical replication slot synchronization</a> will use the same
135 settings later on. The dbname will be recorded only if the dbname was
136 specified explicitly in the connection string or <a class="link" href="libpq-envars.html" title="32.15. Environment Variables">
137 environment variable</a>.
138 </p></dd><dt><span class="term"><code class="option">-t <em class="replaceable"><code>target</code></em></code><br /></span><span class="term"><code class="option">--target=<em class="replaceable"><code>target</code></em></code></span></dt><dd><p>
139 Instructs the server where to place the base backup. The default target
140 is <code class="literal">client</code>, which specifies that the backup should
141 be sent to the machine where <span class="application">pg_basebackup</span>
142 is running. If the target is instead set to
143 <code class="literal">server:/some/path</code>, the backup will be stored on
144 the machine where the server is running in the
145 <code class="literal">/some/path</code> directory. Storing a backup on the
146 server requires superuser privileges or having privileges of the
147 <code class="literal">pg_write_server_files</code> role. If the target is set to
148 <code class="literal">blackhole</code>, the contents are discarded and not
149 stored anywhere. This should only be used for testing purposes, as you
150 will not end up with an actual backup.
152 Since WAL streaming is implemented by
153 <span class="application">pg_basebackup</span> rather than by the server,
154 this option cannot be used together with <code class="literal">-Xstream</code>.
155 Since that is the default, when this option is specified, you must also
156 specify either <code class="literal">-Xfetch</code> or <code class="literal">-Xnone</code>.
157 </p></dd><dt><span class="term"><code class="option">-T <em class="replaceable"><code>olddir</code></em>=<em class="replaceable"><code>newdir</code></em></code><br /></span><span class="term"><code class="option">--tablespace-mapping=<em class="replaceable"><code>olddir</code></em>=<em class="replaceable"><code>newdir</code></em></code></span></dt><dd><p>
158 Relocates the tablespace in directory <em class="replaceable"><code>olddir</code></em>
159 to <em class="replaceable"><code>newdir</code></em> during the backup. To be
160 effective, <em class="replaceable"><code>olddir</code></em> must exactly match the
161 path specification of the tablespace as it is defined on the source
162 server. (But it is not an error if there is no tablespace
163 in <em class="replaceable"><code>olddir</code></em> on the source server.)
164 Meanwhile <em class="replaceable"><code>newdir</code></em> is a directory in the
165 receiving host's filesystem. As with the main target directory,
166 <em class="replaceable"><code>newdir</code></em> need not exist already, but if
167 it does exist it must be empty.
168 Both <em class="replaceable"><code>olddir</code></em>
169 and <em class="replaceable"><code>newdir</code></em> must be absolute paths. If
170 either path needs to contain an equal sign (<code class="literal">=</code>),
171 precede that with a backslash. This option can be specified multiple
172 times for multiple tablespaces.
174 If a tablespace is relocated in this way, the symbolic links inside
175 the main data directory are updated to point to the new location. So
176 the new data directory is ready to be used for a new server instance
177 with all tablespaces in the updated locations.
179 Currently, this option only works with plain output format; it is
180 ignored if tar format is selected.
181 </p></dd><dt><span class="term"><code class="option">--waldir=<em class="replaceable"><code>waldir</code></em></code></span></dt><dd><p>
182 Sets the directory to write WAL (write-ahead log) files to.
183 By default WAL files will be placed in
184 the <code class="filename">pg_wal</code> subdirectory of the target
185 directory, but this option can be used to place them elsewhere.
186 <em class="replaceable"><code>waldir</code></em> must be an absolute path.
187 As with the main target directory,
188 <em class="replaceable"><code>waldir</code></em> need not exist already, but if
189 it does exist it must be empty.
190 This option can only be specified when
191 the backup is in plain format.
192 </p></dd><dt><span class="term"><code class="option">-X <em class="replaceable"><code>method</code></em></code><br /></span><span class="term"><code class="option">--wal-method=<em class="replaceable"><code>method</code></em></code></span></dt><dd><p>
193 Includes the required WAL (write-ahead log) files in the
194 backup. This will include all write-ahead logs generated during
195 the backup. Unless the method <code class="literal">none</code> is specified,
196 it is possible to start a postmaster in the target
197 directory without the need to consult the WAL archive, thus
198 making the output a completely standalone backup.
200 The following <em class="replaceable"><code>method</code></em>s for collecting the
201 write-ahead logs are supported:
203 </p><div class="variablelist"><dl class="variablelist"><dt><span class="term"><code class="literal">n</code><br /></span><span class="term"><code class="literal">none</code></span></dt><dd><p>
204 Don't include write-ahead logs in the backup.
205 </p></dd><dt><span class="term"><code class="literal">f</code><br /></span><span class="term"><code class="literal">fetch</code></span></dt><dd><p>
206 The write-ahead log files are collected at the end of the backup.
207 Therefore, it is necessary for the source server's
208 <a class="xref" href="runtime-config-replication.html#GUC-WAL-KEEP-SIZE">wal_keep_size</a> parameter to be set high
209 enough that the required log data is not removed before the end
210 of the backup. If the required log data has been recycled
211 before it's time to transfer it, the backup will fail and be
214 When tar format is used, the write-ahead log files will be
215 included in the <code class="filename">base.tar</code> file.
216 </p></dd><dt><span class="term"><code class="literal">s</code><br /></span><span class="term"><code class="literal">stream</code></span></dt><dd><p>
217 Stream write-ahead log data while the backup is being taken.
218 This method will open a second connection to the server and
219 start streaming the write-ahead log in parallel while running
220 the backup. Therefore, it will require two replication
221 connections not just one. As long as the client can keep up
222 with the write-ahead log data, using this method requires no
223 extra write-ahead logs to be saved on the source server.
225 When tar format is used, the write-ahead log files will be
226 written to a separate file named <code class="filename">pg_wal.tar</code>
227 (if the server is a version earlier than 10, the file will be named
228 <code class="filename">pg_xlog.tar</code>).
230 This value is the default.
231 </p></dd></dl></div></dd><dt><span class="term"><code class="option">-z</code><br /></span><span class="term"><code class="option">--gzip</code></span></dt><dd><p>
232 Enables gzip compression of tar file output, with the default
233 compression level. Compression is only available when using
234 the tar format, and the suffix <code class="filename">.gz</code> will
235 automatically be added to all tar filenames.
236 </p></dd><dt><span class="term"><code class="option">-Z <em class="replaceable"><code>level</code></em></code><br /></span><span class="term"><code class="option">-Z [{client|server}-]<em class="replaceable"><code>method</code></em>[:<em class="replaceable"><code>detail</code></em>]</code><br /></span><span class="term"><code class="option">--compress=<em class="replaceable"><code>level</code></em></code><br /></span><span class="term"><code class="option">--compress=[{client|server}-]<em class="replaceable"><code>method</code></em>[:<em class="replaceable"><code>detail</code></em>]</code></span></dt><dd><p>
237 Requests compression of the backup. If <code class="literal">client</code> or
238 <code class="literal">server</code> is included, it specifies where the
239 compression is to be performed. Compressing on the server will reduce
240 transfer bandwidth but will increase server CPU consumption. The
241 default is <code class="literal">client</code> except when
242 <code class="literal">--target</code> is used. In that case, the backup is not
243 being sent to the client, so only server compression is sensible.
244 When <code class="literal">-Xstream</code>, which is the default, is used,
245 server-side compression will not be applied to the WAL. To compress
246 the WAL, use client-side compression, or
247 specify <code class="literal">-Xfetch</code>.
249 The compression method can be set to <code class="literal">gzip</code>,
250 <code class="literal">lz4</code>, <code class="literal">zstd</code>,
251 <code class="literal">none</code> for no compression or an integer (no
252 compression if 0, <code class="literal">gzip</code> if greater than 0).
253 A compression detail string can optionally be specified.
254 If the detail string is an integer, it specifies the compression
255 level. Otherwise, it should be a comma-separated list of items,
257 <em class="replaceable"><code>keyword</code></em> or
258 <em class="replaceable"><code>keyword=value</code></em>.
259 Currently, the supported keywords are <code class="literal">level</code>,
260 <code class="literal">long</code>, and <code class="literal">workers</code>.
261 The detail string cannot be used when the compression method
262 is specified as a plain integer.
264 If no compression level is specified, the default compression level
265 will be used. If only a level is specified without mentioning an
266 algorithm, <code class="literal">gzip</code> compression will be used if the
267 level is greater than 0, and no compression will be used if the level
270 When the tar format is used with <code class="literal">gzip</code>,
271 <code class="literal">lz4</code>, or <code class="literal">zstd</code>, the suffix
272 <code class="filename">.gz</code>, <code class="filename">.lz4</code>, or
273 <code class="filename">.zst</code>, respectively, will be automatically added to
274 all tar filenames. When the plain format is used, client-side
275 compression may not be specified, but it is still possible to request
276 server-side compression. If this is done, the server will compress the
277 backup for transmission, and the client will decompress and extract it.
279 When this option is used in combination with
280 <code class="literal">-Xstream</code>, <code class="literal">pg_wal.tar</code> will
281 be compressed using <code class="literal">gzip</code> if client-side gzip
282 compression is selected, but will not be compressed if any other
283 compression algorithm is selected, or if server-side compression
285 </p></dd></dl></div><p>
287 The following command-line options control the generation of the
288 backup and the invocation of the program:
290 </p><div class="variablelist"><dl class="variablelist"><dt><span class="term"><code class="option">-c {fast|spread}</code><br /></span><span class="term"><code class="option">--checkpoint={fast|spread}</code></span></dt><dd><p>
291 Sets checkpoint mode to fast (immediate) or spread (the default)
292 (see <a class="xref" href="continuous-archiving.html#BACKUP-LOWLEVEL-BASE-BACKUP" title="25.3.4. Making a Base Backup Using the Low Level API">Section 25.3.4</a>).
293 </p></dd><dt><span class="term"><code class="option">-C</code><br /></span><span class="term"><code class="option">--create-slot</code></span></dt><dd><p>
294 Specifies that the replication slot named by the
295 <code class="literal">--slot</code> option should be created before starting
296 the backup. An error is raised if the slot already exists.
297 </p></dd><dt><span class="term"><code class="option">-l <em class="replaceable"><code>label</code></em></code><br /></span><span class="term"><code class="option">--label=<em class="replaceable"><code>label</code></em></code></span></dt><dd><p>
298 Sets the label for the backup. If none is specified, a default value of
299 <span class="quote">“<span class="quote"><code class="literal">pg_basebackup base backup</code></span>”</span> will be used.
300 </p></dd><dt><span class="term"><code class="option">-n</code><br /></span><span class="term"><code class="option">--no-clean</code></span></dt><dd><p>
301 By default, when <code class="command">pg_basebackup</code> aborts with an
302 error, it removes any directories it might have created before
303 discovering that it cannot finish the job (for example, the target
304 directory and write-ahead log directory). This option inhibits
305 tidying-up and is thus useful for debugging.
307 Note that tablespace directories are not cleaned up either way.
308 </p></dd><dt><span class="term"><code class="option">-N</code><br /></span><span class="term"><code class="option">--no-sync</code></span></dt><dd><p>
309 By default, <code class="command">pg_basebackup</code> will wait for all files
310 to be written safely to disk. This option causes
311 <code class="command">pg_basebackup</code> to return without waiting, which is
312 faster, but means that a subsequent operating system crash can leave
313 the base backup corrupt. Generally, this option is useful for testing
314 but should not be used when creating a production installation.
315 </p></dd><dt><span class="term"><code class="option">-P</code><br /></span><span class="term"><code class="option">--progress</code></span></dt><dd><p>
316 Enables progress reporting. Turning this on will deliver an approximate
317 progress report during the backup. Since the database may change during
318 the backup, this is only an approximation and may not end at exactly
319 <code class="literal">100%</code>. In particular, when WAL log is included in the
320 backup, the total amount of data cannot be estimated in advance, and
321 in this case the estimated target size will increase once it passes the
322 total estimate without WAL.
323 </p></dd><dt><span class="term"><code class="option">-r <em class="replaceable"><code>rate</code></em></code><br /></span><span class="term"><code class="option">--max-rate=<em class="replaceable"><code>rate</code></em></code></span></dt><dd><p>
324 Sets the maximum transfer rate at which data is collected from the
325 source server. This can be useful to limit the impact
326 of <span class="application">pg_basebackup</span> on the server. Values
327 are in kilobytes per second. Use a suffix of <code class="literal">M</code>
328 to indicate megabytes per second. A suffix of <code class="literal">k</code>
329 is also accepted, and has no effect. Valid values are between 32
330 kilobytes per second and 1024 megabytes per second.
332 This option always affects transfer of the data directory. Transfer of
333 WAL files is only affected if the collection method
334 is <code class="literal">fetch</code>.
335 </p></dd><dt><span class="term"><code class="option">-S <em class="replaceable"><code>slotname</code></em></code><br /></span><span class="term"><code class="option">--slot=<em class="replaceable"><code>slotname</code></em></code></span></dt><dd><p>
336 This option can only be used together with <code class="literal">-X
337 stream</code>. It causes WAL streaming to use the specified
338 replication slot. If the base backup is intended to be used as a
339 streaming-replication standby using a replication slot, the standby
340 should then use the same replication slot name as
341 <a class="xref" href="runtime-config-replication.html#GUC-PRIMARY-SLOT-NAME">primary_slot_name</a>. This ensures that the
342 primary server does not remove any necessary WAL data in the time
343 between the end of the base backup and the start of streaming
344 replication on the new standby.
346 The specified replication slot has to exist unless the
347 option <code class="option">-C</code> is also used.
349 If this option is not specified and the server supports temporary
350 replication slots (version 10 and later), then a temporary replication
351 slot is automatically used for WAL streaming.
352 </p></dd><dt><span class="term"><code class="option">--sync-method=<em class="replaceable"><code>method</code></em></code></span></dt><dd><p>
353 When set to <code class="literal">fsync</code>, which is the default,
354 <code class="command">pg_basebackup</code> will recursively open and synchronize
355 all files in the backup directory. When the plain format is used, the
356 search for files will follow symbolic links for the WAL directory and
357 each configured tablespace.
359 On Linux, <code class="literal">syncfs</code> may be used instead to ask the
360 operating system to synchronize the whole file system that contains the
361 backup directory. When the plain format is used,
362 <code class="command">pg_basebackup</code> will also synchronize the file systems
363 that contain the WAL files and each tablespace. See
364 <a class="xref" href="runtime-config-error-handling.html#GUC-RECOVERY-INIT-SYNC-METHOD">recovery_init_sync_method</a> for information about
365 the caveats to be aware of when using <code class="literal">syncfs</code>.
367 This option has no effect when <code class="option">--no-sync</code> is used.
368 </p></dd><dt><span class="term"><code class="option">-v</code><br /></span><span class="term"><code class="option">--verbose</code></span></dt><dd><p>
369 Enables verbose mode. Will output some extra steps during startup and
370 shutdown, as well as show the exact file name that is currently being
371 processed if progress reporting is also enabled.
372 </p></dd><dt><span class="term"><code class="option">--manifest-checksums=<em class="replaceable"><code>algorithm</code></em></code></span></dt><dd><p>
373 Specifies the checksum algorithm that should be applied to each file
374 included in the backup manifest. Currently, the available
375 algorithms are <code class="literal">NONE</code>, <code class="literal">CRC32C</code>,
376 <code class="literal">SHA224</code>, <code class="literal">SHA256</code>,
377 <code class="literal">SHA384</code>, and <code class="literal">SHA512</code>.
378 The default is <code class="literal">CRC32C</code>.
380 If <code class="literal">NONE</code> is selected, the backup manifest will
381 not contain any checksums. Otherwise, it will contain a checksum
382 of each file in the backup using the specified algorithm. In addition,
383 the manifest will always contain a <code class="literal">SHA256</code>
384 checksum of its own contents. The <code class="literal">SHA</code> algorithms
385 are significantly more CPU-intensive than <code class="literal">CRC32C</code>,
386 so selecting one of them may increase the time required to complete
389 Using a SHA hash function provides a cryptographically secure digest
390 of each file for users who wish to verify that the backup has not been
391 tampered with, while the CRC-32C algorithm provides a checksum that is
392 much faster to calculate; it is good at catching errors due to accidental
393 changes but is not resistant to malicious modifications. Note that, to
394 be useful against an adversary who has access to the backup, the backup
395 manifest would need to be stored securely elsewhere or otherwise
396 verified not to have been modified since the backup was taken.
398 <a class="xref" href="app-pgverifybackup.html" title="pg_verifybackup"><span class="refentrytitle"><span class="application">pg_verifybackup</span></span></a> can be used to check the
399 integrity of a backup against the backup manifest.
400 </p></dd><dt><span class="term"><code class="option">--manifest-force-encode</code></span></dt><dd><p>
401 Forces all filenames in the backup manifest to be hex-encoded.
402 If this option is not specified, only non-UTF8 filenames are
403 hex-encoded. This option is mostly intended to test that tools which
404 read a backup manifest file properly handle this case.
405 </p></dd><dt><span class="term"><code class="option">--no-estimate-size</code></span></dt><dd><p>
406 Prevents the server from estimating the total
407 amount of backup data that will be streamed, resulting in the
408 <code class="structfield">backup_total</code> column in the
409 <code class="structname">pg_stat_progress_basebackup</code> view
410 always being <code class="literal">NULL</code>.
412 Without this option, the backup will start by enumerating
413 the size of the entire database, and then go back and send
414 the actual contents. This may make the backup take slightly
415 longer, and in particular it will take longer before the first
416 data is sent. This option is useful to avoid such estimation
417 time if it's too long.
419 This option is not allowed when using <code class="option">--progress</code>.
420 </p></dd><dt><span class="term"><code class="option">--no-manifest</code></span></dt><dd><p>
421 Disables generation of a backup manifest. If this option is not
422 specified, the server will generate and send a backup manifest
423 which can be verified using <a class="xref" href="app-pgverifybackup.html" title="pg_verifybackup"><span class="refentrytitle"><span class="application">pg_verifybackup</span></span></a>.
424 The manifest is a list of every file present in the backup with the
425 exception of any WAL files that may be included. It also stores the
426 size, last modification time, and an optional checksum for each file.
427 </p></dd><dt><span class="term"><code class="option">--no-slot</code></span></dt><dd><p>
428 Prevents the creation of a temporary replication slot
431 By default, if log streaming is selected but no slot name is given
432 with the <code class="option">-S</code> option, then a temporary replication
433 slot is created (if supported by the source server).
435 The main purpose of this option is to allow taking a base backup when
436 the server has no free replication slots. Using a replication slot
437 is almost always preferred, because it prevents needed WAL from being
438 removed by the server during the backup.
439 </p></dd><dt><span class="term"><code class="option">--no-verify-checksums</code></span></dt><dd><p>
440 Disables verification of checksums, if they are enabled on the server
441 the base backup is taken from.
443 By default, checksums are verified and checksum failures will result
444 in a non-zero exit status. However, the base backup will not be
445 removed in such a case, as if the <code class="option">--no-clean</code> option
446 had been used. Checksum verification failures will also be reported
447 in the <a class="link" href="monitoring-stats.html#MONITORING-PG-STAT-DATABASE-VIEW" title="27.2.17. pg_stat_database">
448 <code class="structname">pg_stat_database</code></a> view.
449 </p></dd></dl></div><p>
451 The following command-line options control the connection to the source
454 </p><div class="variablelist"><dl class="variablelist"><dt><span class="term"><code class="option">-d <em class="replaceable"><code>connstr</code></em></code><br /></span><span class="term"><code class="option">--dbname=<em class="replaceable"><code>connstr</code></em></code></span></dt><dd><p>
455 Specifies parameters used to connect to the server, as a <a class="link" href="libpq-connect.html#LIBPQ-CONNSTRING" title="32.1.1. Connection Strings">connection string</a>; these
456 will override any conflicting command line options.
458 This option is called <code class="literal">--dbname</code> for consistency with other
459 client applications, but because <span class="application">pg_basebackup</span>
460 doesn't connect to any particular database in the cluster, any database
461 name included in the connection string will be ignored by the server.
462 However, a database name supplied that way overrides the default
463 database name (<code class="literal">replication</code>) for purposes of
464 looking up the replication connection's password
465 in <code class="filename">~/.pgpass</code>. Similarly, middleware or proxies
466 used in connecting to <span class="productname">PostgreSQL</span> might
467 utilize the name for purposes such as connection routing. The
468 database name can also be used
469 by <a class="link" href="logicaldecoding-explanation.html#LOGICALDECODING-REPLICATION-SLOTS-SYNCHRONIZATION" title="47.2.3. Replication Slot Synchronization">
470 logical replication slot synchronization</a>.
471 </p></dd><dt><span class="term"><code class="option">-h <em class="replaceable"><code>host</code></em></code><br /></span><span class="term"><code class="option">--host=<em class="replaceable"><code>host</code></em></code></span></dt><dd><p>
472 Specifies the host name of the machine on which the server is
473 running. If the value begins with a slash, it is used as the
474 directory for a Unix domain socket. The default is taken
475 from the <code class="envar">PGHOST</code> environment variable, if set,
476 else a Unix domain socket connection is attempted.
477 </p></dd><dt><span class="term"><code class="option">-p <em class="replaceable"><code>port</code></em></code><br /></span><span class="term"><code class="option">--port=<em class="replaceable"><code>port</code></em></code></span></dt><dd><p>
478 Specifies the TCP port or local Unix domain socket file
479 extension on which the server is listening for connections.
480 Defaults to the <code class="envar">PGPORT</code> environment variable, if
481 set, or a compiled-in default.
482 </p></dd><dt><span class="term"><code class="option">-s <em class="replaceable"><code>interval</code></em></code><br /></span><span class="term"><code class="option">--status-interval=<em class="replaceable"><code>interval</code></em></code></span></dt><dd><p>
483 Specifies the number of seconds between status packets sent back to
484 the source server. Smaller values allow more accurate monitoring of
485 backup progress from the server.
486 A value of zero disables periodic status updates completely,
487 although an update will still be sent when requested by the server, to
488 avoid timeout-based disconnects. The default value is 10 seconds.
489 </p></dd><dt><span class="term"><code class="option">-U <em class="replaceable"><code>username</code></em></code><br /></span><span class="term"><code class="option">--username=<em class="replaceable"><code>username</code></em></code></span></dt><dd><p>
490 Specifies the user name to connect as.
491 </p></dd><dt><span class="term"><code class="option">-w</code><br /></span><span class="term"><code class="option">--no-password</code></span></dt><dd><p>
492 Prevents issuing a password prompt. If the server requires
493 password authentication and a password is not available by
494 other means such as a <code class="filename">.pgpass</code> file, the
495 connection attempt will fail. This option can be useful in
496 batch jobs and scripts where no user is present to enter a
498 </p></dd><dt><span class="term"><code class="option">-W</code><br /></span><span class="term"><code class="option">--password</code></span></dt><dd><p>
499 Forces <span class="application">pg_basebackup</span> to prompt for a
500 password before connecting to the source server.
502 This option is never essential, since
503 <span class="application">pg_basebackup</span> will automatically prompt
504 for a password if the server demands password authentication.
505 However, <span class="application">pg_basebackup</span> will waste a
506 connection attempt finding out that the server wants a password.
507 In some cases it is worth typing <code class="option">-W</code> to avoid the extra
509 </p></dd></dl></div><p>
511 Other options are also available:
513 </p><div class="variablelist"><dl class="variablelist"><dt><span class="term"><code class="option">-V</code><br /></span><span class="term"><code class="option">--version</code></span></dt><dd><p>
514 Prints the <span class="application">pg_basebackup</span> version and exits.
515 </p></dd><dt><span class="term"><code class="option">-?</code><br /></span><span class="term"><code class="option">--help</code></span></dt><dd><p>
516 Shows help about <span class="application">pg_basebackup</span> command line
517 arguments, and exits.
518 </p></dd></dl></div><p>
519 </p></div><div class="refsect1" id="id-1.9.4.10.7"><h2>Environment</h2><p>
520 This utility, like most other <span class="productname">PostgreSQL</span> utilities,
521 uses the environment variables supported by <span class="application">libpq</span>
522 (see <a class="xref" href="libpq-envars.html" title="32.15. Environment Variables">Section 32.15</a>).
524 The environment variable <code class="envar">PG_COLOR</code> specifies whether to use
525 color in diagnostic messages. Possible values are
526 <code class="literal">always</code>, <code class="literal">auto</code> and
527 <code class="literal">never</code>.
528 </p></div><div class="refsect1" id="id-1.9.4.10.8"><h2>Notes</h2><p>
529 At the beginning of the backup, a checkpoint needs to be performed on the
530 source server. This can take some time (especially if the option
531 <code class="literal">--checkpoint=fast</code> is not used), during
532 which <span class="application">pg_basebackup</span> will appear to be idle.
534 The backup will include all files in the data directory and tablespaces,
535 including the configuration files and any additional files placed in the
536 directory by third parties, except certain temporary files managed by
537 PostgreSQL and operating system files. But only regular files and
538 directories are copied, except that
539 symbolic links used for tablespaces are preserved. Symbolic links pointing
540 to certain directories known to PostgreSQL are copied as empty directories.
541 Other symbolic links and special device files are skipped.
542 See <a class="xref" href="protocol-replication.html" title="54.4. Streaming Replication Protocol">Section 54.4</a> for the precise details.
544 In plain format, tablespaces will be backed up to the same path
545 they have on the source server, unless the
546 option <code class="literal">--tablespace-mapping</code> is used. Without
547 this option, running a plain format base backup on the same host as the
548 server will not work if tablespaces are in use, because the backup would
549 have to be written to the same directory locations as the original
552 When tar format is used, it is the user's responsibility to unpack each
553 tar file before starting a PostgreSQL server that uses the data. If there
554 are additional tablespaces, the
555 tar files for them need to be unpacked in the correct locations. In this
556 case the symbolic links for those tablespaces will be created by the server
557 according to the contents of the <code class="filename">tablespace_map</code> file that is
558 included in the <code class="filename">base.tar</code> file.
560 <span class="application">pg_basebackup</span> works with servers of the same
561 or older major version, down to 9.1. However, WAL streaming mode (<code class="literal">-X
562 stream</code>) only works with server version 9.3 and later, the tar format
563 (<code class="literal">--format=tar</code>) only works with server version 9.5
564 and later, and incremental backup (<code class="literal">--incremental</code>) only works
565 with server version 17 and later.
567 <span class="application">pg_basebackup</span> will preserve group permissions
568 for data files if group permissions are enabled on the source cluster.
569 </p></div><div class="refsect1" id="id-1.9.4.10.9"><h2>Examples</h2><p>
570 To create a base backup of the server at <code class="literal">mydbserver</code>
571 and store it in the local directory
572 <code class="filename">/usr/local/pgsql/data</code>:
573 </p><pre class="screen">
574 <code class="prompt">$</code> <strong class="userinput"><code>pg_basebackup -h mydbserver -D /usr/local/pgsql/data</code></strong>
577 To create a backup of the local server with one compressed
578 tar file for each tablespace, and store it in the directory
579 <code class="filename">backup</code>, showing a progress report while running:
580 </p><pre class="screen">
581 <code class="prompt">$</code> <strong class="userinput"><code>pg_basebackup -D backup -Ft -z -P</code></strong>
584 To create a backup of a single-tablespace local database and compress
585 this with <span class="productname">bzip2</span>:
586 </p><pre class="screen">
587 <code class="prompt">$</code> <strong class="userinput"><code>pg_basebackup -D - -Ft -X fetch | bzip2 > backup.tar.bz2</code></strong>
589 (This command will fail if there are multiple tablespaces in the
592 To create a backup of a local database where the tablespace in
593 <code class="filename">/opt/ts</code> is relocated
594 to <code class="filename">./backup/ts</code>:
595 </p><pre class="screen">
596 <code class="prompt">$</code> <strong class="userinput"><code>pg_basebackup -D backup/data -T /opt/ts=$(pwd)/backup/ts</code></strong>
598 To create a backup of the local server with one tar file for each tablespace
599 compressed with <span class="application">gzip</span> at level 9, stored in the
600 directory <code class="filename">backup</code>:
601 </p><pre class="screen">
602 <code class="prompt">$</code> <strong class="userinput"><code>pg_basebackup -D backup -Ft --compress=gzip:9</code></strong>
603 </pre></div><div class="refsect1" id="id-1.9.4.10.10"><h2>See Also</h2><span class="simplelist"><a class="xref" href="app-pgdump.html" title="pg_dump"><span class="refentrytitle"><span class="application">pg_dump</span></span></a>, <a class="xref" href="progress-reporting.html#BASEBACKUP-PROGRESS-REPORTING" title="27.4.6. Base Backup Progress Reporting">Section 27.4.6</a></span></div></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="app-pgamcheck.html" title="pg_amcheck">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="reference-client.html" title="PostgreSQL Client Applications">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="pgbench.html" title="pgbench">Next</a></td></tr><tr><td width="40%" align="left" valign="top"><span class="application">pg_amcheck</span> </td><td width="20%" align="center"><a accesskey="h" href="index.html" title="PostgreSQL 18.0 Documentation">Home</a></td><td width="40%" align="right" valign="top"> <span class="application">pgbench</span></td></tr></table></div></body></html>
begriffs.com / projects / ai-pg / blob