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>25.2. File System Level Backup</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="backup-dump.html" title="25.1. SQL Dump" /><link rel="next" href="continuous-archiving.html" title="25.3. Continuous Archiving and Point-in-Time Recovery (PITR)" /></head><body id="docContent" class="container-fluid col-10"><div class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="5" align="center">25.2. File System Level Backup</th></tr><tr><td width="10%" align="left"><a accesskey="p" href="backup-dump.html" title="25.1. SQL Dump">Prev</a> </td><td width="10%" align="left"><a accesskey="u" href="backup.html" title="Chapter 25. Backup and Restore">Up</a></td><th width="60%" align="center">Chapter 25. Backup and Restore</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="continuous-archiving.html" title="25.3. Continuous Archiving and Point-in-Time Recovery (PITR)">Next</a></td></tr></table><hr /></div><div class="sect1" id="BACKUP-FILE"><div class="titlepage"><div><div><h2 class="title" style="clear: both">25.2. File System Level Backup <a href="#BACKUP-FILE" class="id_link">#</a></h2></div></div></div><p>
3 An alternative backup strategy is to directly copy the files that
4 <span class="productname">PostgreSQL</span> uses to store the data in the database;
5 <a class="xref" href="creating-cluster.html" title="18.2. Creating a Database Cluster">Section 18.2</a> explains where these files
6 are located. You can use whatever method you prefer
7 for doing file system backups; for example:
9 </p><pre class="programlisting">
10 tar -cf backup.tar /usr/local/pgsql/data
13 There are two restrictions, however, which make this method
14 impractical, or at least inferior to the <span class="application">pg_dump</span>
17 </p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><p>
18 The database server <span class="emphasis"><em>must</em></span> be shut down in order to
19 get a usable backup. Half-way measures such as disallowing all
20 connections will <span class="emphasis"><em>not</em></span> work
21 (in part because <code class="command">tar</code> and similar tools do not take
22 an atomic snapshot of the state of the file system,
23 but also because of internal buffering within the server).
24 Information about stopping the server can be found in
25 <a class="xref" href="server-shutdown.html" title="18.5. Shutting Down the Server">Section 18.5</a>. Needless to say, you
26 also need to shut down the server before restoring the data.
27 </p></li><li class="listitem"><p>
28 If you have dug into the details of the file system layout of the
29 database, you might be tempted to try to back up or restore only certain
30 individual tables or databases from their respective files or
31 directories. This will <span class="emphasis"><em>not</em></span> work because the
32 information contained in these files is not usable without
34 <code class="filename">pg_xact/*</code>, which contain the commit status of
35 all transactions. A table file is only usable with this
36 information. Of course it is also impossible to restore only a
37 table and the associated <code class="filename">pg_xact</code> data
38 because that would render all other tables in the database
39 cluster useless. So file system backups only work for complete
40 backup and restoration of an entire database cluster.
41 </p></li></ol></div><p>
43 An alternative file-system backup approach is to make a
44 <span class="quote">“<span class="quote">consistent snapshot</span>”</span> of the data directory, if the
45 file system supports that functionality (and you are willing to
46 trust that it is implemented correctly). The typical procedure is
47 to make a <span class="quote">“<span class="quote">frozen snapshot</span>”</span> of the volume containing the
48 database, then copy the whole data directory (not just parts, see
49 above) from the snapshot to a backup device, then release the frozen
50 snapshot. This will work even while the database server is running.
51 However, a backup created in this way saves
52 the database files in a state as if the database server was not
53 properly shut down; therefore, when you start the database server
54 on the backed-up data, it will think the previous server instance
55 crashed and will replay the WAL log. This is not a problem; just
56 be aware of it (and be sure to include the WAL files in your backup).
57 You can perform a <code class="command">CHECKPOINT</code> before taking the
58 snapshot to reduce recovery time.
60 If your database is spread across multiple file systems, there might not
61 be any way to obtain exactly-simultaneous frozen snapshots of all
62 the volumes. For example, if your data files and WAL log are on different
63 disks, or if tablespaces are on different file systems, it might
64 not be possible to use snapshot backup because the snapshots
65 <span class="emphasis"><em>must</em></span> be simultaneous.
66 Read your file system documentation very carefully before trusting
67 the consistent-snapshot technique in such situations.
69 If simultaneous snapshots are not possible, one option is to shut down
70 the database server long enough to establish all the frozen snapshots.
71 Another option is to perform a continuous archiving base backup (<a class="xref" href="continuous-archiving.html#BACKUP-BASE-BACKUP" title="25.3.2. Making a Base Backup">Section 25.3.2</a>) because such backups are immune to file
72 system changes during the backup. This requires enabling continuous
73 archiving just during the backup process; restore is done using
74 continuous archive recovery (<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>).
76 Another option is to use <span class="application">rsync</span> to perform a file
77 system backup. This is done by first running <span class="application">rsync</span>
78 while the database server is running, then shutting down the database
79 server long enough to do an <code class="command">rsync --checksum</code>.
80 (<code class="option">--checksum</code> is necessary because <code class="command">rsync</code> only
81 has file modification-time granularity of one second.) The
82 second <span class="application">rsync</span> will be quicker than the first,
83 because it has relatively little data to transfer, and the end result
84 will be consistent because the server was down. This method
85 allows a file system backup to be performed with minimal downtime.
87 Note that a file system backup will typically be larger
88 than an SQL dump. (<span class="application">pg_dump</span> does not need to dump
89 the contents of indexes for example, just the commands to recreate
90 them.) However, taking a file system backup might be faster.
91 </p></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="backup-dump.html" title="25.1. SQL Dump">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="backup.html" title="Chapter 25. Backup and Restore">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="continuous-archiving.html" title="25.3. Continuous Archiving and Point-in-Time Recovery (PITR)">Next</a></td></tr><tr><td width="40%" align="left" valign="top">25.1. <acronym class="acronym">SQL</acronym> Dump </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"> 25.3. Continuous Archiving and Point-in-Time Recovery (PITR)</td></tr></table></div></body></html>