1PG_STANDBY(1) PostgreSQL 9.2.24 Documentation PG_STANDBY(1)
2
6 pg_standby - supports the creation of a PostgreSQL warm standby server
7
9 pg_standby [option...] archivelocation nextwalfile xlogfilepath
10 [restartwalfile]
11
13 pg_standby supports creation of a “warm standby” database server. It is
14 designed to be a production-ready program, as well as a customizable
15 template should you require specific modifications.
16 pg_standby is designed to be a waiting restore_command, which is needed
18 to turn a standard archive recovery into a warm standby operation.
19 Other configuration is required as well, all of which is described in
20 the main server manual (see Section 25.2, “Log-Shipping Standby
21 Servers”, in the documentation).
22 To configure a standby server to use pg_standby, put this into its
24 recovery.conf configuration file:
25 restore_command = 'pg_standby archiveDir %f %p %r'
27 where archiveDir is the directory from which WAL segment files should
29 be restored.
30 If restartwalfile is specified, normally by using the %r macro, then
32 all WAL files logically preceding this file will be removed from
33 archivelocation. This minimizes the number of files that need to be
34 retained, while preserving crash-restart capability. Use of this
35 parameter is appropriate if the archivelocation is a transient staging
36 area for this particular standby server, but not when the
37 archivelocation is intended as a long-term WAL archive area.
38 pg_standby assumes that archivelocation is a directory readable by the
40 server-owning user. If restartwalfile (or -k) is specified, the
41 archivelocation directory must be writable too.
42 There are two ways to fail over to a “warm standby” database server
44 when the master server fails:
45 Smart Failover
47 In smart failover, the server is brought up after applying all WAL
48 files available in the archive. This results in zero data loss,
49 even if the standby server has fallen behind, but if there is a lot
50 of unapplied WAL it can be a long time before the standby server
51 becomes ready. To trigger a smart failover, create a trigger file
52 containing the word smart, or just create it and leave it empty.
53 Fast Failover
55 In fast failover, the server is brought up immediately. Any WAL
56 files in the archive that have not yet been applied will be
57 ignored, and all transactions in those files are lost. To trigger a
58 fast failover, create a trigger file and write the word fast into
59 it. pg_standby can also be configured to execute a fast failover
60 automatically if no new WAL file appears within a defined interval.
61
63 pg_standby accepts the following command-line arguments:
64 -c
66 Use cp or copy command to restore WAL files from archive. This is
67 the only supported behavior so this option is useless.
68 -d
70 Print lots of debug logging output on stderr.
71 -k
73 Remove files from archivelocation so that no more than this many
74 WAL files before the current one are kept in the archive. Zero (the
75 default) means not to remove any files from archivelocation. This
76 parameter will be silently ignored if restartwalfile is specified,
77 since that specification method is more accurate in determining the
78 correct archive cut-off point. Use of this parameter is deprecated
79 as of PostgreSQL 8.3; it is safer and more efficient to specify a
80 restartwalfile parameter. A too small setting could result in
81 removal of files that are still needed for a restart of the standby
82 server, while a too large setting wastes archive space.
83 -r maxretries
85 Set the maximum number of times to retry the copy command if it
86 fails (default 3). After each failure, we wait for sleeptime *
87 num_retries so that the wait time increases progressively. So by
88 default, we will wait 5 secs, 10 secs, then 15 secs before
89 reporting the failure back to the standby server. This will be
90 interpreted as end of recovery and the standby will come up fully
91 as a result.
92 -s sleeptime
94 Set the number of seconds (up to 60, default 5) to sleep between
95 tests to see if the WAL file to be restored is available in the
96 archive yet. The default setting is not necessarily recommended;
97 consult Section 25.2, “Log-Shipping Standby Servers”, in the
98 documentation for discussion.
99 -t triggerfile
101 Specify a trigger file whose presence should cause failover. It is
102 recommended that you use a structured file name to avoid confusion
103 as to which server is being triggered when multiple servers exist
104 on the same system; for example /tmp/pgsql.trigger.5432.
105 -V, --version
107 Print the pg_standby version and exit.
108 -w maxwaittime
110 Set the maximum number of seconds to wait for the next WAL file,
111 after which a fast failover will be performed. A setting of zero
112 (the default) means wait forever. The default setting is not
113 necessarily recommended; consult Section 25.2, “Log-Shipping
114 Standby Servers”, in the documentation for discussion.
115 -?, --help
117 Show help about pg_standby command line arguments, and exit.
118
120 pg_standby is designed to work with PostgreSQL 8.2 and later.
121 PostgreSQL 8.3 provides the %r macro, which is designed to let
123 pg_standby know the last file it needs to keep. With PostgreSQL 8.2,
124 the -k option must be used if archive cleanup is required. This option
125 remains available in 8.3, but its use is deprecated.
126 PostgreSQL 8.4 provides the recovery_end_command option. Without this
128 option a leftover trigger file can be hazardous.
129 pg_standby is written in C and has an easy-to-modify source code, with
131 specifically designated sections to modify for your own needs
132
134 On Linux or Unix systems, you might use:
135 archive_command = 'cp %p .../archive/%f'
137 restore_command = 'pg_standby -d -s 2 -t /tmp/pgsql.trigger.5442 .../archive %f %p %r 2>>standby.log'
139 recovery_end_command = 'rm -f /tmp/pgsql.trigger.5442'
141 where the archive directory is physically located on the standby
143 server, so that the archive_command is accessing it across NFS, but the
144 files are local to the standby (enabling use of ln). This will:
145 · produce debugging output in standby.log
147 · sleep for 2 seconds between checks for next WAL file availability
149 · stop waiting only when a trigger file called
151 /tmp/pgsql.trigger.5442 appears, and perform failover according to
152 its content
153 · remove the trigger file when recovery ends
155 · remove no-longer-needed files from the archive directory
157 On Windows, you might use:
159 archive_command = 'copy %p ...\\archive\\%f'
161 restore_command = 'pg_standby -d -s 5 -t C:\pgsql.trigger.5442 ...\archive %f %p %r 2>>standby.log'
163 recovery_end_command = 'del C:\pgsql.trigger.5442'
165 Note that backslashes need to be doubled in the archive_command, but
167 not in the restore_command or recovery_end_command. This will:
168 · use the copy command to restore WAL files from archive
170 · produce debugging output in standby.log
172 · sleep for 5 seconds between checks for next WAL file availability
174 · stop waiting only when a trigger file called C:\pgsql.trigger.5442
176 appears, and perform failover according to its content
177 · remove the trigger file when recovery ends
179 · remove no-longer-needed files from the archive directory
181 The copy command on Windows sets the final file size before the file is
183 completely copied, which would ordinarily confuse pg_standby. Therefore
184 pg_standby waits sleeptime seconds once it sees the proper file size.
185 GNUWin32's cp sets the file size only after the file copy is complete.
186 Since the Windows example uses copy at both ends, either or both
188 servers might be accessing the archive directory across the network.
189
191 Simon Riggs <simon@2ndquadrant.com>
192
194 pg_archivecleanup(1)
195PostgreSQL 9.2.24 2017-11-06 PG_STANDBY(1)