1PG_UPGRADE(1) PostgreSQL 14.3 Documentation PG_UPGRADE(1)
2
6 pg_upgrade - upgrade a PostgreSQL server instance
7
9 pg_upgrade -b oldbindir [-B newbindir] -d oldconfigdir -D newconfigdir
10 [option...]
11
13 pg_upgrade (formerly called pg_migrator) allows data stored in
14 PostgreSQL data files to be upgraded to a later PostgreSQL major
15 version without the data dump/reload typically required for major
16 version upgrades, e.g., from 9.5.8 to 9.6.4 or from 10.7 to 11.2. It is
17 not required for minor version upgrades, e.g., from 9.6.2 to 9.6.3 or
18 from 10.1 to 10.2.
19 Major PostgreSQL releases regularly add new features that often change
21 the layout of the system tables, but the internal data storage format
22 rarely changes. pg_upgrade uses this fact to perform rapid upgrades by
23 creating new system tables and simply reusing the old user data files.
24 If a future major release ever changes the data storage format in a way
25 that makes the old data format unreadable, pg_upgrade will not be
26 usable for such upgrades. (The community will attempt to avoid such
27 situations.)
28 pg_upgrade does its best to make sure the old and new clusters are
30 binary-compatible, e.g., by checking for compatible compile-time
31 settings, including 32/64-bit binaries. It is important that any
32 external modules are also binary compatible, though this cannot be
33 checked by pg_upgrade.
34 pg_upgrade supports upgrades from 8.4.X and later to the current major
36 release of PostgreSQL, including snapshot and beta releases.
37
39 pg_upgrade accepts the following command-line arguments:
40 -b bindir
42 --old-bindir=bindir
43 the old PostgreSQL executable directory; environment variable
44 PGBINOLD
45 -B bindir
47 --new-bindir=bindir
48 the new PostgreSQL executable directory; default is the directory
49 where pg_upgrade resides; environment variable PGBINNEW
50 -c
52 --check
53 check clusters only, don't change any data
54 -d configdir
56 --old-datadir=configdir
57 the old database cluster configuration directory; environment
58 variable PGDATAOLD
59 -D configdir
61 --new-datadir=configdir
62 the new database cluster configuration directory; environment
63 variable PGDATANEW
64 -j njobs
66 --jobs=njobs
67 number of simultaneous processes or threads to use
68 -k
70 --link
71 use hard links instead of copying files to the new cluster
72 -o options
74 --old-options options
75 options to be passed directly to the old postgres command; multiple
76 option invocations are appended
77 -O options
79 --new-options options
80 options to be passed directly to the new postgres command; multiple
81 option invocations are appended
82 -p port
84 --old-port=port
85 the old cluster port number; environment variable PGPORTOLD
86 -P port
88 --new-port=port
89 the new cluster port number; environment variable PGPORTNEW
90 -r
92 --retain
93 retain SQL and log files even after successful completion
94 -s dir
96 --socketdir=dir
97 directory to use for postmaster sockets during upgrade; default is
98 current working directory; environment variable PGSOCKETDIR
99 -U username
101 --username=username
102 cluster's install user name; environment variable PGUSER
103 -v
105 --verbose
106 enable verbose internal logging
107 -V
109 --version
110 display version information, then exit
111 --clone
113 Use efficient file cloning (also known as “reflinks” on some
114 systems) instead of copying files to the new cluster. This can
115 result in near-instantaneous copying of the data files, giving the
116 speed advantages of -k/--link while leaving the old cluster
117 untouched.
118 File cloning is only supported on some operating systems and file
120 systems. If it is selected but not supported, the pg_upgrade run
121 will error. At present, it is supported on Linux (kernel 4.5 or
122 later) with Btrfs and XFS (on file systems created with reflink
123 support), and on macOS with APFS.
124 -?
126 --help
127 show help, then exit
128
130 These are the steps to perform an upgrade with pg_upgrade:
131 1. Optionally move the old cluster: If you are using a
133 version-specific installation directory, e.g., /opt/PostgreSQL/14,
134 you do not need to move the old cluster. The graphical installers
135 all use version-specific installation directories.
136 If your installation directory is not version-specific, e.g.,
138 /usr/local/pgsql, it is necessary to move the current PostgreSQL
139 install directory so it does not interfere with the new PostgreSQL
140 installation. Once the current PostgreSQL server is shut down, it
141 is safe to rename the PostgreSQL installation directory; assuming
142 the old directory is /usr/local/pgsql, you can do:
143 mv /usr/local/pgsql /usr/local/pgsql.old
145 to rename the directory.
147 2. For source installs, build the new version: Build the new
149 PostgreSQL source with configure flags that are compatible with the
150 old cluster. pg_upgrade will check pg_controldata to make sure all
151 settings are compatible before starting the upgrade.
152 3. Install the new PostgreSQL binaries: Install the new server's
154 binaries and support files. pg_upgrade is included in a default
155 installation.
156 For source installs, if you wish to install the new server in a
158 custom location, use the prefix variable:
159 make prefix=/usr/local/pgsql.new install
161 4. Initialize the new PostgreSQL cluster: Initialize the new cluster
163 using initdb. Again, use compatible initdb flags that match the old
164 cluster. Many prebuilt installers do this step automatically. There
165 is no need to start the new cluster.
166 5. Install extension shared object files: Many extensions and custom
168 modules, whether from contrib or another source, use shared object
169 files (or DLLs), e.g., pgcrypto.so. If the old cluster used these,
170 shared object files matching the new server binary must be
171 installed in the new cluster, usually via operating system
172 commands. Do not load the schema definitions, e.g., CREATE
173 EXTENSION pgcrypto, because these will be duplicated from the old
174 cluster. If extension updates are available, pg_upgrade will report
175 this and create a script that can be run later to update them.
176 6. Copy custom full-text search files: Copy any custom full text
178 search files (dictionary, synonym, thesaurus, stop words) from the
179 old to the new cluster.
180 7. Adjust authentication: pg_upgrade will connect to the old and new
182 servers several times, so you might want to set authentication to
183 peer in pg_hba.conf or use a ~/.pgpass file (see Section 34.16).
184 8. Stop both servers: Make sure both database servers are stopped
186 using, on Unix, e.g.:
187 pg_ctl -D /opt/PostgreSQL/9.6 stop
189 pg_ctl -D /opt/PostgreSQL/14 stop
190 or on Windows, using the proper service names:
192 NET STOP postgresql-9.6
194 NET STOP postgresql-14
195 Streaming replication and log-shipping standby servers can remain
197 running until a later step.
198 9. Prepare for standby server upgrades: If you are upgrading standby
200 servers using methods outlined in section Step 11, verify that the
201 old standby servers are caught up by running pg_controldata against
202 the old primary and standby clusters. Verify that the “Latest
203 checkpoint location” values match in all clusters. (There will be a
204 mismatch if old standby servers were shut down before the old
205 primary or if the old standby servers are still running.) Also,
206 make sure wal_level is not set to minimal in the postgresql.conf
207 file on the new primary cluster.
208 10. Run pg_upgrade: Always run the pg_upgrade binary of the new
210 server, not the old one. pg_upgrade requires the specification of
211 the old and new cluster's data and executable (bin) directories.
212 You can also specify user and port values, and whether you want the
213 data files linked or cloned instead of the default copy behavior.
214 If you use link mode, the upgrade will be much faster (no file
216 copying) and use less disk space, but you will not be able to
217 access your old cluster once you start the new cluster after the
218 upgrade. Link mode also requires that the old and new cluster data
219 directories be in the same file system. (Tablespaces and pg_wal can
220 be on different file systems.) Clone mode provides the same speed
221 and disk space advantages but does not cause the old cluster to be
222 unusable once the new cluster is started. Clone mode also requires
223 that the old and new data directories be in the same file system.
224 This mode is only available on certain operating systems and file
225 systems.
226 The --jobs option allows multiple CPU cores to be used for
228 copying/linking of files and to dump and reload database schemas in
229 parallel; a good place to start is the maximum of the number of CPU
230 cores and tablespaces. This option can dramatically reduce the time
231 to upgrade a multi-database server running on a multiprocessor
232 machine.
233 For Windows users, you must be logged into an administrative
235 account, and then start a shell as the postgres user and set the
236 proper path:
237 RUNAS /USER:postgres "CMD.EXE"
239 SET PATH=%PATH%;C:\Program Files\PostgreSQL\14\bin;
240 and then run pg_upgrade with quoted directories, e.g.:
242 pg_upgrade.exe
244 --old-datadir "C:/Program Files/PostgreSQL/9.6/data"
245 --new-datadir "C:/Program Files/PostgreSQL/14/data"
246 --old-bindir "C:/Program Files/PostgreSQL/9.6/bin"
247 --new-bindir "C:/Program Files/PostgreSQL/14/bin"
248 Once started, pg_upgrade will verify the two clusters are
250 compatible and then do the upgrade. You can use pg_upgrade --check
251 to perform only the checks, even if the old server is still
252 running. pg_upgrade --check will also outline any manual
253 adjustments you will need to make after the upgrade. If you are
254 going to be using link or clone mode, you should use the option
255 --link or --clone with --check to enable mode-specific checks.
256 pg_upgrade requires write permission in the current directory.
257 Obviously, no one should be accessing the clusters during the
259 upgrade. pg_upgrade defaults to running servers on port 50432 to
260 avoid unintended client connections. You can use the same port
261 number for both clusters when doing an upgrade because the old and
262 new clusters will not be running at the same time. However, when
263 checking an old running server, the old and new port numbers must
264 be different.
265 If an error occurs while restoring the database schema, pg_upgrade
267 will exit and you will have to revert to the old cluster as
268 outlined in Step 17 below. To try pg_upgrade again, you will need
269 to modify the old cluster so the pg_upgrade schema restore
270 succeeds. If the problem is a contrib module, you might need to
271 uninstall the contrib module from the old cluster and install it in
272 the new cluster after the upgrade, assuming the module is not being
273 used to store user data.
274 11. Upgrade streaming replication and log-shipping standby servers: If
276 you used link mode and have Streaming Replication (see
277 Section 27.2.5) or Log-Shipping (see Section 27.2) standby servers,
278 you can follow these steps to quickly upgrade them. You will not be
279 running pg_upgrade on the standby servers, but rather rsync on the
280 primary. Do not start any servers yet.
281 If you did not use link mode, do not have or do not want to use
283 rsync, or want an easier solution, skip the instructions in this
284 section and simply recreate the standby servers once pg_upgrade
285 completes and the new primary is running. Install the new
286 PostgreSQL binaries on standby servers: Make sure the new binaries
287 and support files are installed on all standby servers. Make sure
288 the new standby data directories do not exist: Make sure the new
289 standby data directories do not exist or are empty. If initdb was
290 run, delete the standby servers' new data directories. Install
291 extension shared object files: Install the same extension shared
292 object files on the new standbys that you installed in the new
293 primary cluster. Stop standby servers: If the standby servers are
294 still running, stop them now using the above instructions. Save
295 configuration files: Save any configuration files from the old
296 standbys' configuration directories you need to keep, e.g.,
297 postgresql.conf (and any files included by it),
298 postgresql.auto.conf, pg_hba.conf, because these will be
299 overwritten or removed in the next step. Run rsync: When using
300 link mode, standby servers can be quickly upgraded using rsync. To
301 accomplish this, from a directory on the primary server that is
302 above the old and new database cluster directories, run this on the
303 primary for each standby server:
304 rsync --archive --delete --hard-links --size-only --no-inc-recursive old_cluster new_cluster remote_dir
306 where old_cluster and new_cluster are relative to the current
308 directory on the primary, and remote_dir is above the old and new
309 cluster directories on the standby. The directory structure under
310 the specified directories on the primary and standbys must match.
311 Consult the rsync manual page for details on specifying the remote
312 directory, e.g.,
313 rsync --archive --delete --hard-links --size-only --no-inc-recursive /opt/PostgreSQL/9.5 \
315 /opt/PostgreSQL/9.6 standby.example.com:/opt/PostgreSQL
316 You can verify what the command will do using rsync's --dry-run
318 option. While rsync must be run on the primary for at least one
319 standby, it is possible to run rsync on an upgraded standby to
320 upgrade other standbys, as long as the upgraded standby has not
321 been started.
322 What this does is to record the links created by pg_upgrade's link
324 mode that connect files in the old and new clusters on the primary
325 server. It then finds matching files in the standby's old cluster
326 and creates links for them in the standby's new cluster. Files that
327 were not linked on the primary are copied from the primary to the
328 standby. (They are usually small.) This provides rapid standby
329 upgrades. Unfortunately, rsync needlessly copies files associated
330 with temporary and unlogged tables because these files don't
331 normally exist on standby servers.
332 If you have tablespaces, you will need to run a similar rsync
334 command for each tablespace directory, e.g.:
335 rsync --archive --delete --hard-links --size-only --no-inc-recursive /vol1/pg_tblsp/PG_9.5_201510051 \
337 /vol1/pg_tblsp/PG_9.6_201608131 standby.example.com:/vol1/pg_tblsp
338 If you have relocated pg_wal outside the data directories, rsync
340 must be run on those directories too. Configure streaming
341 replication and log-shipping standby servers: Configure the servers
342 for log shipping. (You do not need to run pg_start_backup() and
343 pg_stop_backup() or take a file system backup as the standbys are
344 still synchronized with the primary.)
345 12. Restore pg_hba.conf: If you modified pg_hba.conf, restore its
347 original settings. It might also be necessary to adjust other
348 configuration files in the new cluster to match the old cluster,
349 e.g., postgresql.conf (and any files included by it),
350 postgresql.auto.conf.
351 13. Start the new server: The new server can now be safely started,
353 and then any rsync'ed standby servers.
354 14. Post-upgrade processing: If any post-upgrade processing is
356 required, pg_upgrade will issue warnings as it completes. It will
357 also generate script files that must be run by the administrator.
358 The script files will connect to each database that needs
359 post-upgrade processing. Each script should be run using:
360 psql --username=postgres --file=script.sql postgres
362 The scripts can be run in any order and can be deleted once they
364 have been run.
365 Caution
367 In general it is unsafe to access tables referenced in rebuild
368 scripts until the rebuild scripts have run to completion; doing
369 so could yield incorrect results or poor performance. Tables
370 not referenced in rebuild scripts can be accessed immediately.
371 15. Statistics: Because optimizer statistics are not transferred by
373 pg_upgrade, you will be instructed to run a command to regenerate
374 that information at the end of the upgrade. You might need to set
375 connection parameters to match your new cluster.
376 16. Delete old cluster: Once you are satisfied with the upgrade, you
378 can delete the old cluster's data directories by running the script
379 mentioned when pg_upgrade completes. (Automatic deletion is not
380 possible if you have user-defined tablespaces inside the old data
381 directory.) You can also delete the old installation directories
382 (e.g., bin, share).
383 17. Reverting to old cluster: If, after running pg_upgrade, you wish
385 to revert to the old cluster, there are several options:
386 • If the --check option was used, the old cluster was unmodified;
388 it can be restarted.
389 • If the --link option was not used, the old cluster was
391 unmodified; it can be restarted.
392 • If the --link option was used, the data files might be shared
394 between the old and new cluster:
395 • If pg_upgrade aborted before linking started, the old
397 cluster was unmodified; it can be restarted.
398 • If you did not start the new cluster, the old cluster was
400 unmodified except that, when linking started, a .old suffix
401 was appended to $PGDATA/global/pg_control. To reuse the old
402 cluster, remove the .old suffix from
403 $PGDATA/global/pg_control; you can then restart the old
404 cluster.
405 • If you did start the new cluster, it has written to shared
407 files and it is unsafe to use the old cluster. The old
408 cluster will need to be restored from backup in this case.
409
411 pg_upgrade creates various working files, such as schema dumps, in the
412 current working directory. For security, be sure that that directory is
413 not readable or writable by any other users.
414 pg_upgrade launches short-lived postmasters in the old and new data
416 directories. Temporary Unix socket files for communication with these
417 postmasters are, by default, made in the current working directory. In
418 some situations the path name for the current directory might be too
419 long to be a valid socket name. In that case you can use the -s option
420 to put the socket files in some directory with a shorter path name. For
421 security, be sure that that directory is not readable or writable by
422 any other users. (This is not supported on Windows.)
423 All failure, rebuild, and reindex cases will be reported by pg_upgrade
425 if they affect your installation; post-upgrade scripts to rebuild
426 tables and indexes will be generated automatically. If you are trying
427 to automate the upgrade of many clusters, you should find that clusters
428 with identical database schemas require the same post-upgrade steps for
429 all cluster upgrades; this is because the post-upgrade steps are based
430 on the database schemas, and not user data.
431 For deployment testing, create a schema-only copy of the old cluster,
433 insert dummy data, and upgrade that.
434 pg_upgrade does not support upgrading of databases containing table
436 columns using these reg* OID-referencing system data types:
437 regcollation
438 regconfig
439 regdictionary
440 regnamespace
441 regoper
442 regoperator
443 regproc
444 regprocedure
445 (regclass, regrole, and regtype can be upgraded.)
446 If you are upgrading a pre-PostgreSQL 9.2 cluster that uses a
448 configuration-file-only directory, you must pass the real data
449 directory location to pg_upgrade, and pass the configuration directory
450 location to the server, e.g., -d /real-data-directory -o '-D
451 /configuration-directory'.
452 If using a pre-9.1 old server that is using a non-default Unix-domain
454 socket directory or a default that differs from the default of the new
455 cluster, set PGHOST to point to the old server's socket location. (This
456 is not relevant on Windows.)
457 If you want to use link mode and you do not want your old cluster to be
459 modified when the new cluster is started, consider using the clone
460 mode. If that is not available, make a copy of the old cluster and
461 upgrade that in link mode. To make a valid copy of the old cluster, use
462 rsync to create a dirty copy of the old cluster while the server is
463 running, then shut down the old server and run rsync --checksum again
464 to update the copy with any changes to make it consistent. (--checksum
465 is necessary because rsync only has file modification-time granularity
466 of one second.) You might want to exclude some files, e.g.,
467 postmaster.pid, as documented in Section 26.3.3. If your file system
468 supports file system snapshots or copy-on-write file copies, you can
469 use that to make a backup of the old cluster and tablespaces, though
470 the snapshot and copies must be created simultaneously or while the
471 database server is down.
472
474 initdb(1), pg_ctl(1), pg_dump(1), postgres(1)
475PostgreSQL 14.3 2022 PG_UPGRADE(1)