The {[project]} Releases detail each version of the software and lists the changes made in each version. release numbers consist of two parts, major and minor. A major release may break compatibility with the prior major release, for instance that 1.XX line is not compatible with the 0.XX line. Minor releases can include bug fixes and features but do not change the interface, naming, or the repository format. The notes for a release may also contain Additional Notes but changes in the this section are only to documentation and the test suite and have no impact the on the codebase. David Steele dwsteele Chris Barber gamerscomplete Evan Benoit evanbenoit Christoph Berg ChristophBerg Dmitry Didovicher anarazel Andres Freund anarazel Stephen Frost sfrost John Harvey crunchyjohn Jason O'Donnell Dwaligon Janice Parkinson jpabt Michael Renner terrorobe Eric Radman eradman Cynthia Shang cmwshang Greg Smith gregscds Adrian Vondendriesch disco-stu Jan Wieck wieck

Fixed an issue where keep-alives could be starved out by lots of small files during multi-threaded operation and were completely absent during single-threaded operation when resuming from a previous incomplete backup.

Fixed an issue where the expire command would refuse to run when explicitly called from the command line if the db-host option was set. This was not an issue when expire was run after a backup, which is the usual case.

Added check command to validate that is configured correctly for archiving and backups.

Added the protocol-timeout option. Previously protocol-timeout was set as db-timeout + 30 seconds.

Failure to shutdown remotes at the end of the backup no longer throws an exception. A warning is still generated that recommends a higher protocol-timeout.

The pg_xlogfile_name() function is no longer used to construct WAL filenames from LSNs. While this function is convenient it is not available on a standby. Instead, the archive is searched for the LSN in order to find the timeline. If due to some misadventure the LSN appears on multiple timelines then an error will be thrown, whereas before this condition would have passed unnoticed.

Option handling is now far more strict. Previously it was possible for a command to use an option that was not explicitly assigned to it. This was especially true for the backup-host and db-host options which are used to determine locality.

Fixed DTD search path that did not work properly when --doc-path was used.

Fixed -specific xml that was loaded for non- projects.

Fixed section names being repeated in the info output when multiple --require options depended on the same sections.

Fixed config sections being blank in the output when not loaded from cache.

Allow hidden options to be added to a command. This allows certain commands (like apt-get) to be forced during the build without making that a part of the documentation.

Allow command summaries to be inserted anywhere in the documentation to avoid duplication.

Update TeX Live to 2016 version.

Recommended install location for modules is now /usr/share/perl5 since /usr/lib/perl5 has been removed from the search path in newer versions of Perl.

Added instructions for removing prior versions of .

New, consolidated implementation for link rendering.

version is now a variable to allow multi-version documentation.

Containers now use a squid proxy for apt/yum to speed builds.

Obsolete containers are removed by the --vm-force option.

Greatly reduced the quantity of Docker containers built by default. Containers are only built for versions specified in db-minimal and those required to build documentation. Additional containers can be built with --db-version=all or by specifying a version, e.g. --db-version=9.4.

Fix usage of sprintf() due to new constraints in Perl 5.22. Parameters not referenced in the format string are no longer allowed.

Log directory create and file open now using FileCommon functions which produce more detailed error messages on failure.

Fixed syntax that was not compatible with Perl 5.2X.

Fixed absolute paths that were used for the PDF logo.

Release notes are now broken into sections so that bugs, features, and refactors are clearly delineated. An Additional Notes section has been added for changes to documentation and the test suite that do not affect the core code.

Added man page generation.

Added an execution cache so that documentation can be generated without setting up the full container environment. This is useful for packaging, keeps the documentation consistent for a release, and speeds up generation when no changes are made in the execution list.

The change log was the last piece of documentation to be rendered in Markdown only. Wrote a converter so the document can be output by the standard renderers. The change log will now be located on the website and has been renamed to Releases.

Remove function constants and pass strings directly to logDebugParam(). The function names were only used once so creating constants for them was wasteful.

Lists can now be used outside of p and text tags for more flexible document structuring.

Replaced overzealous perl -cW check which failed on Perl 5.22 with perl -cw.

Added Ubuntu 16.04 (Xenial) and Debian 8 (Jessie) to the regression suite.

Upgraded doc/test VM to Ubuntu 16.04. This will help catch Perl errors in the doc code since it is not run across multiple distributions like the core and test code. It is also to be hoped that a newer kernel will make Docker more stable.

Test release version against the executable using change-log.xml instead of CHANGELOG.md.

Enhanced text output of info command to include timestamps, sizes, and the reference list for all backups.

Allow selective restore of databases from a cluster backup. This feature can result in major space and time savings when only specific databases are restored. Unrestored databases will not be accessible but must be manually dropped before they will be removed from the shared catalogue.

Experimental support for non-exclusive backups in 9.6 beta1. Changes to the control/catalog/WAL versions in subsequent betas may break compatibility but will be updated with each release to keep pace.

IMPORTANT NOTE: This flag day release breaks compatibility with older versions of . The manifest format, on-disk structure, configuration scheme, and the exe/path names have all changed. You must create a new repository to hold backups for this version of and keep your older repository for a time in case you need to do a restore. Restores from the prior repository will require the prior version of but because of name changes it is possible to have 1.00 and a prior version of installed at the same time. See the notes below for more detailed information on what has changed.

Implemented a new configuration scheme which should be far simpler to use. See the User Guide and Configuration Reference for details but for a simple configuration all options can now be placed in the stanza section. Options that are shared between stanzas can be placed in the [global] section. More complex configurations can still make use of command sections though this should be a rare use case.

The repo-path option now always refers to the repository where backups and archive are stored, whether local or remote, so the repo-remote-path option has been removed. The new spool-path option can be used to define a location for queueing WAL segments when archiving asynchronously. A local repository is no longer required.

The default configuration filename is now pgbackrest.conf instead of pg_backrest.conf. This was done for consistency with other naming changes but also to prevent old config files from being loaded accidentally when migrating to 1.00.

The default repository name was changed from /var/lib/backup to /var/lib/pgbackrest.

Lock files are now stored in /tmp/pgbackrest by default. These days /run/pgbackrest is the preferred location but that would require init scripts which are not part of this release. The lock-path option can be used to configure the lock directory.

Log files are now stored in /var/log/pgbackrest by default and no longer have the date appended so they can be managed with logrotate. The log-path option can be used to configure the lock directory.

Executable filename changed from pg_backrest to pgbackrest.

All files and directories linked from PGDATA are now included in the backup. By default links will be restored directly into PGDATA as files or directories. The {[dash]}-link-all option can be used to restore all links to their original locations. The {[dash]}-link-map option can be used to remap a link to a new location.

Removed {[dash]}-tablespace option and replaced with {[dash]}-tablespace-map-all option which should more clearly indicate its function.

Added detail log level which will output more information than info without being as verbose as debug.

Fixed an issue where the master process was passing {[dash]}-repo-remote-path instead of {[dash]}-repo-path to the remote and causing the lock files to be created in the default repository directory (/var/lib/backup), generally ending in failure. This was only an issue when {[dash]}-repo-remote-path was defined on the command line rather than in pg_backrest.conf.

IMPORTANT BUG FIX FOR TABLESPACES: A change to the repository format was accidentally introduced in 0.90 which means the on-disk backup was no longer a valid cluster when the backup contained tablespaces. This only affected users who directly copied the backups to restore clusters rather than using the restore command. However, the fix breaks compatibility with older backups that contain tablespaces no matter how they are being restored ( will throw errors and refuse to restore). New full backups should be taken immediately after installing version 0.91 for any clusters that contain tablespaces. If older backups need to be restored then use a version of that matches the backup version.

Fixed repository incompatibility introduced in 0.90.

Copy global/pg_control last during backups.

Write .info and .manifest files to temp before moving them to their final locations and fsync'ing.

Rename {[dash]}-no-start-stop option to {[dash]}-no-online.

Static source analysis using Perl-Critic, currently passes on gentle.

Fixed an issue where specifying {[dash]}-no-archive-check would throw a configuration error.

Fixed an issue where a temp WAL file left over after a well-timed system crash could cause the next archive-push to fail.

The retention-archive option can now be be safely set to less than backup retention (retention-full or retention-diff) without also specifying archive-copy=n. The WAL required to make the backups that fall outside of archive retention consistent will be preserved in the archive. However, in this case PITR will not be possible for the backups that fall outside of archive retention.

When backing up and restoring tablespaces only operates on the subdirectory created for the version of being run against. Since multiple versions can live in a tablespace (especially during a binary upgrade) this prevents too many files from being copied during a backup and other versions possibly being wiped out during a restore. This only applies to >= 9.0 — prior versions of could not share a tablespace directory.

Generate an error when archive-check=y but archive_command does not execute pg_backrest.

Improved error message when repo-path or repo-remote-path does not exist.

Added checks for {[dash]}-delta and {[dash]}-force restore options to ensure that the destination is a valid $PGDATA directory. will check for the presence of PG_VERSION or backup.manifest (left over from an aborted restore). If neither file is found then {[dash]}-delta and {[dash]}-force will be disabled but the restore will proceed unless there are files in the $PGDATA directory (or any tablespace directories) in which case the operation will be aborted.

When restore {[dash]}-set=latest (the default) the actual backup restored will be output to the log.

Support for 9.5 partial WAL segments and recovery_target_action setting. The archive_mode = 'always' setting is not yet supported.

Support for recovery_target = 'immediate' recovery setting introduced in 9.4.

The following tablespace checks have been added: paths or files in pg_tblspc, relative links in pg_tblspc, tablespaces in $PGDATA. All three will generate errors.

Fixed an issue where document generation failed because some OSs are not tolerant of having multiple installed versions of . A separate VM is now created for each version. Also added a sleep after database starts during document generation to ensure the database is running before the next command runs.

Fixed an issue where longer-running backups/restores would timeout when remote and threaded. Keepalives are now used to make sure the remote for the main process does not timeout while the thread remotes do all the work. The error message for timeouts was also improved to make debugging easier.

Allow restores to be performed on a read-only repository by using {[dash]}-no-lock and {[dash]}-log-level-file=off. The {[dash]}-no-lock option can only be used with restores.

Minor styling changes, clarifications and rewording in the user guide.

The dev branch has been renamed to master and for the time being the master branch has renamed to release, though it will probably be removed at some point {[dash]}- thus ends the gitflow experiment for . It is recommended that any forks get re-forked and clones get re-cloned.

Fixed an issue where the start/stop commands required the {[dash]}-config option.

Fixed an issue where log files were being overwritten instead of appended.

Fixed an issue where backup-user was not optional.

Symlinks are no longer created in backup directories in the repository. These symlinks could point virtually anywhere and potentially be dangerous. Symlinks are still recreated during a restore.

Added better messaging for backup expiration. Full and differential backup expirations are logged on a single line along with a list of all dependent backups expired.

Archive retention is automatically set to full backup retention if not explicitly configured.

Added documentation in the user guide for delta restores, expiration, dedicated backup hosts, starting and stopping , and replication.

The backup_label.old and recovery.done files are now excluded from backups.

Added a new user guide that covers basics and some advanced topics including PITR. Much more to come, but it's a start.

The website, markdown, and command-line help are now all generated from the same XML source.

Fixed an issue where an error could be returned after a backup or restore completely successfully.

Fixed an issue where a resume would fail if temp files were left in the root backup directory when the backup failed. This scenario was likely if the backup process got terminated during the copy phase.

Added stop and start commands to prevent processes from running on a system where is shutdown or the system needs to be quiesced for some other reason.

Experimental support for 9.5 beta1. This may break when the control version or WAL magic changes in future versions but will be updated in each release to keep pace. All regression tests pass except for {[dash]}-target-resume tests (this functionality has changed in 9.5) and there is no testing yet for .partial WAL segments.

Removed dependency on IO::String module.

Fixed an issue where resumed compressed backups were not preserving existing files.

Fixed an issue where resume and incr/diff would not ensure that the prior backup had the same compression and hardlink settings.

Fixed an issue where a cold backup using {[dash]}-no-start-stop could be started on a running cluster without {[dash]}-force specified.

Fixed an issue where a thread could be started even when none were requested.

Fixed an issue where the version number was not being updated in backup.info and archive.info after an upgrade/downgrade.

Fixed an issue where the info command was throwing an exception when the repository contained no stanzas.

Fixed an issue where the pg_stop_backup() NOTICEs were being output to stderr.

Experimental support for 9.5 alpha2. This may break when the control version or WAL magic changes in future versions but will be updated in each release to keep pace. All regression tests pass except for {[dash]}-target-resume tests (this functionality has changed in 9.5) and there is no testing yet for .partial WAL segments.

Renamed recovery-setting option and section to recovery-option to be more consistent with naming conventions.

Code cleanup and refactoring to standardize on patterns that have evolved over time.

Added dynamic module loading to speed up commands, especially asynchronous archiving.

Command-line help is now extracted from the same XML source that is used for the other documentation and includes much more detail.

Expiration tests are now synthetic rather than based on actual backups. This will allow development of more advanced expiration features.

Fixed an issue that caused the formatted timestamp for both the oldest and newest backups to be reported as the current time by the info command. Only text output was affected {[dash]}- json output reported the correct epoch values.

Fixed protocol issue that was preventing ssh errors (especially on connection) from being logged.

The repository is now created and updated with consistent directory and file modes. By default umask is set to 0000 but this can be disabled with the neutral-umask setting.

Added the stop-auto option to allow failed backups to automatically be stopped when a new backup starts.

Added the db-timeout option to limit the amount of time will wait for pg_start_backup() and pg_stop_backup() to return.

Remove pg_control file at the beginning of the restore and copy it back at the very end. This prevents the possibility that a partial restore can be started by .

Added checks to be sure the db-path setting is consistent with db-port by comparing the data_directory as reported by the cluster against the db-path setting and the version as reported by the cluster against the value read from pg_control. The db-socket-path setting is checked to be sure it is an absolute path.

Experimental support for 9.5 alpha1. This may break when the control version or WAL magic changes in future versions but will be updated in each release to keep pace. All regression tests pass except for {[dash]}-target-resume tests (this functionality has changed in 9.5) and there is no testing yet for .partial WAL segments.

Now using Perl DBI and DBD::Pg for connections to rather than psql. The cmd-psql and cmd-psql-option settings have been removed and replaced with db-port and db-socket-path. Follow the instructions in the Installation Guide to install DBD::Pg on your operating system.

Major refactoring of the protocol layer to support future development.

Split most of README.md out into USERGUIDE.md and CHANGELOG.md because it was becoming unwieldy. Changed most references to database in the user guide to database cluster for clarity.

Changed most references to database in the user guide to database cluster for clarity.

Added vagrant test configurations for Ubuntu 14.04 and CentOS 7.

Removed dependency on CPAN packages for multi-threaded operation. While it might not be a bad idea to update the threads and Thread::Queue packages, it is no longer necessary.

Modified wait backoff to use a Fibonacci rather than geometric sequence. This will make wait time grow less aggressively while still giving reasonable values.

Added vagrant test configurations for Ubuntu 12.04 and CentOS 6.

More options for regression tests and improved code to run in a variety of environments.

Added file and directory syncs to the File object for additional safety during backup/restore and archiving.

Added support for Perl 5.10.1 and OpenSSH 5.3 which are default for CentOS/RHEL 6.

Improved error message when backup is run without archive_command set and without {[dash]}-no-archive-check specified.

Removed pg_backrest_remote and added the functionality to pg_backrest as the remote command.

Moved version number out of the VERSION file to Version.pm to better support packaging.

Replaced IPC::System::Simple and Net::OpenSSH with IPC::Open3 to eliminate CPAN dependency for multiple operating systems.

IMPORTANT NOTE: This flag day release breaks compatibility with older versions of . The manifest format, on-disk structure, and the binary names have all changed. You must create a new repository to hold backups for this version of and keep your older repository for a time in case you need to do a restore. The pg_backrest.conf file has not changed but you'll need to change any references to pg_backrest.pl in cron (or elsewhere) to pg_backrest (without the .pl extension).

Added the info command.

Logging now uses unbuffered output. This should make log files that are being written by multiple threads less chaotic.

Experimental support for 9.5. This may break when the control version or WAL magic changes but will be updated in each release.

More efficient file ordering for backup. Files are copied in descending size order so a single thread does not end up copying a large file at the end. This had already been implemented for restore.

Fixed an issue where archive-copy would fail on an incr/diff backup when hardlink=n. In this case the pg_xlog path does not already exist and must be created.

Fixed an issue in async archiving where archive-push was not properly returning 0 when archive-max-mb was reached and moved the async check after transfer to avoid having to remove the stop file twice. Also added unit tests for this case and improved error messages to make it clearer to the user what went wrong.

Fixed a locking issue that could allow multiple operations of the same type against a single stanza. This appeared to be benign in terms of data integrity but caused spurious errors while archiving and could lead to errors in backup/restore.

Allow duplicate WAL segments to be archived when the checksum matches. This is necessary for some recovery scenarios.

Allow comments/disabling in pg_backrest.conf using the # character. Only # characters in the forst character of the line are honored.

Better logging before pg_start_backup() to make it clear when the backup is waiting on a checkpoint.

Various command behavior and logging fixes.

Replaced JSON module with JSON::PP which ships with core Perl.

Various help fixes.

Fixed an issue where an absolute path was not written into recovery.conf when the restore was run with a relative path.

Better resume support. Resumed files are checked to be sure they have not been modified and the manifest is saved more often to preserve checksums as the backup progresses. More unit tests to verify each resume case.

Resume is now optional. Use the resume setting or {[dash]}-no-resume from the command line to disable.

More info messages during restore. Previously, most of the restore messages were debug level so not a lot was output in the log.

Added tablespace setting to allow tablespaces to be restored into the pg_tblspc path. This produces compact restores that are convenient for development, staging, etc. Currently these restores cannot be backed up as expects only links in the pg_tblspc path.

Fixed a buffering error that could occur on large, highly-compressible files when copying to an uncompressed remote destination. The error was detected in the decompression code and resulted in a failed backup rather than corruption so it should not affect successful backups made with previous versions.

Pushing duplicate WAL now generates an error. This worked before only if checksums were disabled.

Database System IDs are used to make sure that all WAL in an archive matches up. This should help prevent misconfigurations that send WAL from multiple clusters to the same archive.

Improved threading model by starting threads early and terminating them late.

Regression tests working back to 8.3.

Fixed broken checksums and now they work with normal and resumed backups. Finally realized that checksums and checksum deltas should be functionally separated and this simplified a number of things. Issue #28 has been created for checksum deltas.

Fixed an issue where a backup could be resumed from an aborted backup that didn't have the same type and prior backup.

Added restore functionality.

All options can now be set on the command-line making pg_backrest.conf optional.

De/compression is now performed without threads and checksum/size is calculated in stream. That means file checksums are no longer optional.

Added option {[dash]}-no-start-stop to allow backups when Postgres is shut down. If postmaster.pid is present then {[dash]}-force is required to make the backup run (though if Postgres is running an inconsistent backup will likely be created). This option was added primarily for the purpose of unit testing, but there may be applications in the real world as well.

Checksum for backup.manifest to detect a corrupted/modified manifest.

Link latest always points to the last backup. This has been added for convenience and to make restores simpler.

Removed dependency on Moose. It wasn't being used extensively and makes for longer startup times.

More comprehensive unit tests in all areas.

Complete rewrite of BackRest::File module to use a custom protocol for remote operations and Perl native GZIP and SHA operations. Compression is performed in threads rather than forked processes.

Removed dependency on Storable and replaced with a custom ini file implementation.

Numerous other changes that can only be identified with a diff.

Added much needed documentation

Fairly comprehensive unit tests for all the basic operations. More work to be done here for sure, but then there is always more work to be done on unit tests.

Found and squashed a nasty bug where file_copy() was defaulted to ignore errors. There was also an issue in file_exists() that was causing the test to fail when the file actually did exist. Together they could have resulted in a corrupt backup with no errors, though it is very unlikely.

Worked on improving error handling in the File object. This is not complete, but works well enough to find a few errors that have been causing us problems (notably, find is occasionally failing building the archive async manifest when system is under load).

The archive-get command now returns a 1 when the archive file is missing to differentiate from hard errors (ssh connection failure, file copy error, etc.) This lets know that that the archive stream has terminated normally. However, this does not take into account possible holes in the archive stream.

If an archive directory which should be empty could not be deleted backrest was throwing an error. There's a good fix for that coming, but for the time being it has been changed to a warning so processing can continue. This was impacting backups as sometimes the final archive file would not get pushed if the first archive file had been in a different directory (plus some bad luck).

Added RequestTTY=yes to ssh sessions. Hoping this will prevent random lockups.

Added archive-get functionality to aid in restores.

Added option to force a checkpoint when starting the backup, start-fast=y.

Removed master_stderr_discard option on database SSH connections. There have been occasional lockups and they could be related to issues originally seen in the file code.

Changed lock file conflicts on backup and expire commands to ERROR. They were set to DEBUG due to a copy-and-paste from the archive locks.

No restore functionality, but the backup directories are consistent data directories. You'll need to either uncompress the files or turn off compression in the backup. Uncompressed backups on a ZFS (or similar) filesystem are a good option because backups can be restored locally via a snapshot to create logical backups or do spot data recovery.

Archiving is single-threaded. This has not posed an issue on our multi-terabyte databases with heavy write volume. Recommend a large WAL volume or to use the async option with a large volume nearby.

Backups are multi-threaded, but the Net::OpenSSH library does not appear to be 100% thread-safe so it will very occasionally lock up on a thread. There is an overall process timeout that resolves this issue by killing the process. Yes, very ugly.

Checksums are lost on any resumed backup. Only the final backup will record checksum on multiple resumes. Checksums from previous backups are correctly recorded and a full backup will reset everything.

The backup.manifest is being written as Storable because Config::IniFile does not seem to handle large files well. Would definitely like to save these as human-readable text.

Absolutely no documentation (outside the code). Well, excepting these release notes.