Bug Fixes:
* Fixed an issue where an archive-push error would not be retried and would instead return errors to PostgreSQL indefinitely (unless the .error file was manually deleted). (Reported by Jens Wilke.)
* Fixed a race condition in parallel archiving where creation of new paths generated an error when multiple processes attempted to do so at the same time. (Reported by Jens Wilke.)
Refactoring:
* Improved performance of wal archive min/max provided by the info command. (Suggested by Jens Wilke.)
* Documentation can now be built with reusable blocks to reduce duplication.
* Added ability to pass options to containers within the documentation.
* Add proper tag to slightly emphasize proper nouns.
IMPORTANT NOTE: The new implementation of asynchronous archiving no longer copies WAL to a separate queue. If there is any WAL left over in the old queue after upgrading to 1.13, it will be abandoned and not pushed to the repository.
To prevent this outcome, stop archiving by setting archive_command = false. Next, drain the async queue by running pgbackrest --stanza=[stanza-name] archive-push and wait for the process to complete. Check that the queue in [spool-path]/archive/[stanza-name]/out is empty. Finally, install 1.13 and restore the original archive_command.
IMPORTANT NOTE: The stanza-create command is not longer optional and must be executed before backup or archiving can be performed on a new stanza. Pre-existing stanzas do not require stanza-create to be executed.
Bug Fixes:
* Fixed const assignment giving compiler warning in C library. (Fixed by Adrian Vondendriesch.)
* Fixed a few directory syncs that were missed for the --repo-sync option.
* Fixed an issue where a missing user/group on restore could cause an "uninitialized value" error in File->owner(). (Reported by Leonardo Avellar.)
* Fixed an issue where protocol mismatch errors did not output the expected value.
* Fixed a spurious archive-get log message that indicated an exit code of 1 was an abnormal termination.
Features:
* Improved, multi-process implementation of asynchronous archiving.
* Improved stanza-create command so that it can repair broken repositories in most cases and is robust enough to be made mandatory. (Contributed by Cynthia Shang.)
* Improved check command to run on a standby, though only basic checks are done because pg_switch_xlog() cannot be executed on a replica. (Contributed by Cynthia Shang.)
* Added archive and backup WAL ranges to the info command.
* Added warning to update pg_tablespace.spclocation when remapping tablespaces in PostgreSQL < 9.2. (Contributed by blogh.)
* Remove remote lock requirements for the archive-get, restore, info, and check commands since they are read-only operations. (Suggested by Michael Vitale.)
Refactoring:
* Refactor File and BackupCommon modules to improve test coverage.
* Return proper error code when unable to convert a relative path to an absolute path. (Suggested by Yogesh Sharma.)
* Log file banner is not output until the first log entry is written. (Suggested by Jens Wilke.)
* Moved File->manifest() into the FileCommon.pm module.
* Moved the Archive modules to the Archive directory and split the archive-get and archive-push commands into separate modules.
* Split the check command out of the Archive.pm module.
* Allow logging to be suppressed via logDisable() and logEnable().
* Allow for locks to be taken more than once in the same process without error.
* Lock directories can be created when more than one directory level is required.
* Clean up optionValid()/optionTest() logic in Lock.pm.
* Added Exception::exceptionCode() and Exception::exceptionMessage() to simplify error handling logic.
* Represent .gz extension with a constant.
* Allow empty files to be created with FileCommon::fileStringWrite() and use temp files to avoid partial reads.
* Refactor process IO and process master/minion code out from the common protocol code.
* Reduced the likelihood of torn pages causing a false positive in page checksums by filtering on start backup LSN.
* Remove Intel-specific optimization from C library build flags. (Contributed by Adrian Vondendriesch.)
* Removed --lock option. This option was introduced before the lock directory could be located outside the repository and is now obsolete.
* Added --log-timestamp option to allow timestamps to be suppressed in logging. This is primarily used to avoid filters in the automated documentation.
* Fixed alignment issues with multiline logging.
* Allow for locks to be taken more than once in the same process without error.
* Lock directories can be created when more than one directory level is required.
* Clean up optionValid()/optionTest() logic in Lock.pm.
* Allow logging to be suppressed via logDisable() and logEnable().
* Added more flexibility in initializing and cleaning up after modules and tests.
* testResult() suppresses logging and reports exceptions.
* testException() allows messages to be matched with regular expressions.
* Refactor name/locations of common modules that setup test environments.
IMPORTANT NOTE: In prior releases it was possible to specify options on the command-line that were invalid for the current command without getting an error. An error will now be generated for invalid options so it is important to carefully check command-line options in your environment to prevent disruption.
Bug Fixes:
* Fixed an issue where options that were invalid for the specified command could be provided on the command-line without generating an error. The options were ignored and did not cause any change in behavior, but it did lead to some confusion. Invalid options will now generate an error. (Reported by Nikhilchandra Kulkarni.)
* Fixed an issue where internal symlinks were not being created for tablespaces in the repository. This issue was only apparent when trying to bring up clusters in-place manually using filesystem snapshots and did not affect normal backup and restore.
* Fixed an issue that prevented errors from being output to the console before the logging system was initialized, i.e. while parsing options. Error codes were still being returned accurately so this would not have made a process look like it succeeded when it did not. (Reported by Adrian Vondendriesch.)
* Fixed an issue where the db-port option specified on the backup server would not be properly passed to the remote unless it was from the first configured database. (Reported by Michael Vitale.)
Features:
* Added the --checksum-page option to allow pgBackRest to validate page checksums in data files when checksums are enabled on PostgreSQL >= 9.3. Note that this functionality requires a C library which may not initially be available in OS packages. The option will automatically be enabled when the library is present and checksums are enabled on the cluster. (Suggested by Stephen Frost.)
* Added the --repo-link option to allow internal symlinks to be suppressed when the repository is located on a filesystem that does not support symlinks. This does not affect any pgBackRest functionality, but the convenience link latest will not be created and neither will internal tablespace symlinks, which will affect the ability to bring up clusters in-place manually using filesystem snapshots.
* Added the --repo-sync option to allow directory syncs in the repository to be disabled for file systems that do not support them, e.g. NTFS.
* Added a predictable log entry to signal that a command has completed successfully. For example a backup ends successfully with: INFO: backup command end: completed successfully. (Suggested by Jens Wilke.)
Refactoring:
* Abstracted code to determine which database cluster is the master and which are standbys. (Contributed by Cynthia Shang.)
* Improved consistency and flexibility of the protocol layer by using JSON for all messages.
* File copy protocol now accepts a function that can do additional processing on the copy buffers and return a result to the calling process.
* Improved IO->bufferRead to always return requested number of bytes until EOF.
* For simplicity, the pg_control file is now copied with the rest of the files instead of by itself of at the end of the process. The backup command does not require this behavior and the restore copies to a temporary file which is renamed at the end of the restore.
* Simplified the result hash of File->manifest(), Db->tablespaceMapGet(), and Db->databaseMapGet().
* Improved errors returned from child processes by removing redundant error level and code.
* Code cleanup in preparation for improved stanza-create command. (Contributed by Cynthia Shang.)
* Improved parameter/result logging in debug/trace functions.
This option allows pgBackRest to validate page checksums in data files when checksums are enabled on PostgreSQL >= 9.3. Note that this functionality requires a C library which may not initially be available in OS packages. The option will automatically be enabled when the library is present and checksums are enabled on the cluster.
* The options were ignored and did not cause any change in behavior, but it did lead to some confusion. Invalid options will now generate an error.
* Removed erroneous --no-config option in help test module.
* Changed the --no-fork test option to --fork with negation to match all other boolean parameters.
That is, while parsing options. Error codes were still being returned accurately so this would not have made a process look like it succeeded when it did not.
Allow internal symlinks to be suppressed when the repository is located on a filesystem that does not support symlinks. This does not affect any pgBackRest functionality, but the convenience link latest will not be created and neither will internal tablespace symlinks, which will affect the ability to bring up clusters in-place manually using filesystem snapshots.
This was introduced in an effort to make the html output XHTML 1.0 STRICT compliant because the standard does not allow / characters in anchors.
However, the / characters were changed to . in the anchors but not in the links. For now revert the anchors to / so further though can be given to this issue.
Bug Fixes:
* Fixed and issue that suppressed exceptions in PDF builds.
Features:
* Allow a source to be included as a section so large documents can be broken up.
* Added section link support to Markdown output.
* Added list support to PDF output.
* Added include option to explicitly build sources (complements the exclude option though both cannot be used in the same invocation).
* Added keyword-add option to add keywords without overriding the default keyword.
* Added debug option to doc.pl to easily add the debug keyword to documentation builds.
* Added pre option to doc.pl to easily add the pre keyword to documentation builds.
Refactoring:
* Improvements to markdown rendering.
* Remove code dependency on project variable, instead use title param.
Bug Fixes:
* Fixed an issue where asynchronous archiving was transferring one file per execution instead of transferring files in batches. This regression was introduced in v1.09 and affected efficiency only, all WAL segments were correctly archived in asynchronous mode. (Reported by Stephen Frost.)
This regression was introduced in v1.09 and affected efficiency only, all WAL segments were correctly archived in asynchronous mode.
Reported by Stephen Frost.
Bug Fixes:
* Fixed an issue where a backup could error if no changes were made to a database between backups and only pg_control changed.
* Fixed an issue where tablespace paths with the same prefix would cause an invalid link error. (Reported by Nikhilchandra Kulkarni.)
Features:
* Added the stanza-create command to formalize creation of stanzas in the repository. (Contributed by Cynthia Shang.)
Refactoring:
* Removed extraneous use lib directives from Perl modules. (Suggested by Devrim Gündüz.)
Bug Fixes:
* Fixed missing variable replacements.
* Removed hard-coded host names from configuration file paths.
Documentation Features:
* Allow command-line length to be configured using cmd-line-len param.
* Added compact param to allow CSS to be embedded in HTML file.
* Added pretty param to produce HTML with proper indenting.
* Only generate HTML menu when required and don't require index page.
* Assign numbers to sections by default.
* VM mount points are now optional.
Bug Fixes:
* Fixed the check command to prevent an error message from being logged if the backup directory does not exist. (Fixed by Cynthia Shang.)
* Fixed error message to properly display the archive command when an invalid archive command is detected. (Reported by Jason O'Donnell.)
* Fixed an issue where the async archiver would not be started if archive-push did not have enough space to queue a new WAL segment. This meant that the queue would never be cleared without manual intervention (such as calling archive-push directly). PostgreSQL now receives errors when there is not enough space to store new WAL segments but the async process will still be started so that space is eventually freed. (Reported by Jens Wilke.)
* Fixed a remote timeout that occurred when a local process generated checksums (during resume or restore) but did not copy files, allowing the remote to go idle. (Reported by Jens Wilke.)
Features:
* Non-exclusive backups will automatically be used on PostgreSQL 9.6.
* Added the cmd-ssh option to allow the ssh client to be specified. (Suggested by Jens Wilke.)
* Added the log-level-stderr option to control whether console log messages are sent to stderr or stdout. By default this is set to warn which represents a change in behavior from previous versions, even though it may be more intuitive. Setting log-level-stderr=off will preserve the old behavior. (Suggested by Sascha Biberhofer.)
* Set application_name to "pgBackRest [command]" for database connections. (Suggested by Jens Wilke.)
* Check that archive_mode is enabled when archive-check option enabled.
Refactoring:
* Clarified error message when unable to acquire pgBackRest advisory lock to make it clear that it is not a PostgreSQL backup lock. (Suggested by Jens Wilke.)
* pgBackRest version number included in command start INFO log output.
* Process ID logged for local process start/stop INFO log output.
The timeout occurred when a local process generated checksums (during resume or restore) but did not copy files, allowing the remote to go idle.
Reported by Jens Wilke.
This meant that the queue would never be cleared without manual intervention (such as calling archive-push directly). PostgreSQL now receives errors when there is not enough space to store new WAL segments but the async process will still be started so that space is eventually freed.
Reported by Jens Wilke.
Controls whether console log messages are sent to stderr or stdout. By default this is set to warn which represents a change in behavior from previous versions, even though it may be more intuitive. Setting log-level-stderr=off will preserve the old behavior.
Suggested by Sascha Biberhofer.
* Fixed error message to properly display the archive command when an invalid archive command is detected.
* Check that archive_mode is enabled when archive-check option enabled.
* pgBackRest version number included in command start INFO log output.
* Process ID logged for local process start/stop INFO log output.
* Fixed missing expect output for help module.
Bug Fixes:
* Fixed an issue where local processes were not disconnecting when complete and could later timeout. (Reported by Todd Vernick.)
* Fixed an issue where the protocol layer could timeout while waiting for WAL segments to arrive in the archive. (Reported by Todd Vernick.)
Refactoring:
* Cache file log output until the file is created to create a more complete log.
* Fixed an issue where local processes were not disconnecting when complete and could later timeout. (Reported by Todd Vernick.)
* Fixed an issue where the protocol layer could timeout while waiting for WAL segments to arrive in the archive. (Reported by Todd Vernick.)
Bug Fixes:
* Fixed an issue where tablespaces were copied from the master during standby backup.
* Fixed the check command so backup info is checked remotely and not just locally. (Fixed by Cynthia Shang.)
* Fixed an issue where retention-archive was not automatically being set when retention-archive-type=diff, resulting in a less aggressive than intended expiration of archive. (Fixed by Cynthia Shang.)
Features:
* Converted Perl threads to processes to improve compatibility and performance.
* Exclude contents of $PGDATA/pg_replslot directory so that replication slots on the master do not become part of the backup.
* The archive-start and archive-stop settings are now filled in backup.manifest even when archive-check=n.
* Additional warnings when archive retention settings may not have the intended effect or would allow indefinite retention. (Contributed by Cynthia Shang.)
* Experimental support for non-exclusive backups in PostgreSQL 9.6 rc1. Changes to the control/catalog/WAL versions in subsequent release candidates may break compatibility but pgBackRest will be updated with each release to keep pace.
Refactoring:
* Refactor of protocol minions in preparation for the new local minion.
* Remove obsolete thread index variable from File() module.
* Changed temporary file names to consistently use the .pgbackrest.tmp extension even if the destination file is compressed or has an appended checksum.
* Improve ASSERT error handling, safely check eval blocks, and convert $@ to $EVAL_ERROR.
* Fixed an issue where retention-archive was not automatically being set when retention-archive-type=diff, resulting in a less aggressive than intended expiration of archive.
* Additional warnings when archive retention settings may not have the intended effect or would allow indefinite retention.
* Closed#235: "Retention policy question" by adding documentation for archive retention.
Contributed by Cynthia Shang.
Bug Fixes:
* Fixed an issue where a tablespace link that referenced another link would not produce an error, but instead skip the tablespace entirely. (Reported by Michael Vitale.)
* Fixed an issue where options that should not allow multiple values could be specified multiple times in pgbackrest.conf without an error being raised. (Reported by Michael Vitale.)
* Fixed an issue where the protocol-timeout option was not automatically increased when the db-timeout option was increased. (Reported by Todd Vernick.)
Features:
* Backup from a standby cluster. A connection to the primary cluster is still required to start/stop the backup and copy files that are not replicated, but the vast majority of files are copied from the standby in order to reduce load on the master.
* More flexible configuration for databases. Master and standby can both be configured on the backup server and pgBackRest will automatically determine which is the master. This means no configuration changes for backup are required after failing over from a master to standby when a separate backup server is used.
* Exclude directories during backup that are cleaned, recreated, or zeroed by PostgreSQL at startup. These include pgsql_tmp and pg_stat_tmp. The postgresql.auto.conf.tmp file is now excluded in addition to files that were already excluded: backup_label.old, postmaster.opts, postmaster.pid, recovery.conf, recovery.done.
* Experimental support for non-exclusive backups in PostgreSQL 9.6 beta4. Changes to the control/catalog/WAL versions in subsequent betas may break compatibility but pgBackRest will be updated with each release to keep pace.
Refactoring:
* Simplify protocol creation and identifying which host is local/remote.
* Removed all OP_* function constants that were used only for debugging, not in the protocol, and replaced with __PACKAGE__.
* Improvements in Db module: separated out connect() function, allow executeSql() calls that do not return data, and improve error handling.
* Improve error message for links that reference links in manifest build.
* Added hints to error message when relative paths are detected in archive-push or archive-get.
* Improve backup log messages to indicate which host the files are being copied from.
A connection to the primary cluster is still required to start/stop the backup and copy files that are not replicated, but the vast majority of files are copied from the standby in order to reduce load on the master.
Master and standby can both be configured on the backup server and pgBackRest will automatically determine which is the master. This means no configuration changes for backup are required after failing over from a master to standby when a separate backup server is used.
These include (depending on the version where they were introduced): pgsql_tmp, pg_dynshmem, pg_notify, pg_replslot, pg_serial, pg_snapshots, pg_stat_tmp, pg_subtrans. The postgresql.auto.conf.tmp file is now excluded in addition to files that were already excluded: backup_label.old, postmaster.opts, postmaster.pid, recovery.conf, recovery.done.
Bug Fixes:
* Fixed an issue where tablespace paths that had $PGDATA as a substring would be identified as a subdirectories of $PGDATA even when they were not. Also hardened relative path checking a bit. (Reported by Chris Fort.)
* Tablespace paths that had $PGDATA as a substring would be identified as a subdirectories of $PGDATA even when they were not.
* Also hardened relative path checking a bit.
Bug Fixes:
* Fixed an issue an where an extraneous remote was created causing threaded backup/restore to possibly timeout and/or throw a lock conflict. (Reported by Michael Vitale.)
* Fixed an issue where db-path was not required for the check command so an assert was raised when it was missing rather than a polite error message. (Reported by Michael Vitale.)
* Fixed check command to throw an error when database version/id does not match that of the archive. (Fixed by Cynthia Shang.)
* Fixed an issue where a remote could try to start its own remote when the backup-host option was not present in pgbackrest.conf on the database server. (Reported by Lardière Sébastien.)
* Fixed an issue where the contents of pg_xlog were being backed up if the directory was symlinked. This didn't cause any issues during restore but was a waste of space.
* Fixed an invalid log() call in lock routines.
Features:
* Experimental support for non-exclusive backups in PostgreSQL 9.6 beta3. Changes to the control/catalog/WAL versions in subsequent betas may break compatibility but pgBackRest will be updated with each release to keep pace.
Refactoring:
* Enhancements to the protocol layer for improved reliability and error handling.
* All remote types now take locks. The exceptions date to when the test harness and pgBackRest were running in the same VM and no longer apply.
* Exceptions are now passed back from threads as messages when possible rather than raised directly.
* Temp files created during backup are now placed in the same directory as the target file.
* Output lock file name when a lock cannot be acquired to aid in debugging.
* Reduce calls to protocolGet() in backup/restore.
* Suppress banners on SSH protocol connections.
* Improved remote error messages to identify the host where the error was raised.
This is a better approach than 93320b8 (reverted in this commit) because it ensures that the remote type will be none so any functions that utilize optionRemoteTypeTest will work correctly.
This bug was only an issue when backup-host was not properly configured on the database host.
Bug Fixes:
* Fixed an issue where keep-alives could be starved out by lots of small files during multi-threaded backup. They were also completely absent from single/multi-threaded backup resume and restore checksumming. (Reported by Janice Parkinson, Chris Barber.)
* 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 automatically after a backup (Reported by Chris Barber.)
* Fixed an issue where validation was being running on archive_command even when the archive-check option was disabled.
Features:
* Added check command to validate that pgBackRest is configured correctly for archiving and backups. (Contributed by Cynthia Shang.)
* 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. Instead a warning is generated that recommends a higher protocol-timeout.
* Experimental support for non-exclusive backups in PostgreSQL 9.6 beta2. Changes to the control/catalog/WAL versions in subsequent betas may break compatibility but pgBackRest will be updated with each release to keep pace.
Refactoring:
* 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.
* Improved handling of users/groups captured during backup that do not exist on the restore host. Also explicitly handle the case where user/group is not mapped to a name.
* Changed version variable to a constant. It had originally been designed to play nice with a specific packaging tool but that tool was never used.
Improved handling of users/groups captured during backup that do not exist on the restore host. Also explicitly handle the case where user/group is not mapped to a name.
This was worked out as part of the test suite refactor [c8f806a] but not committed with it because of the large number of expect logs changes involved. Keeping them separate made it easier to audit the changes in the refactor.
* Make the code more modular and object-oriented.
* Multiple Docker containers can now be created for a single test to simulate more realistic environments.
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.
* 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.
Reported by Janice Parkinson.
* 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.
* 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.
* 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.
Reported by Chris Barber.
* Containers now use a squid proxy for apt/yum to speed builds.
* Obsolete containers are removed by the <br-option>--vm-force</br-option> option.
* Greatly reduced the quantity of Docker containers built by default. Containers are only built for PostgreSQL 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.
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.
* Recommended install location for pgBackRest 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 pgBackRest.
* Fix usage of sprintf() due to new constraints in Perl 5.22. Parameters not referenced in the format string are no longer allowed. (Fixed by Adrian Vondendriesch.)
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.
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.
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".
