File: ChangeLog

package info (click to toggle)
barman 2.6-1
links: PTS, VCS
area: main
in suites: buster
size: 2,896 kB
sloc: python: 9,937; makefile: 50; sh: 38
file content (4369 lines) | stat: -rw-r--r-- 151,182 bytes
2019-01-31  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Prepare release 2.6

2019-01-30  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fix flake8 errors

2019-01-29  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Improved documentation on geo-redundancy
	Contains also minor fixes and typos fixes

2019-01-10  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add colored output for check and error/warning messages
	Use ANSI escape codes for color selection

2018-11-15  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Geographic redundancy implementation
	Introduce three new commands (`sync-info`, `sync-backup` and `sync-wals`)
	and one global/server configuration option called `primary_ssh_command`.

	When the latter is specified globally, the whole instance of Barman is
	an async copy of the server reached via SSH by `primary_ssh_command`.
	If specified on a single server definition, that server in Barman is
	an async copy of the same server that is defined on another Barman
	installation.

	Geo-redundancy is asynchronous and based on SSH connections between
	Barman servers. Cascading backup is supported.

2019-01-04  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Implement put-wal command

2019-01-08  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add utils.fsync_file() method

2019-01-10  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Clarify output of list-backup on 'WAITING_FOR_WALS' backups

2018-12-24  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Do not treat lock file busy as an error when validating a backup
	This patch prevents the `barman backup` command to terminate with a
	failure because of a race condition with the cron starting a
	backup validation.

2018-12-06  Abhijit Menon-Sen  <ams@2ndQuadrant.com>

	Fix typo: consistencty → consistency

2018-11-23  Martín Marqués  <martin.marques@2ndquadrant.com>

	Typo in NEWS file refering to xlogdb

2018-11-24  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix current_action in concurrent stop backup errors

2018-11-21  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Documentation: exclude PGDG repository from Barman RPM management
	Document how to exclude any Barman related software from getting
	updated via PGDG RPM repositories.

2018-11-13  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix WAL compression detection algorithm
	This patch fixes a misdetection of the compression status of WAL files
	into the archive when compression method changes.

2018-11-09  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Set version to 2.6a1

2018-11-08  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Honour archiver locking in `wait_for_wal` method
	Fixes a race between the archiver run by the `cron` and
	the archiver run by `wait_for_wal` method.
	Now only one archiver can run at the same time.

2018-11-05  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.com>

	Make code compliant with the newer Flake8

	Fix switch-wal on standby with WAL dir empty
	This patch fixes an error happening using the `switch-wal` command on a
	standby server when the WAL directory is empty.

	Closes: #5993

2018-10-22  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Set version to 2.5

2018-10-17  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.com>

	Avoid checking failed backups
	If a backup is in a different state than WAITING_FOR_WALS, we don't need
	to check it for consistency.

2018-10-17  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Prepared release notes for 2.5 with tentative release date

2018-10-16  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Avoid being verbose about missing WAL files in `check-backup`
	The number of missing WAL files can be huge, and storing such list or
	displaying it in the error message is consuming and not useful.
	The check now stops as soon as it finds a missing file, and the message
	has changed to display only that file.

2018-10-15  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.com>

	Support switch-wal on standby server
	The switch-wal command can also be used on standby servers, but only for
	the `--archive` option.

	Use XLOG segment size in check_backup
	The `check_backup` command wasn't using the current XLOG segment size to
	generate the WAL file names.

2018-10-15  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Implement atomic BackupInfo writes

2018-10-15  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.com>

	Notify users when attempting to recover a WAITING_FOR_WALS backup
	When a backup which is still in WAITING_FOR_WALS state will be
	recovered, a warning message will be raised if the get-wal feature
	is not in use.

	Another warning message will be raised if the backup is still in the
	same state after having copied the data files.

2018-10-15  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Always update the backup status during `check-backup` execution
	This change aims to detect cases where the content of the WAL archive
	has been modified outside Barman.

2018-10-15  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Cover WAL streaming in documentation for backup from a standby

2018-10-14  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	In PostgreSQL 11 `wal_segment_size` is returned in bytes

2018-10-04  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.com>

	Add `check-backup` command
	Make sure that all the required WAL files to check the consistency
	of a physical backup (that is, from the beginning to the end of
	the full backup, as outlined in the backup label) are correctly archived.
	This command is automatically invoked by the `cron` command,
	and at the end of every backup operation.

2018-10-09  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Set version to 2.5a1

2018-09-14  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Pin py<1.6.0 to keep 2.6 support

2018-08-02  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix decoding errors reading external commands output
	Completed the fix started in commit f14e405e9b36de912899d0735563eed76f479164

	The StreamLineProcessor implementation is enough to ensure that
	the Command out and err property are proper unicode strings or None

	Closes: #174

	Pin pytest-timeout<1.2.1 to continue supporting Python 2.6

2018-05-23  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Set version to 2.4

2018-05-23  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Fix doc about RPM/APT repositories

2018-05-18  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

2018-05-15  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fix invalid "Last Archived WAL" for Postgres < 9.4
	Closes #110 #113

2018-05-14  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Fix links syntax for Markdown in documentation

	Add 2ndQuadrant public repositories info

	Prepared release notes for 2.4 with tentative release date

2018-05-11  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Downgraded sync standby names messages to debug
	Fixes #89

2018-05-10  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.com>

	Ensure CommandWrapper has an unicode output
	Depending on the version of Python and on the usage, the CommandWrapper
	was collecting the output of the wrapper command as a byte string or as
	an unicode string.

	This patch makes the `out` and the `err` properties of the
	CommandWrapper be always an unicode string.

	Closes #150
	Closes #143

2018-05-09  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.com>

	Treat slot names as lowercase
	PostgreSQL replication slot names can contain only lowercase letters,
	numbers and the underscore character.

	This patch makes Barman treating the slot names as lowercase and
	reporting configuration errors when invalid characters (beside uppercase
	letters) are used.

	Closes: #170

2018-05-09  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Improve usability of point-in-time recovery
	Add a check to stop the recovery immediately if the recovery time is
	before the backup end time.

	If the timezone is not specified, the target recovery time is
	interpreted according to the Barman server timezone. This is a
	behavior change, because previously the specified time was simply
	passed to PostgreSQL and thus interpreted according the PostgreSQL
	settings.

2018-05-10  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.com>

	Fix target_action handling when not specified

2018-05-10  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Make DataTransferFailure.from_command_error() more resilient
	Closes: #86

2018-05-09  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Remove workaround for gforcada/flake8-isort#9

	Fix remote get_file_content method
	Recover was always failing to pull .barman-recover.info because of a
	typo in remote get_file_content method.

	Closes: #151

2018-05-09  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.com>

	Ensure item_class is specified for resources to copy
	Add an assertion to the RsyncCopyController item preventing the code to
	create items without item_class set.

2018-05-08  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Suggest stricter `archive_command` in documentation
	Add an example of stricter `archive_command` that checks that
	it is executed only on a specific server. Suggested by
	Florent Xicluna (https://github.com/florentx)

	Closes #11

2018-05-08  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.com>

	Add '--standby-mode' option to barman recover
	This option is only available for PostgreSQL older than 9.0 and will
	make 'barman recover' generate the recovery.conf file even if the
	get-wal and PITR are not active.

2018-01-12  Todd Seidelmann  <seidelma@wharton.upenn.edu>

	Add hook script for recovery
	The list of new hook scripts is:

	* pre_recovery_script
	* pre_recovery_retry_script
	* post_recovery_retry_script
	* post_recovery_script

	Closes GitHub #137

2018-05-08  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add hook script for wal deletion
	The list of new hook scripts is:

	* pre_wal_delete_script
	* pre_wal_delete_retry_script
	* post_wal_delete_retry_script
	* post_wal_delete_script

2018-05-07  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add hook script for delete command
	The list of new hook scripts is:

	* pre_delete_script
	* pre_delete_retry_script
	* post_delete_retry_script
	* post_delete_script

2018-05-08  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.com>

	Add '--target-action' option to barman recover
	The option is only active when Barman creates a recovery.conf suitable
	for PITR. Possible values are:

	- None (default, for back portability)
	- shutdown (for PostgreSQL 9.5+)
	- pause (for PostgreSQL 9.1+)
	- promote (for PostgreSQL 9.5+)

2017-08-06  Reto Zingg  <g.d0b3rm4n@gmail.com>

	rename command() to barman_command()
	Fixes #118

2017-07-22  Matthew Hawn  <steamraven@yahoo.com>

	Initialise synchronous standby names list if not set
	Fixes #111

2018-05-07  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Documentation: add limitations about .partial files and recovery
	Closes Github #79

	Documentation: remind that recovery requires a shutdown Postgres
	Closes GitHub #96

2018-01-15  Feike Steenbergen  <feike.steenbergen@adyen.com>

	Correct placeholders ordering
	The number of backups on disk and the configured redundancy
	were reported in the wrong order, causing some confusion.

	Closes GitHub #138

2017-06-28  Ian Barwick  <ian@2ndquadrant.com>

	Add recovery completed message

2016-10-21  Christoph Moench-Tegeder  <christoph@2ndquadrant.de>

	force the datestyle to iso for replication connections
	this works around a bug in some psycopg2 versions

2018-05-07  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Drop support for Python 3.3
	Python 3.3 is EOL since 2017-09-29.

2018-05-04  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fix 2.6 compatibility
	Fix version of pyton-dateutils to version < 2.7.0, as more recent
	versions dropped support for py 2.6

2018-05-04  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Fix misleading message about PITR when having just `get-wal`

	Returns error if `delete` command does not remove the backup

2018-03-18  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Improve max_wal_senders documentation

2018-03-02  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Enable more flake8 tests and update the copyright year

2018-02-13  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Better way to pin wheel package to support testing on python 2.6
	Using setup_requires instead of install_requires we avoid depending
	on wheel at runtime

2018-02-09  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix missing documentation in MANIFEST.in

2017-12-19  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Improve documentation for pg_receivewal/pg_receivexlog

2018-02-06  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Pin wheel package on python 2.6 to avoid build failures

	Fix error reported by latest version of flake8-isort

2018-02-02  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Split manpages generation code to ease the management.
	With this patch we also support using pandoc 2.x to generate the
	documentation.

2018-01-15  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Ignore vanished files in streaming directory
	A file could get renamed by pg_receivewal while Barman
	reads the content of the streaming directory, resulting in an error
	message in the logs. With this patch, Barman stops treating it as an
	error, and ignores it instead.

2018-01-10  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Disable minimal tox env in travis-ci
	The minimal tox env requires psycopg2==2.4.2, but that version does
	not compile anymore in the travis-ci environment because it does not
	understand the libpq version returned by PostgreSQL 10.

2017-12-12  Jonathan Battiato  <jonathan.battiato@2ndquadrant.it>

	Fix exception when calling is_power_of_two(None)
	The function `is_power_of_two(number)` must return None
	if the `number` argument is None, instead of
	printing a Traceback.

2017-12-04  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Safer handling of types in FieldListFile class

2017-11-29  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Require pytest < 3.3.0 as we are still support Python 2.6

2017-11-29  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add '--wal-method=none' when pg_basebackup >= 10
	Fixes #133

2017-10-03  Flavio Curella  <flavio.curella@gmail.com>

	fix typo in hooks docs

2017-11-15  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Stop process manager module from ovewriting lock files content
	Previously a bug in the process manager code was writing its own PID
	in all the existing lock files which didn't have an owner.

	Due to a bug, the content of all the lock files without owner
	was overwritten with the PID of every barman command interacting with
	a server object.

	Because of this barman cron was wrongly identifying other commands
	as a running `receive-wal` process and then terminating them.

	The code has now changed to avoid modifying the content of the lock
	file when accessed by the process manager.

2017-10-26  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix some stylistic issues raised by flake8

2017-10-18  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add Postgres 10 to feature matrix (docs)

2017-09-27  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Relax the rules for rsync output parsing
	In rare cases the rsync error message about vanished files is emitted
	from the receiver process instead of from the generator.
	Furthermore recent versions of rsync have changed the capitalization
	of error messages, so the regular expression Barman uses to parse the
	output has been generalized and made case insensitive.

	Set version to 2.4a1

2017-09-04  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Set version to 2.3

2017-09-01  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Support `path_prefix` for PostgreSQL 10
	With PostgreSQL 10, `pg_receivexlog` has been renamed to
	`pg_receivewal`. Barman will try both and will pick `pg_receivewal` in
	case it find both of them.

	The user can override this behavior by using the `path_prefix` setting
	and prepending a directory containing only the desired command
	version.

2017-08-30  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Improve naming consistency with PostgreSQL 10
	The `switch-xlog` command has been renamed to `switch-wal`.

	In commands output, the `xlog` word has been changed to `WAL` and
	`location` has been changed to `LSN` when appropriate.

	Improve documentation about recovery target

2017-08-29  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Fix typo in doc for createuser

2017-08-26  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add `--target-immediate` option to `recover` command
	Throught the `--target-immediate` option, Barman writes the
	`recovery_target = 'immediate'` line to recovery.conf.
	By doing so, PostgreSQL will exit from recovery mode as soon as a
	consistent state is reached (end of the base backup).

	Only available for PostgreSQL 9.4 or newer.

2017-08-25  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Show cluster state with `barman status`
	Similarly to `pg_controldata`, the `status` command now
	displays the cluster state of a PostgreSQL server ('in production'
	or 'in archive recovery').

2017-08-23  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Support PostgreSQL 10 in `barman replication-status` command

	Invoke pg_basebackup with --no-slot option in PostgreSQL 10+
	Starting from PostgreSQL 10 pg_basebackup uses a temporary
	replication slot unless explicitly instructed.

	Barman don't need it because it already stores the full
	WAL stream

2017-08-02  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix exception during exclusive backup with PostgreSQL 10

2017-08-01  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add `--network-compression` option to remote recovery
	Enable/disable network compression at run-time to remote recovery
	with `--network-compression` and `--no-network-compression` options.

2017-08-01  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Describe the actual error when invoking archiving hooks script
	In case of error the `BARMAN_ERROR` variable will contain a string
	starting with the name of the actual WALFileException subclass.

2017-07-28  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Ignore `.tmp` files in `incoming` and `streaming` directories
	The pg_receivewal utility occasionally writes to a `.tmp` file and
	then renames it to the final name to ensure an atomic creation.
	Barman must ignore this temporary file.
	Other programs could do the same, so Barman will ignore any file name
	with a `.tmp` suffix in the incoming and streaming directories.

2017-07-20  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.com>

	Add support to PostgreSQL 10
	PostgreSQL 10 renames SQL functions, tools and options that references
	`xlog` to `wal`. Also WAL related functions referencing `location` have
	been renamed to `lsn`.

	That change have been made to improve consistency of terminology.

	This patch improves Barman adding support to PostgreSQL 10.

2017-07-26  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Make flake8 happy again

	Improve `barman diagnose` robustness

2017-07-22  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix high memory usage with parallel_jobs > 1
	This patch avoids bloating the memory with objects duplicated by
	pickle/unpickle cycles.

	Fixes #116

2017-07-21  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Improve travis integration

	Fix unit test for FileWalArchiver to work with Python 3.6
	Python 3.6 is stricter about what the `os.path.join` function can
	receive as input. This change has highlighted missing prerequisite
	setting in the FileWalArchiver.archive test.

	Set version 2.3a1

	Fix exception in `barman get-wal --peek` requiring history files

	Fix exception during check if ssh works and conninfo is invalid

	Make sure that parallel workers are cleaned up in case of errors
	In CopyController code, at the end of parallel copy, it is necessary
	to terminate all the parallel workers, whether the copy finishes
	naturally or it is terminated by an error.

	Closes: #114

2017-07-17  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

2017-07-14  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Set version to 2.2

	Update copyright

2017-07-12  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Log get-wal source host when called through SSH

2017-07-06  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.com>

	Check the number of files in the archiving queue
	Add a new configuration parameter, named `max_incoming_wals_queue`, that
	enables a new check for the number of files in the incoming directories.

	This check will fail if the number of WALs to be archived
	goes over the allowed threshold.

	This check does not block normal operations.

	The purpose of this check is to improve robustness of the Barman
	system when integrated in an alerting/monitoring infrastructure,
	by warning users in case Barman is not archiving fast enough.

2017-07-10  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix missing mock in test_pg_stat_archiver_status

2017-07-11  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.com>

	Show-backup fails on incomplete backups
	The barman `show-backup` command was failing on failed backups
	with a stacktrace. Fixed unit tests as well as the bug.

2017-07-05  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update copy_controller unit tests

2017-07-04  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Improve copy output messages

2017-07-03  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.com>

	Support dynamic WAL size
	Barman now asks PostgreSQL for the current WAL file size at every start of
	backup. This information is stored in the backup info and then used while
	showing statistics.

	Previously the size of WAL files was hardcoded into Barman, giving wrong
	statistics when PostgreSQL was compiled with a different value.

	IMPORTANT: Release notes must mention that the format of the backup info
	file has changed. A backup made with Barman 2.2 cannot be read by a
	previous version of Barman (in the rare event that a user downgrades
	version).

2017-07-05  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.com>

	Make cuncurrent backup killable
	Barman was able to recover from a KeyboardInterrupt while
	executing a backup, so you were not able to interrupt a backup anymore.

	This patch disable the handling of the SIGINT signal from the worker processes.
	The signal is now handled by the parent process, which is responsible
	for the termination of the children processes.
	The backup will be then stopped in the usual way.

2017-07-04  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Allow replication-status to work against a standby

2017-07-04  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Close all PostgreSQL connections before starting pg_basebackup
	With this commit we avoid keeping an idle transaction during the file
	copy and fix incorrect end time reported at the end of the backup.

	Closes: #108, #104

	Fix exception in show-backup with postgres backup_method

2017-06-01  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Safely handle paths containing special characters
	This patch makes Barman correctly handle path containing those
	characters having special meaning to the Unix shells such as quotes,
	double quotes, spaces, and so on.

2017-06-12  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Improve backup completed message

2017-06-05  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Display statistics about the copy gathered during backup
	The statistics will be displayed in `barman show-backup` output as
	'Copy time' and 'Estimated throughput' values.

	Fix an error message interpolation

	Make the method rsync_factory private
	The method rsync_factory should be private, as it must be used only from the
	copy controller.

2017-05-29  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Ensure incoming directory is empty when `archiver=off`
	This commit adds a check that warns the user when there are files in
	the incoming directory but the `archiver` option is disabled.

	This is something that could happen when the archiver is disabled in
	favor of the `streaming_archiver`, but PostgreSQL is still configured to
	ship WAL files to the incoming directory.

	This check is a non blocking check with the purpose of warning the
	user about the disk filling up with unarchived WALs.

	Closes: #80

2017-05-25  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add `external_configuration` to `backup_options`
	It is now possible to instruct Barman that we are aware that
	configuration files reside outside of PGDATA, through the
	`external_configuration` value for the `backup_options` option.
	By doing so, users will suppress any warning regarding external
	configuration files during backup.

2017-05-23  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Command line options to enable/disable `get-wal` usage during recovery
	Add two options `--get-wal` and `--no-get-wal` to the `barman recover`
	command. They will control the usage of the `get-wal` option during
	the recovery, overriding what is specified in the configuration file.

2017-05-26  William Ivanski  <william.ivanski@2ndquadrant.com>

	Documentation on parallel copy
	Updated documentation and man pages regarding parallel
	copy feature for both backup and recovery.

2017-05-22  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Archive .partial files coming from promotion of streaming source
	Change the way Barman handles multiple partial files during
	the archiving process.

	Archive older partial files as normal files, instead of aborting.

	Fix .gitignore syntax under sphinx directory

2017-05-12  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Update TODO

2017-04-03  Mitchel Humpherys  <mitch@directangular.com>

	Document how to stop `cron`
	It's not immediately clear from the docs how to stop all of the processes
	that get started by the `cron` command.  Add some instructions to the
	section on `cron` documenting how to stop things properly.

	Fixes #97

	Documentation improvement: clarification about cron
	It can be a bit surprising to discover that barman doesn't have any sort of
	daemon program, as many long-running services do.  Make this clear up front
	to avoid confusion.

2017-03-20  Cédric Villemain  <cedric@2ndquadrant.fr>

	Documentation improvements

2017-05-05  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Ensure that PGCONTROL_CLASS items are copied as last step

2017-05-04  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Implement parallel copy for backup/recovery
	When `parallel_jobs` configuration option (or the `--jobs` run-time
	option) is set to a value greater than 1, the files to be copied
	are equally distributed between a number of buckets equal to the
	number of workers.

	The current maximum size of a bucket is 10GB.

2017-05-03  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	In RsyncCopyController use an in-memory file_list instead of a file
	This provides enough information to the _generate_job to implement a
	job sharding strategy in a future commit.

2017-05-02  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Use a process pool to execute backup and recovery
	Add initial multiprocessing infrastructure to support parallel backup
	and recovery using rsync.

	Added the 'parallel_jobs' global/server configuration option,
	that specifies the number of parallel processes used for both
	backup and recovery operations. Default value is 1 (for
	retro-compatibility).

	Added also the `--jobs` and `-j` run-time option for the `backup`
	and `recover` commands to override the server behaviour.

	At the moment, when the value is set to 2 or more, copy operations
	are parallelized, but the logic to split them in smaller chunks is not
	yet implemented.

2017-04-18  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Recursively create directories during recovery
	Closes: SF#44

2017-04-05  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Reduce time ServerXLOGDBLock is held while archiving
	The ServerXLOGDBLock was held from the start to the end of the
	archiving process, but that was not how it was supposed to be.

	The implicit contract of a waiting lock such as the ServerXLOGDBLock
	requires that the lock is kept for the shortest possible time.
	However, that is not the case when archiving a large number of WAL
	files or when executing the pre/post archive hooks.

	This commit changes the acquisition strategy to acquire and
	release the lock on a per-WAL file basis.

	Fixes #99

2017-03-31  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix a docstring typo

2017-03-31  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Remove `tablespace_map` file during `recover`
	Barman takes care of tablespaces relocation during `recover` by creating
	the appropriate links into `pg_tblspc` directory. The eventual
	presence of `tablespace_map` file would instruct PostgreSQL to
	rollback these changes, so it is mandatory to remove this file during
	`recover`.

	For the same reason the code that was writing the `tablespace_map`
	file during `backup` was disabled.

	Fixes #95

2017-03-09  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Improve log messages when a backup fails

2017-02-27  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Improve log message for streaming connections
	Add the server name in log messages when opening
	a streaming connection to PostgreSQL fails.

	Reported by Andreas Kretschmer <andreas.kretschmer@2ndquadrant.com>

2017-02-06  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Properly detect broken Postgres connections
	When a Postgres cursor is required, issue a 'SELECT 1'
	query to check that connection is properly working.

2017-02-01  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Cosmetic change in documentation

2017-02-01  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Reconnect to PostgreSQL if the connection drops
	Now Barman will try to reconnect automatically if the connection to
	PostgreSQL drops.

	Closes: SF#82

2017-01-20  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add a init_check method to CheckStrategy
	Add a init_check method to the CheckStrategy object, the method
	will help keeping track of the check execution and produce a clear
	output in case of timeout failure.

	Set version 2.2a1

2017-01-05  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Update the ChangeLog file

	Set version to 2.1

2016-12-22  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Avoid importing barman.utils module twice

2017-01-03  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Make retry hook script options global (bugfix)

2016-12-27  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

2016-12-26  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Release 2.1a1

2016-12-21  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add PostgreSQL 10 versioning scheme compatibility
	Fix #73

	Exclude extraneous files during tablespace copy
	PostgreSQL stores the tablespace data in a directory defined in the
	PostgreSQL source as:

	    "PG_" PG_MAJORVERSION "_" CATALOG_VERSION_NO

	The CATALOG_VERSION_NO is not easy to retrieve, so we copy

	    "PG_" + PG_MAJORVERSION + "_*"

	It could select some spurious directory if a development or a beta
	version have been used, but it's good enough for a production system
	as it filters out other major versions.

	Fix #74

2016-12-20  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Align the list of excluded files to pg_basebackup
	Fixes #72

2016-12-18  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Return failure when 'barman get-wal' command is used on inactive servers
	The 'barman get-wal' command is meant to be used used as PostgreSQL's
	restore_command, so it must succeed only if the target WAL file has
	been actually returned.

2016-12-17  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add last archived WAL info to diagnose output
	Show in diagnose the last archived WAL per timeline

	switch-xlog --archive waits for the closed WAL file
	Modify the switch-xlog --archive behaviour to wait for the specific WAL
	file, instead of a generic one. This is faster and it avoids possible
	issues when a higher timeline is present in the archive.

2016-12-03  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add --archive and --archive-timeout options to switch-xlog command
	Improve the switch-xlog command adding the --archive and
	--archive-timeout options.

	The --archive option will start an archiver process for the selected
	server, and will wait for 30 seconds for a new xlog to be archived.

	The --archive-timeout TIMEOUT will allow the user to increase or
	decrease the number of seconds the archiver process will wait
	for the archival of a new xlog file, before exiting.

2016-11-26  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Make streaming application name options global
	Make both 'streaming_archiver_name' and 'streaming_backup_name'
	global options, as per documentation. Bugfix.

	Fixes #57

2016-11-27  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix rsync failures due to files truncated during transfer
	Fixes #64

2016-11-26  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Fix typo in documentation

2016-11-17  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Convert README in README.rst
	Suggested by ckeeney. Closes #69

2016-11-24  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Correctly handle compressed history files
	Fix how barman behave when a compressed history files is found.

	Fixes: #66

	Avoid dereferencing symlinks in pg_tblspc when preparing recovery
	This commit fixes a bug triggered by the presence of a broken tablespace
	symlink in the target pg_tblspc directory when recovery starts.

	Fixes: #55

2016-10-06  Martin Swiech  <martin.swiech@zonky.cz>

	Fix comparation of last archiving failure
	Compare the time of last failure instead of the WAL name to detect
	archiving failures.

	Closes: #40, #58

2016-11-13  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Avoid failing recovery if postgresql.conf is not writable
	Fixes: #68

2016-11-08  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add start time and execution time to delete output
	Improve the output of `barman delete` command by adding some execution
	time statistics.

	Additionally, the utility function that prints human readable time deltas
	has been improved to correctly handle a wider range of values.

	Closes #67

2016-10-26  Ian Barwick  <ian@2ndquadrant.com>

	Change references from non-existent `path` option to `path_prefix`

2016-10-28  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Exclude files within pg_replslot from backups
	Thanks to Magnus Hagander for reporting the issue.

	Ref: https://www.postgresql.org/docs/current/static/continuous-archiving.html#BACKUP-LOWLEVEL-BASE-BACKUP-DATA

	Fixes #65

2016-10-20  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Workaroud to setuptools testrunner bug.
	Ref links:

	* https://bitbucket.org/pypa/setuptools/issue/196
	* https://github.com/pypa/setuptools/issues/196

2016-10-17  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fix output of the replication-status command.
	The replication-status command used to emit an exception in some
	PostgreSQL releases, due to the different structure of the
	`pg_stat_replication` view.

	This patch fixes the issue checking for the existence of the fields
	before writing them to the output.

	Fixes #56

2016-10-11  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Move isort configuration in setup.cfg

2016-10-11  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add a note in docs about concurrent backup for 9.6

2016-09-30  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Split the logic of the copy method in CopyController
	This patch creates an analisys phase and a copy phase. the analisys
	phase is responsible for the preparation of the list of files that we
	want to copy. The copy phase is responsible for the execution of the
	actual copy using rsync.

2016-10-02  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Fix level of section about WAL verification

2016-09-27  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Update sphinx index page

2016-09-26  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Set version 2.1a1

2016-09-26  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

2016-09-26  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Update architecture diagrams

2016-09-26  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Review of example configurations in doc/barman.d/

2016-09-26  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Fix image size in manual

2016-09-26  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Set version to 2.0

2016-09-25  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	General review of the manual before 2.0 release

2016-09-23  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add section `Upgrade from Barman 1.X` to manual.

2016-09-23  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Improve documentation of get-wal command
	This patch improves the documentation of the get-wal command for the
	local recovery, specifying that the command must be executed as `barman`
	user and not as the user which is running the PostgreSQL server.
	It also includes details about the barman-wal-restore command.

2016-09-23  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Update INSTALL, README and TODO files

2016-09-23  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Add Windows server documentation
	Add a Windows server section in the "Setup a new server in Barman"
	chapter.

2016-09-21  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Fix wrongly named max_replication_slots configuration parameter

2016-09-20  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

2016-09-19  csawyerYumaed  <csawyer@yumaed.org>

	Improve barman(5) manual page, adding conninfo
	Add additional information to documentation around conninfo.

	Closes #53

2016-09-20  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Set version to 2.0b1

	Remove the quickstart guide
	At the moment, it is an empty document. It may be reintroduced later.

2016-09-20  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Improve configurantion error message
	The configuration error message raised when a parameter has an invalid
	value was not telling the wrong value but only the key.

	This patch improves the error message in that case.

2016-09-16  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add troubleshooting section to manual

2016-09-16  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Improve "Features in detail" section

2016-09-19  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Avoid logging tracebacks in case of expected failures
	There was a few places in the code where an expected error was producing
	a traceback in the logs. They have been replaced with a simple error
	output.

2016-09-16  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Integrate barman-wal-restore in barman recovery
	During a remote restore, use the barman-wal-restore command instead of
	calling `barman get-wal` on the remote machine. It requires the
	barman-cli package to be installed on the target server.

	Force special tox environments to use a specific Python version
	Use Python 2.7 in the 'flake8' environment, because it doesn't work with
	Python 2.6.

	Use Python 2.6 in the 'minimal' envorinment, because it doesn't work
	with Python 3.x due to the presence of Python-2-only modules.

	Enable the minimal target in Travis-CI again, as it was failing due to
	the default Python version used there (Python 3.5).

2016-09-19  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix wrongly named database parameter in configuration examples
	Thanks to Abhijit Menon-Sen <ams@2ndQuadrant.com> who found it.

2016-09-14  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Add "server commands" documentation section
	This patch complete the section "Server commands" inside the Barman manual

2016-09-14  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Backup command section of the manual

2016-09-14  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Add the "general commands" documentation section
	This patch complete the section "general commands" inside the Barman
	manual.

2016-09-13  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Add the "setup new server" documentation section
	This patch complete the work on the "Setup of a new server in Barman"
	documentation section.

2016-09-13  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Make sure 'dbname' parameter is not set for streaming connections
	Also fix an issue in the parse_dsn method, many thanks to
	Théophile Helleboid <chtitux@gmail.com> for reporting it.

2016-09-15  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix flake8 plugin versions

2016-09-13  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Make sure path_prefix is honored in every Command subclass
	Commit ad8002745326fe20a2c4ce2a5e9115a26c5a5f49 contains a typo
	so it was not working at all. The typo is fixed in this commit.

2016-09-13  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Pass `--no-password` option to pg_basebackup and pg_receivexlog
	Barman will never send any password through the pg_basebackup or
	pg_receivexlog standard input channel. Invoking them with the
	`--no-password` option will produce a cleaner error message in case of
	authentication failure.

	Modify tests accordingly.

2016-09-12  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Improve the warning message for configuration files outside PGDATA
	Now the `backup_method=postgres` code emits the warning message for
	configuration files that are located outside PGDATA only when that's
	the actual case.

2016-09-12  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Moved barman-wal-restore in barman-cli project

2016-09-12  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Enable `show-server` command on inactive or disabled servers

2016-09-09  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add missing release date on manual front page

	Disable minimal tox env on travis-ci as it doesn't work on py35

	Update the ChangeLog file

	Prepare release 2.0a1

	Run minimal and flake8 tox envs in travis-ci

2016-09-09  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Fix flake8 error

2016-08-26  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	First stub of documentation for 2.0
	This is a first attempt to change the documentation in Barman,
	which creates two artifacts: a manual and a quickstart guide.

	The work is currently incomplete, but it is ok for a first
	alpha release, given that all required information from a
	technical point of view is in man pages.

	The reason for this is that now Barman does not have a sequential
	installation procedure, but different options for backup and
	WAL archiving that produce several possibile combinations of
	setups.

2016-09-09  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Set last supported version by argh for Python 2.x
	Add minimal checks in Tox

2016-09-09  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Release notes for 2.0a1

2016-09-09  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Rename barman-wal-restore.py to barman-wal-restore

2016-09-09  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Fix exception in barman check
	Barman check raise an exception when streaming_archiver is on and
	streaming_conninfo is not configured.

	This patch controls if the streaming connection is correctly configured
	before checking the pg_basebackup compatibility, removing the exception
	cause.

2016-09-09  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Remove unit test dependency on environment
	Previously unit tests were failing the $USER environment variable was
	not defined. This patch removes this requirement.

2016-09-08  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Make sure path is used in every Command class

	Switch xlog after a concurrent backup
	For a backup to be usable, all the XLOG files written between the backup
	start and the backup end must be archived.

	This patch makes PostgreSQL switch the current XLOG file after a
	concurrent backup either using rsync or streaming backup mode, in this
	way the WAL file containing the backup end position will be archived as
	soon as possible.

2016-09-07  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add retry support when `backup_method=postgres`
	Modify the backup executor to handle the retry of a streaming
	backup with pg_basebackup (`backup_method=postgres`).
	Between every retry attempt, the PGDATA and the directories that
	contain the tablespaces are removed. Therefore, subsequent
	backup attempts restart from scratch.

2016-09-07  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Fix E303 too many blank lines (flake8)

2016-09-07  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Support /etc/barman.d in rpm packaging

2016-09-01  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Redesign retry logic for backup and recovery
	The CommandWrapper class is now responsible to specifically
	design the logic of a retry operation in case of failure.

	Previously, retry was managed at an outer level, by the
	backup manager and the executor classes.

	With this new approach, every single command can now develop its
	own retry logic.

2016-09-05  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Force `archiver=on` when no archiver is active.
	This is a back-compatibility feature that will be kept
	in Barman for a few releases, and mitigate the upgrade
	to version 2.0 (which requires explicit setting of
	an archiving method - before, `archiver` was `on` by default)

2016-09-05  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Change 'wal_level' to 'replica' in unit tests

2016-08-29  Ian Barwick  <ian@2ndquadrant.com>

	Enable barman-wal-restore.py to pass -c/--config to Barman
	This enables barman-wal-restore.py to be used if the Barman
	server uses a non-default configuration file.

2016-09-02  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add slot name to `replication-status` (PostgreSQL 9.5+)

	Handle PostgresConnectionError exception differently
	Changed the way PostgresConnectionError works, in order
	to be more easily used in 'check' methods for both
	standard and streaming connection purposes.

	Specified also a new exception (PostgresAppNameError).

2016-09-01  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Remove any reference to Rsync.smart_copy()
	Remove the smart copy logic from the command wrappers module,
	as it is now part of RsyncCopyController.

	Support 9.6 syntax for `synchronous_standby_names`

2016-09-01  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Cosmetic changes in recovery_executor.py
	Renamed also some methods as they are private

2016-08-30  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Introduce the CopyController module for recovery
	This patch introduces the use of the RsyncCopyController
	inside the recovery_executor.py module.

2016-08-31  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add Config.get_bwlimit() public method
	Move the private method _bwlimit() from RsyncBackupExecutor
	class into Config, and rename it into get_bwlimit().

2016-08-30  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add status() method for archiver classes

2016-08-30  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add StreamingWalArchiver._is_synchronous()
	Make code clearer and easier to read by encapsulating
	checks for synchronous replication in a separate method
	of the StreamingWalArchiver class. Make code more resilient
	by checking that PostgreSQL connection is working.

2016-08-25  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Introduce the CopyController module for backup
	This patch adds a new module called copy_controller.py.

	The new module is responsible for containing the logic of
	a physical copy of for a backup, starting with basic
	rsync support.

	This patch is a refactor of the existing code, with
	no additional functionality yet. Only backup executor
	is using this class for now.

2016-08-30  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix exception in 'check' when postgres is unreachable
	The check method of PostgresBackupExecutor and ConcurrentBackupStrategy
	objects were raising an exception if PostgreSQL connection was not
	working.

	This commit fixes this behaviour by correctly handling this case.

2016-08-26  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Set default value of 'archiver' to 'off'
	This is a milestone in Barman's history as it
	removes the need for standard WAL archiving through
	the `archive_command` configuration option in
	PostgreSQL (even though it is still recommended
	to use it in conjunction with WAL streaming, for
	more robust architectures).

	**IMPORTANT**: When upgrading from 1.X.X versions
	of Barman to 2.X versions, users must explicitly add
	`archiver = on` (either globally or at server level).

	Created configuration templates for streaming and Ssh use cases

2016-08-24  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Enable streaming-only archiver with replication slots
	Remove the previous restriction that required standard WAL archiver
	(`archiver=on`) to be always present. Allow a server to operate
	with just WAL streaming if a replication slot is set (PostgreSQL
	9.4 or higher is required). Replication slots make sure that every
	WAL file is streamed to the Barman server.

	Basic checks have been implemented at configuration level, requiring
	`streaming_archiver=on` and a `slot_name` if `archiver=off`.

2016-08-25  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Improve replication slot error messages
	Barman didn't catch connection related exceptions in the
	`receive-wal` subcommand while creating and deleting a replication
	slot.

	This patch fixes that and reports to the user the connection error,
	avoiding to print a stack trace.

	This commit also adds a specific error message in case of drop of a
	replication slot that is currently in use.

2016-08-23  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Manage 'all replication slots are in use' error during slot creation
	Display an informative message to the user instead of a raw exception
	when the PostgreSQL's max_replication_slots setting is too low.

2016-08-23  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Improve replication slot related messages
	This patch improves the replicaiton slot creation and drop messages
	quoting the names of the servers and of the replication slots.

2016-08-23  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fix error messages during create and drop of replication slots
	Modified error message emitted in case of an already existing
	replication slot during the execution of receive-wal --create-slot.  the
	message have been changed to error from info, replicating the behaviour
	of pg_receivexlog. Now, in case on error, the barman process exit code
	will be 1.

	The error message emitted during the execution of the receive-wal
	--drop-slot in case of a not existing replication slot have been changed
	to error as well.

2016-08-23  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Set version to 2.0 alpha 1
	Anticipating PostgreSQL's new versioning system kicking in
	from version 10.0, development team has decided to adopt the same
	policy to Barman's versioning system starting from version 2.0.

	The development team has also decided that this version deserves
	to be 2.0 due to the scope of the new features that will be introduced,
	in particular streaming-only backup and "zero data loss" backups.

2016-08-23  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Improve error messages for receive-wal subcommand
	Error messages for the receive-wal subcommands were incoherent and this
	patch improves them, using a common form.

	In addition to that, the availability of the 'slot_name' configuration
	parameter was checked before the server version, making that check
	useless if the PostgreSQL server version has no support for replication
	slots.

2016-08-20  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Support zero data loss backup (synchronous WAL streaming)
	Automatically start `pg_receivexlog` with `--synchronous`
	if the following conditions are met:

	- WAL streaming is active (see `streaming_archiver` and
	  `streaming_conninfo`)
	- the WAL streamer application name (see `streaming_archiver_name`)
	  is in the `synchronous_standby_names` priority list of the PostgreSQL
	  server (which defines the synchronous standby servers)
	- PostgreSQL server 9.4 or higher
	- pg_receivexlog 9.5 or higher

	This feature allows a Barman backup solution to act like a
	synchronous standby server in a PostgreSQL business continuity
	cluster, bringing RPO=0 backups (zero data loss).
	Currently, RPO=0 requires a replication slot to be properly configured
	(see `slot_name`).

2016-08-23  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fix exception during `barman diagnose`
	Barman diagnose was trying to serialize a LooseVersion object to the
	JSON stream, which is not serializable by default.

	This patch add the support for (Loose|Strict)Version objects to the
	BarmanEncoder object treting them as a string.

2016-08-22  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fix archiver by catching ENOENT exception
	Fixed a corner-case error that could occour when an unexpected
	file (e.g. .partial file) was not found anymore by the archiver
	process while trying to move it in the errors directory.

	This commit also makes logging messages easier to read.

2016-08-17  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add support for replication slots in `receive-wal` command
	If `slot_name` option is set and `streaming_archiver` is enabled, the
	`receive-wal` command will use the given replication slot to ensure
	that no WAL file is lost before its archival.

	The `check` command will monitor the status of the configured slot
	on the PostgreSQL side, accordingly with the value of the
	`streaming_archiver` option: if `streaming_archiver` is on, it will make
	sure that the slot exists and it is used, otherwise it will warn the
	user in case a stale slot is found.

2016-08-19  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fix directory permissions when backup_method=postgres
	This patch fixes how pg_basebackup manages the permissions of
	the destination directory of a backup on the barman server,
	forcing 0700 at creation time.
	The same rights are applied to the destination directory of every
	tablespace copied.

	Fix misleading message on pg_receivexlog termination
	In case of termination of the pg_receivexlog subprocess using a signal,
	a misleading entry was written in the log file.

	This patch improves the error message.

2016-08-13  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Introduce batch size for WAL archiving
	Previously, the `archive-wal` command would scan the
	incoming directories for both standard and streaming archiving
	and then process all the available WAL files in one single
	run.

	While this remains the default behaviour, two global/server
	options have been added to Barman: `archiver_batch_size` and
	`streaming_archiver_batch_size`.

	The first controls the batch size for the `archiver` process,
	the latter for the `streaming_archiver` process.
	The default value is 0, meaning unlimited batch (traditional
	behaviour in Barman). A value greater than 0 would trigger
	batch processing of WAL archiving in Barman, limiting a single
	run of `archive-wal` for the specific incoming queue (file archival
	or streaming) to that value.

	For example, setting `archiver_batch_size` to `10` would allow
	the `archive_wal` command to stop after processing a maximum of
	10 WAL segments from `archive_command` in the same run (assuming
	`archiver` is enabled). In those contexts where `streaming_archiver`
	is enabled, setting `streaming_archiver_batch_size` to `10` would
	reach the same goal for WAL segments that have been streamed to
	Barman.

	These settings should be tuned based on the server's workload
	and the frequency of the `barman cron`. A more distributed
	WAL processing certainly reduces the probability of xlog.db
	locking (be careful not to set these value too low - the risk
	is that your Barman server stays behind the server's WAL production).

	Renamed WalArchiverBatch class into a more appropriate WalArchiverQueue.

2016-08-02  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Add to receive-wal the '--drop-slot' option
	This patch adds to the 'receive-wal' Barman subcommand a '--drop-slot'
	option. That option drop the replication slot named in the 'slot_name'
	configuration parameter.

	The patch also adds the relative documentation to the Barman man
	page in the section 1.

	TODO: section in the tutorial that explains replication slots

2016-07-30  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Add to receive-wal the '--create-slot' option
	This patch adds to the 'receive-wal' Barman subcommand a
	'--create-slot' option that creates a replication slot using the
	PostgreSQL streaming connection.

	The patch also adds the relative documentation to the Barman man page
	in the section 1.

	TODO: section in the tutorial that explains replication slots

2016-08-12  Leonardo Cecchi  <leonardo.cecchi@2ndQuadrant.it>

	Correctly handle history files from streaming
	Since 9.3, `pg_receivexlog` follows timeline switches and creates
	appropriate `.history` files when needed. Previous implementation
	in Barman of WAL archiving from streaming connections erroneously
	skipped history files and treated them as errors.
	This patch fixes this behaviour.

2016-08-04  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Avoid to lock XLOGDB while checking if WAL archiving is set up
	The patch avoid to take a lock during the checking of the WAL
	archiving setup.

	This is important for invocations of `barman check` to work correctly
	even when there are long activities on the XLOGDB file, i.e. when the
	archive process is managing many WAL files.

2016-07-29  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fix flake8 version for Unit Tests

2016-07-28  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Check tablespace-mapping option in pg_basebackup
	With `postgres` backup method, tablespaces are supported only
	for PostgreSQL servers 9.3 or higher and pg_basebackup 9.4 or higher.

	The reason is that pg_basebackup supports the `tablespace-mapping`
	option only from 9.4, and this option is needed in case there are
	tablespaces in the server. Given that pg_basebackup 9.4 is not
	compatible with PostgreSQL 9.1 and 9.2, these versions are not
	supported in case of tablespace presence.

	This patch adds a check for this scenario, providing users with
	warnings and preventing them from starting a backup in case
	the aforementioned requirements are not met.

	Documentation has been modified accordingly.

2016-07-19  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add the `slot_name` configuration parameter
	Setting the `slot_name` parameter on a server running at least
	PostgreSQL 9.4 will enable the usage of a physical replication slot
	to track the position of the `receive-wal` process.

2016-07-19  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Cosmetic changes in barman-wal-restore.py help page

2016-07-14  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add a check for spurious messages in ssh_command output
	The aim of this check is to early detect potential communication
	problems through the configured ssh_command.

	For example, if the login script on the PostgreSQL server outputs
	any messages, this additional output could confuse Barman and
	therefore trigger random backup failures.

2016-07-13  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Fix replication-status command on PostgreSQL 9.1
	Barman was using the `pg_xlog_location_diff` function to calculate the
	difference in bytes between the master and the standby xlog locations.
	As `pg_xlog_location_diff` is available only from PostgreSQL 9.2, the
	`replication-status` command wasn't working on PostgreSQL 9.1.

	Adds the diff_lsn function to calculate the difference between
	two xlog locations.

	Fixes #45

2016-07-07  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	9.6 backup API support improvement
	PostgreSQL 9.6 provides functions to read the control file,
	so we can avoid parsing the backup_label to know the current
	timeline.

2016-06-16  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Support backup from standby
	This commit allows to backup standby servers using rsync or the new
	PostgreSQL 9.6 backup API.

	Allow invoking current_xlog_info on standby servers
	Previously, `pg_current_xlog_location` was used to get the current
	transaction log location, but this function can't be used on a standby
	server. Now the code checks the server type and use
	`pg_last_xlog_replay_location` if the server is in recovery mode.

	Uniform backup_label parsing
	The backup_label file was previouly parsed in multiple places of code.
	This code has been moved in a dedicated method.

2016-05-30  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add support for concurrent backups with PostgreSQL 9.6
	As of version 9.6 PostgreSQL support concurrent backups natively.
	Add support to the new API, using new calls for concurrent
	backups. Users from PostgreSQL 9.2 to 9.5 can continue using
	the pgespresso extension.

	Code has been refactored accordingly.

2016-06-14  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Introduce PostgreSQLClient base class
	Refactor the PgBaseBackup and PgReceiveXlog classes, by moving common
	code in the PostgreSQLClient base class.

	Add PgBasebackup class unit tests.

2016-06-16  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fix flake8 problem in tox
	See: https://github.com/gforcada/flake8-isort/issues/9

2016-06-13  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Refine the logic to detect unused WAL files.
	When multiple timelines are present in the archive, avoid purging
	WAL files that could be used by a backup on a different timeline.

2016-06-13  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Store detailed backup mode in backup.info file
	The `mode` field is present since long time, but before was always set
	to `default`.

	From now on, the `mode` field contains a descriptive string that shows
	the type of executor involved in the backup (rsync/postgres) and, if
	present, the strategy of the backup executor (exclusive/concurrent)

2016-06-03  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Add a warning message to show-backup checking timelines
	If after a backup there is a timeline switch, the "Last WAL" field from
	the show-backup subcommand will show the last available WAL on the
	timeline on which the backup was taken and not the last archived one.
	This behaviour, while it's correct, can confuse the user.

	Now barman handles such situation and emits a warning if there is more
	than one timeline reachable from the displayed backup.

2016-06-10  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Fix replication-status with no sent and write locations

	Fix bug with reuse_backup and backup_method=postgres

2016-06-09  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add `streaming_backup_name` option
	When `backup_method` is set to `postgres`, the `streaming_backup_name`
	option sets the `application_name` of the managed `pg_basebackup`
	connection, improving monitoring of backups taken using `pg_basebackup`.

	By default, `streaming_backup_name` is set to `barman_streaming_backup`.

	Available for PostgreSQL 9.3 and above.

2016-06-05  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Refine documentation and messages

2016-06-03  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Change psycopg2 cursor type to DictCursor
	Instruct the cursor object to use DictCursor objects to manage the
	results of the execute commands.

	Update calling methods accordingly.

2016-05-31  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Move exceptions in a dedicated module
	Exceptions have been moved in a dedicated `exceptions` module and
	organised into a hierarchy.

2016-03-30  Stefano Bianucci  <stefano.bianucci@2ndquadrant.it>

	Base backups through PostgreSQL streaming replication
	Enable 'streamrep' only base backups with `backup_method=postgres`
	configuration option (global/server). Requires a valid streaming
	replication connection.

	Barman's implementation relies on `pg_basebackup`, therefore the
	following limitations apply:

	- `bandwidth_limit` is available with `pg_basebackup` >= 9.4 only
	- `network_compression`, `reuse_backup` and `tablespace_bandwidth_limit`
	  are not supported
	- configuration files outside PGDATA are not copied

2016-05-30  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add contrib/barman-wal-restore.py
	Python version of the WAL restore script to be used as
	'restore_command' for recovery or fallback method with standby
	servers. This script invokes 'barman get-wal' through
	a secure shell channel, and supports also parallel fetching
	of WAL files (requires Barman 1.6.1+).

	Requires argparse. Supersedes barman-wal-restore for bash.

2016-05-27  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Limit the maximum execution time of 'barman check' command
	Add a configurable timeout for the execution of the 'barman check'
	command. It limits the maximum execution time to 30 seconds per server
	by default. The user can change this value using the 'check_timeout'
	configuration parameter.

2016-05-18  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add version properties to PostgreSQL base class
	Add server_txt_version and server_major_version properties to the
	PostgreSQL class to quickly retrieve the major version of the PostgreSQL
	server.

	Refactor current_xlog_* properties in PostgreSQLConnection
	Introduce the current_xlog_info property that returns detailed info
	about current xlog location.

	Rename current_xlog property as current_xlog_file_name.

	Now current_xlog_file_name and current_xlog_location properties have
	been reimplemented and now they call current_xlog_info.

	Generalise DataTransferFailure exception
	A DataTransferFailure exception is not necessarily related to rsync only.
	Make it generic by accepting the command name as argument in the
	from_command_error factory method.

	Extract _purge_unused_wal_files method from SshBackupExecutor
	For reuse purposed, the code responsible for removing unused wal files
	in the backup method of SshBackupExecutor has been extracted.
	A new method called _purge_unused_wal_files is now part of the base class.

2016-05-11  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Use UnixRemoteCommand to execute remote operations
	When execute operation on remote server use UnixRemoteCommand instead of
	manually build an ssh Command.

	Some pre-existing UnixRemoteCommand invocations have also been fixed to
	honour the path_prefix setting.

2016-05-18  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Set version to 1.6.2a1

	Update the ChangeLog file

	Prepared release 1.6.1

2016-05-16  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Clarify switch-xlog and replication-status failure output
	Change the output messages to be more clear regarding the cause of the
	failure.

2016-05-12  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Fix traceback for switch-xlog with unknown server

2016-05-09  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Avoid logging tracebacks during WAL files deletion
	During WAL files deletion, if the target file doesn't exists,
	print a WARNING message and continue.

2016-05-06  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Prepared version 1.6.1a1

2016-05-06  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Fix "PostgreSQL connection forced cleanup message" during unit tests

2016-05-06  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Modify manifest to include tutorial file in dist tar.

2016-05-06  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Rolling out 1.6.1a1

	Add "replication-status" command
	The "replication-status" command shows live information
	and status of any streaming client attached to one or more
	given servers.

	The --target option allows users to limit the scope to
	only hot standby servers ('hot-standby') or WAL streaming
	clients, such as pg_receivexlog ('wal-streamer'). By default,
	--target is set to 'all'.

2016-05-05  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Check for 'backup_label' on the master
	If a PostgreSQL server is abruptly shut down when an
	exclusive backup is in progress, the 'backup_label' file
	will stay inside PGDATA.

	In case of a failed exclusive backup, if the Postgres server
	is down, "barman check" verifies that a backup_label is present.
	If so, prints "A 'backup_label' is present".

	Fixes #69

2016-05-04  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add check for empty WAL archive
	Add a check, visible in the check output only if fails, that war the
	user about a bad configuration of WAL shipping checking the dimension of
	the xlog.db file.

2016-05-03  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add implementation of the switch-xlog command
	Implement the switch-xlog command that invokes
	pg_switch_xlog() on one or any PostgreSQL server.

	A --force option issues a CHECKPOINT just before the
	execution of the switch, forcing the server to generate
	a new WAL file.

2016-05-05  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Avoid logging a traceback during check with wrong ssh_command
	If ssh_command does not exists avoid logging an exception during checks
	with PostgreSQL versions older than 9.4

2016-05-02  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add `streaming_archiver_name` option
	When `streaming_archiver` is enabled, the `streaming_archiver_name`
	option properly sets the `application_name` of the managed `pg_receivexlog`
	process, improving monitoring of physical streaming replication clients.
	By default, `streaming_archiver_name` is set to `barman_receive_wal`.

	Available for PostgreSQL 9.3 and above.

	Remove redundant initial assignment

2016-04-28  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add reference to external project BarmanAPI

2016-04-22  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Use absolute unit values in pretty_size()

2016-04-12  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Avoid crashing with a malformed conninfo parameter
	When a connection fails, prepend the error message with the connnection
	type ('conninfo' or 'streaming_conninfo').

2016-04-09  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix usage of the --dbname option.
	The --dbname option is available only from version 9.3 of pg_receivexlog.
	This patch fixes the wal_archiver module, forcing barman when
	pg_receivexlog version is <= 9.2 to split the
	conninfo parameter.

	Fixes #36

2016-04-05  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add timestamp and exit code to barman-wal-restore output

2016-04-01  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Make sure PostgreSQL connections are cleanly closed
	Modified the cli module to wrap every Server's top level command
	in a closing(server) context.

	Added also an atexit handler which logs a warning for stale connections.

	During backup command, the PostgreSQL connections are closed as soon as
	they are no longer needed.

2016-03-15  Leonardo Cecchi  <leonardo.cecchi@gmail.com>

	Make Barman unit tests remove temporary folders
	During the recovery executor unit tests, Barman was leaving a few
	temporary folders.

	This patch makes the unit tests remove them and covers a few corner
	cases where Barman, in consequence of a command failure, was not
	deleting the temporary folders.

2016-03-16  Rubens Souza  <rubens.souza@2ndquadrant.it>

	Minor change in man page for streaming_conninfo

2016-03-14  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix warning notices during sphinx execution

2016-03-12  Christoph Moench-Tegeder  <christoph@2ndquadrant.de>

	Mark "failed backups" check as non-fatal
	Allow taking a new backup in case of presence of failed backups.
	This bug was introduced in version 1.6.0.

2016-03-11  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Rename '-x' get-wal option as '-z'
	The old '-x' remains as a deprecated undocumented alias.

2016-03-10  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Add archive_mode=always support for PostgreSQL 9.5
	This patch adds a basic archive_mode=always support for PostgreSQL
	9.5. This archive mode is simply considered equal to "on".

	Thanks to Koichi Suzuki for opening a bug and proposing a fix.

	Fixes #32

2016-03-07  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Small comment fix in scripts/barman-wal-restore

2016-03-04  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Set version to 1.6.1a1

2016-03-03  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Check for PostgreSQL superuser privileges
	* Add the is_superuser property to PostgreSQLConnection
	* Add a related check for the server

	Closes #30

2016-02-26  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add `--peek SIZE` option to `barman get-wal`
	When the `--peek SIZE` option is passed to `get-wal`, it will peek from
	the WAL archive the requested WAL file as well as subsequent files.
	'SIZE' must be an integer >= 1. When invoked with this option, get-wal
	returns a list of zero to 'SIZE' WAL segment names, one per row.

2016-03-01  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Refine dependencies on external software
	Barman requires psycopg2 >= 2.4.2 to support streaming archiving

2016-02-26  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Prepared version 1.6.0

2016-02-24  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fix subtle bug in local recovery

2016-02-20  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Depend on pytest-runner module only when tests are actually invoked

2016-02-18  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Enable E501 check (long lines) in flake8
	Enabled in flake8 the E501 check, and reformatted text according to the
	new line length policy.

	Use pytest-runner module to run tests in setup.py

2016-02-18  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Fix check if ssh is missing
	Catch any 'command not found' errors during ssh check

2016-02-17  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Prepared version 1.6.0b3

2016-02-15  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Stop running tests on Python 3.2

2016-02-11  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Make the code compatible with Python >= 3.2
	All the incompatible constructs were rewritten to support both Python 2.x
	and 3.x. Previously we were running 2to3 during the setup execution.

2016-02-16  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add Leonardo Cecchi to authors

2016-02-15  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Rollback transaction after starting a backup
	Following the complete refactoring of the PostgreSQL connection code,
	part of this 1.6.0 previous versions (alpha1, beta1 and beta2),
	Barman is keeping an IDLE connection while taking a backup.
	This could trigger spurious alerts if a transaction has been left open.

2016-02-15  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fix output of 'barman check --nagios'
	Display the correct number of servers with 'barman check --nagios'

2016-02-12  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Enhance test coverage for the wal_archiver module
	Remove the unused 'is_wal_relevant' method as well.

2016-02-12  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add section about contributing for Barman

	Add a section about .partial files in the documentation

2016-02-12  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Add 'receive-wal running' check
	When `streaming_archiver` is enabled, this check ensures that
	a `receive-wal` process is running.

2016-02-12  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add --reset option to receive-wal comand
	Add the `--reset` option to the `receive-wal` command, allowing
	users to reset the WAL location of the streaming archiver.
	This is useful in case Barman goes out of sync with the master
	server.

2016-02-10  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

2016-02-10  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Prepared version 1.6.0b2

2016-02-05  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Add check for FAILED backups
	The 'barman check' command now checks for the presence of failed backups.

2016-02-08  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Complete integration of RemoteStatusMixin class
	Let classes implementing the the get_remote_status method inherit
	from the RemoteStatusMixin base class, completing the refactoring
	activity started with a previous commit (3179a1e).

2016-02-08  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix unwanted removal of duplicate WAL from archive
	This commit removes an unused and dangerous option from the Compressor
	interface that allowed the removal of the source WAL file after
	compression (introduced in 1.6.0 development branch).

	Fixes: #27

	Remove trashing of old WAL files from the archiver
	The archiver now always archives any file with a valid name that
	finds inside the incoming directory.

	NOTE: The block of code that has been removed changes behaviour
	of existing Barman installations, but in a conservative way.
	We defer to future versions of Barman to better understand
	WAL files' sequentiality (parsing history files, for example),
	by writing specific classes/modules responsible for deeper
	and more robust checks in the archive area.

2016-02-05  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Manage the termination of receive-wal in cron
	When the 'streaming_archive' option is disabled for a server,
	a cron execution makes sure that a running receive-wal process
	is correctly terminated.

	Clearer output and logging messages for `receive-wal`.

2016-02-05  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Add check for archiver errors
	Barman now checks for the presence of files in the error directory.

2016-02-04  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Improve error handling when required server parameters are missing
	Add error handling in case a required parameter (i.e. ssh_command,
	conninfo, ...) is missing, by emitting meaningful error messages.

2016-02-02  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Adjust severity of some messages produced by cron command
	The messages about subprocess execution produced by the cron command
	were emitted for every cron run. They provide little information and
	their presence may make the log difficult to analyse, so they have been
	removed from log.

2016-02-02  Aydan  <Aydan Tasdemir>

	Update docs - Added remote recovery command
	Closes: #26

2016-02-01  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Prepared version 1.6.0b1

2016-01-29  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix python 3.2 error on travis-ci

	Avoid exceptions during ProcessManager initialization
	Previously a LockFileParsingError exception could have been raised if a
	lock file were empty.

	Also, the LockFile class now writes the lock content atomically

2016-01-28  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Remove a spurious error when last_failed_wal is a history file
	Fixes: SF#77

2016-01-27  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fix the 'status' command output with PostgreSQL < 9.4
	Make sure only one 'Last archived WAL' line is present.

	Fix the command used to retrieve the last archived WAL through
	ssh when the pg_stat_archiver view is not available.

	Fixes: #25

2016-01-28  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Set version to 1.6.0b1

2016-01-21  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add --stop option to receive-wal command
	Add the option --stop to the receive-wal command, allowing the user to
	stop a receive-wal process that running in background for a specific server.

	The implementation is based on the new class ProcessManager that will be
	used in the future to manage all barman's sub-processes.

2016-01-24  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add current data size of the server in 'barman status'

2016-01-22  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Updated NEWS with current code

2016-01-19  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Improve archive-wal robustness and log messages
	Display clearer error messages when discarding a WAL file.
	Also, WAL files that are older than the first backup are
	now moved in the 'errors' directory: FAILED backups are now
	ignored in this process.

	TODO: Add a method to check the presence of files in the error
	directory.

	Fixes: #24

2016-01-21  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Move archive-wal code into the wal_archiver module

2016-01-17  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Reorder options in barman.5 man page

2016-01-15  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add include sorting check in tox flake8 environment

2015-12-05  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Implement pygzip and pybzip2 compressors
	The new pygzip and pybzip2 compressors are implemented using Python's
	internal modules for compressing and decompressing gzip and bzip2 files.
	This avoids the cost of forking and executing the shell.

	Initial idea by Christoph Moench-Tegeder <cmt@2ndquadrant.de>

2016-01-14  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update copyright years to 2016

2016-01-08  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add 'receive-wal' cron integration
	Add integration of the new command 'receive-wal' during the execution of
	the cron command.

	If the option 'streaming_archiver' is on, the command 'receive-wal' is
	automatically executed as subprocess by the cron command, allowing
	Barman to receive WAL files from PostgreSQL using the native streaming
	protocol.

	TODO: Implement a way to interrupt a running streaming process

2015-12-09  Stefano Zacchiroli  <zack@upsilon.cc>

	Add support for the pigz compressor
	Pigz on-disk format is the same of gzip, but (de)compression is much
	faster due to the usage of CPU parallelism.

	For write-intense workloads pigz vs gzip might make the difference,
	between struggling to keep up with archiving incoming WALs and being
	able to keep up easily.

2016-01-09  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add check() method in WalArchiver classes
	Move code from the server.check_postgres() method in
	the WalArchiver class, so that each implementation (FileWalArchiver
	and StreamingArchiver) perform their specific checks.
	Created the RemoteStatusMixin class that implements a cacheable
	version of remote status information.

	Currently it is only used by the WalArchiver class tree.
	TODO: propagate it to all other classes implementing the
	get_remote_status() function in a later commit.

2016-01-08  Christoph Moench-Tegeder  <christoph@2ndquadrant.de>

	Permit `archive_mode=always` for 9.5 servers
	PostgreSQL 9.5 introduces archive_mode=always. Even though Barman's
	current framework for checks does not allow yet to correctly monitor
	archiving from standby servers, Barman now allows users to set
	`archive_mode` to `always`, provided that archiving happens from the master.

2016-01-08  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fix server creation and output of check command
	Modified the init method of Server Object, managing errors/exception
	using the msg_list attribute of the configuration.
	Modified the check command output.

2016-01-04  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Rename get_first_backup and get_last_backup methods
	Renamed get_first_backup() and get_last_backup() methods of the
	Server and BackupManager classes, respectively into get_first_backup_id()
	and get_last_backup_id(), for coherence and code clarity.
	Now the fact that the method returns the ID of a backup instead
	of a BackupInfo object has been made explicit.

	Fix management of backup id during backup deletion
	Fix an error that prevents the deletion of an existing
	backup while a backup on the same server is in progress.

	Fixes #22

2015-12-31  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Use isort python package to keep imports ordered

2015-12-14  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Implement `receive-wal` command
	The `receive-wal` command receives the stream of transaction logs for a
	server. The process relies on `pg_receivexlog` to receive WAL files from
	a PostgreSQL server through the streaming protocol.
	It is a foreground process which lasts until it receives a SIGINT.

2015-12-31  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add support for long running Commands
	The Command class now supports the execution of long running
	subprocesses, by passing the subprocess output and error streams
	to a handler, line by line.

	Remove spurious newlines from log lines
	Error messages coming from libpq often terminate with a newline.
	Always call strip() on strings that have been extracted from
	exceptions before logging them.

	Make get_remote_status methods behaviour coherent
	A 'get_remote_status' method should never raise an exception in case of
	error. The expected behaviour is to set missing values to None in the
	resulting dictionary.

2015-12-29  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Fix exception when streaming connection is unavailable
	Previously, when 'streaming_archiver' was on and pg_receivexlog correctly
	installed, Barman would raise an AttributeError exception when streaming
	connection to PostgreSQL was unavailable, as it could not detect the
	server version.

	This commit fixes the error by modifying the management of a failed
	connection to PostgreSQL through streaming connection. It also adds
	an unit test for this scenario, as well as hints for the user during
	`barman check` suggesting a possible streaming connection error.

2015-12-28  Leonardo Cecchi  <leonardo.cecchi@2ndquadrant.it>

	Propagate "-c" option to any Barman subprocess
	The "-c" option lets users specify a custom location for the configuration
	file. It must be propagated to every Barman subprocess for consistency
	of behaviour.

	Fixes #19

2015-12-24  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Update documentation about WAL archiving

2015-12-22  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Manage WAL duplicates in `archive-wal` command
	In case a WAL file is archived twice, `archive-wal` will check
	if the new file is identical to the already archived one (compressed
	WAL files will be decompressed and checked). If yes, the second
	file is discarded, otherwise it is saved in the 'errors' directory.

	Accept compressed WAL files in incoming directory
	In case a WAL file in the incoming directory is already compressed
	using a known algorithm, archive it as is.

	Add streaming support to 'archive-wal' command
	The `archive-wal` command now looks for both archived WAL files
	(shipped from PostgreSQL through its `archive_command`) and streamed
	ones (via the `receive-wal` command).

2015-12-24  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add sleep time option to `barman-wal-restore` script
	Introduce the '-s seconds' option to the barman-wal-restore script.
	After a failure of the get-wal command, the script will sleep for
	the specified amount of seconds (default 0, no wait) before
	exiting the restore_command from PostgreSQL.

	Closes #7

2015-12-22  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add `streaming_wals_directory` and `errors_directory` options

	Make sure WAL archiving is atomic. Fixes: #9,#12

2015-12-15  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Get rid of pytest warnings from pytest-catchlog

2015-12-11  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Restore compatibility with PostgreSQL 8.3
	On PostgreSQL older than 8.4, disable the query that detects those
	settings coming from an included file - as the required pg_setting
	field was only included in that release.

2015-11-24  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fixes possible unhandled exceptions in barman.postgres
	Improve the testing coverage of barman.postgres module.

2015-12-06  Christoph Moench-Tegeder  <cmt@2ndquadrant.com>

	Remove redundant sete[gu]id() calls
	Python's documentation of the os.setuid/os.setgid calls is
	oversimplified to the point of being wrong. Under the hood,
	the setuid(2)/setgid(2) functions are used, thus making these
	calls as POSIX compliant as the underlying operating system.
	As barman won't be installed SUID (it's an interpreter file),
	effective, real and saved user id should always be the same.
	That way, the only reason to use setuid() family calls at all
	is for a privileged user (root) to change over to the barman
	(unprivileged) user. According to POSIX.1-2013, the setuid()
	call changes all three (real, effective and saved) uids if
	called by a privileged user (the same holds for setgid() and
	the group ids). In this construct, the additional seti[gu]id
	calls are redundant no-ops and can be removed.

2015-11-27  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add support for 'streaming_archiver' option
	Add a global/server boolean option called 'streaming_archiver' which
	enables the use of the PostgreSQL's streaming protocol to receive
	transaction logs from a server.
	If set to 'on', Barman expects to find `pg_receivexlog` in the PATH
	(see `path` option) and that streaming connection for the server is
	properly working. Subsequently, it activates both proper connection
	checks and management of WAL files.

	Refactored the code to use a new class, the StreamingWalArchiver,
	which is responsible for managing WAL files using a streaming
	connection.

	Add support for 'archiver' option
	Add a global/server boolean option called 'archiver' which
	enables continuous WAL archiving (through log shipping using
	PostgreSQL's archive_command) for a specific server.
	If set the 'true' (default), Barman expects that PostgreSQL
	continously deposits WAL files in the 'incoming' directory
	of that server, and subsequently activates proper checks and
	management.

	Refactored the code to use a new class, the FileWalArchiver.
	The class is in charge of both archival and compression of WAL files.
	This is the first step towards the implementation of streaming
	archiving support.

2015-11-25  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Reorder constants in barman.config module

2015-11-20  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add `streaming_conninfo` server option
	Added the `streaming_conninfo` option to Barman configuration.
	This option (with server scope only), adds the support in Barman
	for a streaming replication connection to a PostgreSQL server.

	Added also a simple check for the barman check command, that tests
	that the streaming connection to PostgreSQL is properly working.

2015-11-13  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add support for `path_prefix` configuration option
	The `path_prefix` global/server option can contain one
	or more absolute paths, separated by colon, where Barman
	looks for executable files.

	The paths specified in `path_prefix` are tried before the
	ones specified in `PATH` environment variable.

2015-11-10  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Refactored access to PostgreSQL
	Created the 'postgres' module which is responsible for any communication
	with a PostgreSQL server, wrapping all connection and query management
	activities. This module is the only point of access to PostgreSQL.

2015-11-06  Stefano Bianucci  <stefano.bianucci@2ndquadrant.it>

	Rename `copy_method` configuration key to `backup_method`

2015-11-03  Giuseppe Broccolo  <giuseppe.broccolo@2ndQuadrant.it>

	Create an implicit restore point at the end of a backup
	At the end of a backup, if the PostgreSQL version is 9.1 or newer,
	create an implicit restore point using the pg_create_restore_point()
	function.

2015-11-11  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Set version to 1.6.0a1

2015-11-13  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Prepared version 1.5.1

2015-11-13  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Create pg_xlog/archive_status directory at the end of recovery
	On PostgreSQL 8.3, the pg_xlog/archive_status directory is not
	automatically created, if missing, during the startup. To avoid errors
	Barman now creates that directory at the end of any recovery operation.

2015-11-13  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Converted sphinx directory README to Markdown

2015-11-10  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Pin (temporarily) pytest-catchlog to version 1.1
	The pytest-catchlog release 1.2.0 broke Python 2.6 compatibility, so pin
	the 1.1 version until it will be restored.

	Ref: https://github.com/eisensheng/pytest-catchlog/pull/15

	In latest release of pytest-catchlog the result of caplog.records() call
	is a copy of the event list so any modification of that has no global
	effect. Change the code that uses caplog.records() to not rely on the
	previous undocumented behaviour.

2015-11-05  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix a typo in barman-wal-restore help output

2015-11-04  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Avoid forcing arcfour cypher in barman-wal-restore
	OpenSSH 6.7 dropped many low security algorithms, and arcfour is one of
	them. So we stop forcing it in the script, allowing the user to define
	the wanted algorithm in ~/.ssh/config

	Ref: http://www.openssh.com/txt/release-6.7

2015-11-04  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Improve error messaging for missing config files

2015-11-04  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Set version to 1.5.1b1

2015-10-27  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Manage 'wal_level' only for PostgreSQL >= 9.0
	Really fixes #3

2015-10-26  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix all E731 pep8 errors
	Always use a def statement instead of an assignment statement that binds
	a lambda expression directly to a name.

	See: https://www.python.org/dev/peps/pep-0008/#programming-recommendations

2015-10-22  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

2015-10-22  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	NEWS for 1.5.1a1

	Set version to 1.5.1a1
	Revert version to 1.5 branch due to relevant bug fixing

2015-10-12  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add the 'archive-wal' command
	The 'archive-wal' command executes the WAL maintenance operations on a
	given server.

	This command is automatically executed as a subprocess by the cron
	command, allowing the parallel execution of WAL archiving on
	different servers.

2015-10-20  Giuseppe Broccolo  <giuseppe.broccolo@2ndQuadrant.it>

	Avoid 'wal_level' check on PostgreSQL version < 9.0
	Conditionally skip 'wal_level' check in PostgreSQL
	versions prior to 9.0.

	Fixes #3

	Avoid retention policy checks during the recovery
	Additionally, skip any missing tablespace directory during the deletion
	of a backup (like we already do with missing pgdata directory)

2015-10-20  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix more incorrect mock assert calls in unit tests
	Add flake8-mock plugin in flake8 tox environment to make sure it will
	not happen again.

	Some more cleanup of testing infrastructure has been done:

	* Switch from pytest-capturelog (unresponsive upstream) to
	  pytest-catchlog.
	* Remove the version pinning from mock as 1.3.0 supports py2.6 again.
	* Add flake8-copyright and flake8-string-format plugins.

2015-10-09  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Allow parallel cron execution for different servers
	Allow the execution of multiple 'barman cron' processes, each one
	handling a different server. Servers will be handled sequentially,
	skipping those that are already being served by another cron process.

2015-10-22  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix calculation of backup size
	In Barman 1.5.0 the internal directory structure of a backup has been
	changed, moving tablespaces from being inside the data directory to
	being inside the backup directory. The method backup_fsync_and_set_sizes()
	now takes that into account by running against the whole backup directory.

	Fixes #5

2015-10-12  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix some incorrect mock assert calls in unit tests

2015-10-06  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Support for mixed compression types in WAL files (#61)
	Decompress each WAL file using the specific algorithm
	reported in the XLOG archive (xlogdb).

	Improve management of xlog.db errors
	Better management of errors during the decoding of the name
	of an xlog segment. Added a hint that suggests to run
	"barman rebuild-xlogdb" to fix corruptions in the xlogdb file.

2015-10-02  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Started version 1.6.0a1

2015-09-25  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Update Copyright notice

2015-09-25  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Remove obsolete section from tutorial
	Removed a section in the tutorial regarding the 'delete'
	command, as pointed out by Ian Barwick.

2015-09-25  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Enable flake8 in every tox run

2015-09-24  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Prepared version 1.5.0 for final release

2015-09-24  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update tox.ini to allow specifying a target test
	Also add .cache directory, which is created by latest tox, to .gitignore

2015-09-24  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Changed import of call() method.
	Now is compatible with all versions of py.test libraries

2015-09-16  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Support for non-critical checks during backup
	Add a filter for non-critical checks during backup operations.
	This patch fixes a bug that prevented users from taking a new
	backup when minimum redundancy or smelly backup checks failed.

2015-09-15  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

2015-09-15  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Prepared for 1.5.0 alpha 1

2015-09-09  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Converted man pages to Markdown

2015-09-08  Gianni Ciolli  <gianni.ciolli@2ndQuadrant.it>

	Some small fixes to the tutorial

2015-09-01  Francesco Canovai  <francesco.canovai@2ndquadrant.it>

	Updated spec file for RPM building

2015-09-04  Giuseppe Broccolo  <giuseppe.broccolo@2ndQuadrant.it>

	Add barman-wal-restore script
	Add a simple bash script to be used as `restore_command`
	in a standby scenario, as fallback method. The script
	uses ssh to connect to the Barman server and requests
	the required WAL file for recovery.

	The script checks that destination path is not a directory.

2015-08-27  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Convert tutorial to Markdown

2015-08-28  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Manage options without '=' in PostgreSQL configuration files

2015-08-18  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Allow 'pre' retry hook scripts to stop the command
	Add EXIT_ABORT_STOP (63) and EXIT_ABORT_CONTINUE (62) exit codes
	to control how a retry hook script is aborted.

	Termination of a retry hook script with EXIT_ABORT_CONTINUE informs
	Barman to continue with its operations.

	By terminating a retry hook script with EXIT_ABORT_STOP, users
	request Barman to stop its main operation (i.e. backup or WAL archival).

	EXIT_ABORT_CONTINUE is implemented by every retry hook script.

	EXIT_ABORT_STOP is currently implemented only with 'pre_backup_retry_script'
	and 'pre_archive_retry_script'.

	EXIT_ABORT_STOP is currently ignored by 'post_backup_retry_script' and
	'post_archive_retry_script', and its behaviour in these cases is
	identical to EXIT_ABORT_CONTINUE.

2015-08-21  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add flake8 tox environment

2015-08-18  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Documentation for 'barman_lock_directory'

2015-08-18  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Analyse include directives for PostgreSQL
	Check if PostgreSQL administrators take advantage of include
	directives and specify additional files. Include directives
	(include, include_if_exists, include_dir) are not supported
	in Barman for files that reside outside PGDATA.

	During backup, warn users and list the files that require
	manual backup. During recovery, warn users about the presence
	of include directives in PostgreSQL configuration.

2015-08-18  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Workaround for rsync on SUSE Linux
	Many thanks to Christoph Moench-Tegeder <christoph@2ndquadrant.de>
	for his workaround proposal. This fixes the outstanding issue
	on SUSE Linux - previously raised by issues #13 and #26.
	See also: https://bugzilla.opensuse.org/show_bug.cgi?id=898513

2015-08-17  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Added copy_method option (fixed to rsync)

2015-08-06  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Support for 'retry' hook scripts for backup and WAL archiving
	A 'retry' hook script is a special kind of hook scripts that Barman
	tries to run indefinitely until it either returns a SUCCESS (0) or
	an ABORT (63) exit code.

	Safe hook scripts are executed immediately before (pre) and after (post)
	the command execution. Standard hook scripts are executed immediately
	before (pre) and after (post) the retry hook scripts.

	This patch adds support for pre/post retry hook scripts for backup
	and WAL archiving.

	The following four global/server configuration options have been added:

	* pre_backup_retry_script: executed before a backup
	* post_backup_retry_script: executed after a backup
	* pre_archive_retry_script: executed before a WAL archive operation
	* post_archive_retry_script: executed after a WAL archive operation

	By default, no retry hook script is executed. Environment variables
	are identical to the equivalent standard hook script.

2015-08-12  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Updated AUTHORS file

2014-05-29  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Support for the 'get-wal' command
	The 'barman get-wal' command allows users to fetch any WAL file
	from the archive of a specific server in Barman.

2015-08-21  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add simple Travis CI integration

2015-08-11  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Set version 1.5.0a1

2015-07-30  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Preserve Timeline history files. Fixes #70.
	Added check for the management of history files during the removal of a
	backup or during normal maintenance operations (cron).

2015-08-06  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Forbid the delete of a running backup
	Block the execution of a barman delete command on a backup that is in progress.
	Use locks to ensure that there are no running backups on that server.
	Otherwise check for the position of the backup to be deleted in the
	catalogue and its status: if it is the first backup and its status is STARTED or EMPTY,
	the backup is running and delete is forbidden.

2015-08-03  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Follow symlinks when checking directory paths
	Check conflicting paths in configuration files using canonical paths and
	following symlinks when necessary.

2015-07-13  Francesco Canovai  <francesco.canovai@2ndquadrant.it>

	Pin mock==1.0.1 for testing.
	In version 1.1.x mock has changed some behaviours and is currently
	incompatible with our unit tests.

2015-07-08  Stefano Bianucci  <stefano.bianucci@2ndquadrant.it>

	Execute check() before starting a backup
	Execute the full suite of tests from barman check command before
	starting a backup. Skip the execution of a backup in case check
	fails.

	Add a strategy for managing the results of the checks. Now every
	check is properly logged as error (failure) or debug (success).

2015-07-03  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Support code documentation using Sphinx
	Generate code documentation using Sphinx autodocs

2015-07-13  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Modified test for correct management of timezone

2015-07-10  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Properly support BackupInfo serialization in JSON format.

2015-07-01  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Improved management of configuration in tests
	Improved and simplified the management of configurations in tests. Added
	a method for the creation of dictionaries containing all the keys that
	are usually present in a configuration. Updated tests accordingly.

2015-06-18  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Second part of the backup_executor module's refactor
	Refactored and streamlined the executor module. Introduced a specific
	class for physical backup with Rsync through Ssh, as well as a strategy
	pattern for the management of exclusive and concurrent backups (for
	backups from a standby).

2015-06-23  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Standard error management for server commands
	Standardised the process of managing errors with server commands in
	barman/cli.py. By default, inactive servers are skipped (without error)
	as well as temporarily disabled servers (with error). No distinction is
	made between calling a command with one server as target or with a list
	of them (including 'all'). Exceptions are check (no server is skipped,
	errors are returned only for active servers), cron (no error is ever
	returned), list-server and diagnose (both managing active/disabled
	servers with no errors). Inactive servers are the ones with 'active'
	option set to False. Disabled servers are the ones with internal
	directory conflicts (e.g. WALs directory = base backup directory).

2015-06-23  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Asciidoc support for man pages and tutorial

2015-06-16  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fixed error in WAL rate calculation.
	Solved an error during the evaluation of the WAL rate for a backup.
	Added two basic unit tests.

2015-06-12  Stefano Bianucci  <stefano.bianucci@2ndquadrant.it>

	Add check for wal_level
	For better usability, warn users about setting a proper value
	for wal_level setting in PostgreSQL.

2015-05-14  Stefano Bianucci  <stefano.bianucci@2ndquadrant.it>

	Add check for conflicting paths in Barman's configuration
	Added controls for path-clash during the creation of Barman servers.
	If there are conflicting paths, Barman will disable those servers
	containing errors.
	If a potentially destructive command like "backup" is issued on a server
	containing conflicts, Barman exits with an error message.

2015-05-21  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Complete refactor of the 'recover' command
	The main 'recover' command has been completely refactored, through
	the creation of a separate module, called 'recovery_executor'.
	The RecoveryExecutor class now embodies both local and remote
	operations, laying the road for future improvements.

	This patch also fixes #68, by disabling dangerous settings in
	postgresql.auto.conf (available from PostgreSQL 9.4).

	Basic unit tests have been added.

2015-05-19  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Rename class Server to ServerConfig in barman.config module
	Previously both barman.config and barman.server modules had a Server
	class. The former has now been renamed to ServerConfig, hence removing
	the ambiguity.

2015-05-21  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix incompatibility with tox version >= 2

2015-05-08  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Make sure that even an EMPTY backup has a server set.

2015-05-07  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Improve xlog.db integrity (Closes: #67)
	* Execute flush() and fsync() after writing a line in xlog.db
	* Execute fsync() on incoming directory after every WAL is archived

2015-04-13  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Remove unused 'server' argument from WalFile.from_xlogdb_line()
	This also fix regression in 'barman delete' command introduced with
	commit 7ac8fe9c41fd7e5636f370abdc92ca785057263e.

2015-04-13  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Fix exception during error handling in barman recovery (Closes: #65)

2015-03-24  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Improved cache management of backups
	Streamlined the management of the cache of the backups.
	It is now possible to register and remove a backup from the cache.
	The cache structure has been simplified and now it is a simple
	dictionary of backups. Status filters are applied by the
	get_available_backups() method.
	Managed registration and removal of a backup in the cache
	during backup operations.

2015-03-02  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Create backup_executor module and refactor backup.
	Extract all the methods relative to the execution of a backup into a
	dedicated object.
	The RsyncBackupExecutor encapsulates the operation of backing up a
	remote server using rsync and ssh to copy files.

2015-03-27  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Improved management of xlogb file
	The xlogdb file is now correctly fsynced when updated.
	Also, the rebuild-xlogdb command now operates on a temporary
	new file, which overwrites the main one when finished.

2015-03-27  Stefano Bianucci  <stefano.bianucci@2ndquadrant.it>

	Add "active" configuration option for a server
	It is now possible to temporarily disable a server through
	the 'active' configuration option. Defined at server level
	as a boolean, when set to False the server is ignored by
	main operational commands such as cron, backup and recover.
	By default, it is set to True.

2015-03-20  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Modified Barman version following PEP 440 rules.
	Changed Barman version from 1.4.1-alpha1 to 1.4.1a1
	following PEP 440 rules.
	Adapted RPM build scripts to produce packages with
	the correct names.

2015-03-17  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Fix computation of WAL production ratio
	The WAL file ratio reported in the 'show-backup' command output was wrong
	because it was considering only the number of WALS produced during the
	backup instead of the number of WALs produced until next backup.

2015-02-25  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Fix for WAL archival stop working if first backup is EMPTY
	In some rare cases, if an empty backup has left by a failure during a
	backup, the cron could start trashing WAL files, as if there is no
	available backups.

	Closes: #64

2015-03-06  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Added 'barman_lock_directory' global option
	Barman now allows to specify the default directory for locks through
	the global option called 'barman_lock_directory', by default set to
	'barman_home'. Lock files will be created inside this directory.

	Names of lock files have been slightly changed. However, this won't
	affect users of Barman, unless you are relying on their locations and
	paths.

	This patch introduces four classes for managing lock files:
	GlobalCronLock, ServerBackupLock, ServerCronLock and ServerXLOGDBLock.

2015-02-03  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	New backup directory structure
	Inside the backup directory the 'pgdata' has been renamed to 'data'.
	Tablespaces, if present, are stored into individual directories alongside
	the 'data' directory.
	During backup and recovery operations, tablespaces are copied individually
	using a separate rsync command.

2015-02-12  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Improve backup delete command
	Improve robustness and error reporting of backup delete command

2015-02-10  Stefano Bianucci  <stefano.bianucci@2ndquadrant.it>

	Add unit tests for dateutil module compatibility

2015-02-08  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	After a backup, limit cron activity to WAL archiving only (#62)

2015-01-28  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Clarify rpm spec comments about pre-releases

2015-01-28  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Updated backlog (TODO list)

2015-01-23  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update metadata in setup.py
	- Improve barman description
	- Add Python 3.4

2015-01-23  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Started version 1.4.1-alpha.1

	Update the ChangeLog file

	Prepared version 1.4.0

2015-01-20  Francesco Canovai  <francesco.canovai@2ndquadrant.it>

	Updated spec files for RHEL7

2015-01-16  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Delete basebackup dir as last action of a delete.
	Split the delete operation: remove the PGDATA directory first,
	then the related WAL files and, at last, the basebackup directory.

2015-01-13  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Add minimum_redundancy tests in test_retention_policy.py

2015-01-13  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Fix calculation of deduplication ratio

2015-01-12  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Update the ChangeLog file

	Prepared documentation for version 1.4.0-alpha1

2015-01-11  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Store deduplication effects for incremental backup
	When incremental backup is enabled and uses hard links
	(reuse_backup = link), output of 'backup' command reports
	the effects of deduplication. The metrict is stored along
	the backup.info file in the 'deduplicated_size' field.

	IMPORTANT: this metric refers to the increment in size
	of the current backup from the previous backup and reflects
	only the situation at backup time.

2015-01-10  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Prepared version 1.4.0-alpha1

	Updated copyright to 2015

2015-01-09  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix smart_copy of tablespaces when using bwlimit option

2015-01-07  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Add dedicated exception for PostgreSQL connection errors

2015-01-08  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Fix missing argument error in retention policies backup_status method
	Improve test coverage for retention_policy.py module

2015-01-07  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Remove logging of tracebacks on error during backup

2015-01-05  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Avoid logging of tracebacks during smart copy
	While retrieving the list of files on destination for smart copy, log
	any failure as error instead of exception

2014-12-22  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Unit tests for BackupInfo object

2014-12-24  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Change the way BackupInfo are created for testing
	Merge the method build_test_backup_info and the mock_backup_info.
	Now we use real BackupInfo objects instead of a Mock

2011-12-07  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Incremental base backup implementation
	Add support for reuse_backup global/server option, accepting
	three possible values:

	* off:  no incremental backup support (default)
	* copy: uses the last available backup as source (preventing unmodified
	        files from being copied)
	* link: same as copy but uses hard links on destination, if the filesystem
	        on the backup server allows it (reducing the occupied space)

	Add support for command line '--reuse-backup' option (default: link).

2014-12-24  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Allow last_archived_wal to be any xlog file
	Correctly show any xlog file as last_archived_wal for
	pre-pg_stat_archiver cases.
	Improve testing and docstrings for barman/xlog.py module.

2014-12-09  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Improve robustness of ssh_command and conninfo options

2014-12-18  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	pg_stat_archiver support for PostgreSQL 9.4
	Integrate pg_stat_archiver with PostgreSQL 9.4 servers for
	the barman check, show-server and status commands.

2014-11-28  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Improve robustness of retention policy unit tests

2014-12-16  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Fixes retention policies delete bug (#58)
	The method responsible for deleting obsolete backup in
	retention policies enforement, will not raise anymore
	the 'NoneType object is not iterable'. This prevents
	barman from terminating abruptly.

2014-11-28  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Pass list of available backups to retention policy code

2014-12-02  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Include history files in WAL management

2014-12-04  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Added a util method to find an executable in system path.
	If rsync is not present on system, a proper error message is
	displayed to the user when a command using rsync is issued

2014-12-09  Giulio Calacoci  <giulio.calacoci@2ndquadrant.com>

	Changed behaviour if pg_ident.conf is missing from an error to a warning

2014-10-22  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Remove code to replace output stream when quiet
	Previously the '-q' option was handled replacing the standard output
	stream with one which trows away averything it gets. Now it is not
	needed anymore because we haver a proper output module.

2014-09-26  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Remove all remaining output done by yield
	Migrate all the remaining part using yeld to do output to using the new
	output module.

2014-10-07  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Ignore fsync EINVAL errors on directories (#55)
	On some filesystem doing a fsync on a directory raises an EINVAL error.
	Ignoring it is usually safe.

2014-09-23  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Modified output module to access protected properties: quiet and debug

2014-09-10  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix bash autocompleter
	Minor changes:
	* Some code formatting adjustments

	Move cron retention policy management to a separate method

2014-09-05  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix dates in rpm changelog

2014-09-03  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Calculate backup WAL statistics only if the WALs are already processed

2014-09-02  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Change default LockFile behaviour to raise if fails acquisition

2014-09-01  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Invoke WAL maintenance after a successful backup
	* At the end of the 'barman backup' command, maintenance operations
	  are automatically started for successful backups (equivalent to
	  manually executing a 'barman cron' command, just for that server)

	* Trashing of unuseful WALs (part of 'barman cron') has been changed
	  as follows:

	  * in case of one or more backups, delete WAL files older than
	    the start WAL of the first backup
	  * otherwise, trash WAL files in case of exclusive backup server
	    (that is, not concurrent)

2014-09-03  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Remove redundant server argument from HookScriptRunner.env_from_wal_info()

2014-08-27  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Add relpath() and fullpath() methods in WalInfoFile
	* Remove 'full_path' attribute in WalInfoFile
	* Add 'relpath()' method to WalInfoFile, which returns the
	  relative path of a WAL file within the 'wals_directory' directory
	* Add 'fullpath()' method to WalInfoFile, which returns the
	  full path of a WAL file within a server installation (requires
	  a server object)

2014-08-23  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Updated version in .spec file

2014-08-20  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add build_config_from_dicts to testing_helpers module

	Make Config.Server, WalFileInfo and BackupInfo objects json encodable

2014-08-20  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Added unit to JSON representation of a retention policy

	Started version 1.3.4-alpha.1

2014-08-18  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Update the ChangeLog file

	Fixed typo in release date

2014-08-13  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Prepared version 1.3.3

2014-08-12  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add an unit-test for Server.get_wal_full_path() method

2014-08-12  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Refactored building of full path of a WAL file

2014-08-01  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Report which file is about to be archived before actually doing it

2014-07-25  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Remove traceback from output when Barman is interrupted by CTRL-c

	Avoid flushing/fsyncing read only files
	Fixes: #49 EXCEPTION: [Errno 9] Bad file descriptor

2014-07-24  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Added Barman's version number to 'barman diagnose'

2014-07-22  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Move xlogdb_parse_line method in WalFileInfo class

2014-07-23  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Cleanup output API status at the end of test_output.py

2014-07-22  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Estimates WAL production rate for a backup

2014-07-18  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Removed duplicate log message at the end of  'barman recover' wal segments copy

	Fix datetime.timedelta json serialization in 'barman diagnose' command

2014-07-17  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

2014-07-17  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Prepared version 1.3.3-alpha.1 docs

2014-07-17  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Really fix "ssh" version detection in "barman diagnose" command

2014-07-16  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add command line options for retry of backup/recover copy
	Implemented the --retry-times (including --no-retry) and --retry-sleep
	command line options for backup/recovery copy

	Emit warnings in case of unexptected configuration options

2014-07-14  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Reduce the verbosity of the log for "barman cron"
	Currently the "barman cron" command emits one log line for
	every WAL file that's archived (including the server name
	as a prefix). No log line is emitted for an empty cron run.

	Make recovery --target-time option more resilient to wrongly formatted values
	Workaround a bug in dateutil.parser.parse() implementation
	ref: https://bugs.launchpad.net/dateutil/+bug/1247643

	Improved logging for "barman recover" command

	Default log prefix now contains barman process ID (pid)

2014-07-16  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix "ssh" version detection in "barman diagnose" command

2014-07-11  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix wrong variable name in BackupManager.delete_wal()

2014-07-09  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add unit test for LockFile object and server.xlogdb() call
	Minor changes:
	 - converted test_xlog.py to py.test style

2014-07-11  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Make sure remote WAL destination path is a directory
	Add a trailing slash to the remote WAL destination path, in
	order to ensure it is a directory

2014-07-07  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fix serialisation of CvsOption during "barman diagnose" command

2014-07-11  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Use a WalFileInfo object when decoding an xlogdb line

	Add --no-human-readable to rsync --list-only invocation
	In rsync >= 3.1.0 the --list-only format changed adding digit
	groupings by default in "size" field. To obtain the pre 3.1.0 behavior
	you need to add --no-human-readable
	Ref: http://ftp.samba.org/pub/rsync/src/rsync-3.1.0-NEWS

2014-07-09  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Log any hook script failure with its output at warning level

2014-07-08  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Wraps xlogdb() code in a try/finally block

2014-06-28  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix wait parameter logic in LockFile class
	In previous versions the wait argument on the LockFile constructor was
	mistakenly ignored, actually preventing the usage of a waiting lock
	through the Context Manager interface

	Always use utils.mkpath() to create directories that could already exist
	Minor changes:
	 - In retry_backup_copy log the exception which caused the failure

2014-06-27  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Really ignore vanished files errors in rsync smart copy routine

2014-06-27  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Added info messages for the four phases of the new rsync smart copy
	Minor changes:
	- Fix unit tests for basebackup_retry_* config values

	Updated documentation for 1.3.3-alpha1

	Set default for basebackup_retry_times to 0
	For compatibility with previous Barman versions, set
	basebackup_retry_times to 0 as default value.

2014-06-26  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Make sure timestamps are tz-aware anywhere in the code
	Minor changes:
	 - Add basic unit tests for retention policies

2014-06-26  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Close all open descriptors but std{in,out,err} when spawning a child process
	Minor changes:
	 - Remove some dead code
	 - fix missing 'last_backup_maximum_age' as global option

2014-06-24  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Display compression ratio for WALs in show-backup

2014-06-23  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Improved Nagios output for check command

2014-06-25  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Manage KeyboardInterrupt exception in 'barman backup'

2014-06-23  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Added support for PostgreSQL 8.3

2014-06-24  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Updated rpm packaging spec to work with pre-releases
	Minor changes:
	- add rsync dependency to barman.spec file

2014-05-29  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Support for comma separated list options
	Added support for a new data type in configuration options:
	comma separated list values. The first option to be implemented
	is backup_options, now accepting a list of values.

2014-06-18  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Decode binary strings in command_wrapper
	This fixes python2.7 and python3 compatibility

	Minor changes:
	- make scripts/release.sh python3 compatible

2014-06-10  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Support for 'last_backup_max_age'
	This new global/server option allows administrators
	to set the max age of the last backup, making it easier
	to detect any issues with periodical backup execution.

2014-06-18  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Support for "smart" incremental recovery
	Avoid invoking rsync with --checksum option during recovery,
	while maintaining the same level of safety by splitting the
	copy operation in multiple steps.
	Barman will only use the --checksum option on files having identical
	time and size that have been modified after the start of the backup.
	This change greatly improves the speed of "incremental" recovery.

	Minor changes:
	- disable --checksum even for backup. During a backup the rsync
	  destination directory is empty, so it is safe to go with a plain
	  rsync
	- Put a ".barman-recover.info" with backup metadata inside the
	  destination directory during recover.

	Use Postgres' server time for both begin_time and end_time
	Minor changes:
	- make sure exceptions during backup are logged with stacktraces
	- commit on disk the backup status just after issuing the PostgreSQL
	  start_backup command

	Change version to 1.3.3-alpha.1

2014-06-09  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Added fsync() for backup and cron operations

2014-06-06  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix parsing of 'basebackup_retry_times' and 'basebackup_retry_sleep' options

2014-05-30  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Fix for #43 recovery.conf not copied on remote recovery

2014-05-08  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Retry option for base backup
	If a network error happens during rsync, add the ability
	to retry a defined number of time.

	Two options have been added:
	 * basebackup_retry_times: INT (> 0, default 1)
	     maximum number or retry before giving up
	 * basebackup_retry_sleep: INT (> 0, default 10)
	     wait time (seconds) before retrying, after an error

2014-05-29  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Improve robustness of backup code
	Improve error message about stop_backup failure

2014-04-23  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	fixed missing pre/post archive parameters. #41 on sourceforge

2014-04-15  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Update unit tests to match current rsync flags

2014-04-15  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Prepared source code for version 1.3.2

2014-04-15  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Added checks for pg_extension (>= 9.1) and pg_is_in_recovery (>= 9.0)

2014-04-11  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

2014-04-10  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Always pass --checksum to rsync invocations

	Emit a warning if backup_options is set to an invalid value

	Clarify some "permission denied" error

2014-04-08  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Cosmetic change: Pgespresso -> pgespresso

2014-04-07  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update RPM spec file for 1.3.1

2014-04-04  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Prepared documentation for version 1.3.1

2014-04-04  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix 'barman diagnose' python3 support

	Improved logging and error reporting

2014-04-03  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Fixed SourceForge bug #36: Unhandled exception for minimum redundancy

2014-04-03  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Empty strings are now treated as None in Barman configuration

2014-04-02  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Removed spurious "file not found" message in cron output

	Add release information to 'barman diagnose'

	Sort 'barman show-server' output

	Use a Tablespace object to carry tablespace information

2014-03-26  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Protect during recovery tablespaces inside PGDATA
	* When performing a recovery operation, tablespaces that will be recovered
	  inside the new destination directory (PGDATA) are be 'protected' by
	  rsync. This avoids overwrites by rsync when copying PGDATA content.
	* Add debug messages to FS class

2014-03-24  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Implementation of 'barman diagnose' command (JSON output)

2014-03-21  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Concurrent backup using the 'pgespresso' extension
	* Fix bwlimit tablespaces backup (missing destination directory)
	* Purge unused wal files at first backup in concurrent mode
	* Exclusion of recovery.conf during backup

2014-03-19  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix unhandled exception in recover when destination dir is not writable

2014-02-19  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Make -q command line switch working again
	Also demote "another cron is running" message from error to info level.

2014-02-02  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Update RPM spec file for release 1.3.0

	Review of NEWS and AUTHORS files

2014-01-31  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Updated files for final release

2014-01-30  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Improve error messages during remote recovery

2014-01-29  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Use fsync to avoid xlog.db file corruption (Closes #32)

	Add network_compression configuration option (Closes #19)
	When network_compression is enabled, all network transfers are done
	using compression (if available).

2014-01-29  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Check directories exist before executing a backup (#14)

2014-01-28  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Reduce log verbosity during initialisation phase

2014-01-28  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Load configuration files after logger initialisation

2014-01-21  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Avoid tablespaces inside pgdata directory from being copied twice

2014-01-09  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Generalise recovery operations (local/remote)

2014-01-28  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Reviewed documentation of WAL archive hook scripts

2014-01-07  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add pre_archive_script and post_archive_scrip hook scripts

2014-01-23  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Refactor the LockFile management class to report permission errors.

	Fix 'Invalid cross-device link' error in cron when incoming is on a different filesystem (merge request #4 by Holger Hamann)

2014-01-22  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Port 'show-server' command to the new output interface

2014-01-21  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Updated copyright (2014)

2014-01-17  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Port 'status' and 'list-server' commands to the new output interface

2014-01-09  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Port the 'show-backup' command to the new output interface

2014-01-16  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Added implementation for backup command --immediate-checkpoint option and immediate_checkpoint configuration option

2014-01-08  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Bump version number and add release notes for 1.3.0

2013-11-27  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	Add unit tests for infofile and compression modules
	Fix some python3 compatibility bugs highlighted by the tests

2013-10-18  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Move barman._pretty_size() to barman.utils.pretty_size()

2014-01-03  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Implement BackupInfo as a FieldListFile and move it in infofile module.

2014-01-07  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Refactor output to a dedicate module.
	The following commands have been ported to the new interface:

	 * backup
	 * check
	 * list-backup

	A special NagiosOutputWriter has been added to support Nagios compatible
	output for the check command

	WARNING: this code doesn't run due to a circular dependency. The issue
	will be fixed in the next commit

2013-09-12  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Isolate subrocesses' stdin/stdout in command_wrappers module

2014-01-07  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Refactor hooks management

2013-09-12  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Split out logging configuration and userid enforcement from the configuration class.

2013-12-16  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Added rebuild-xlogdb command man page

2013-11-08  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Implement the rebuild-xlogdb command. (Closes #27)

2013-11-19  Giulio Calacoci  <giulio.calacoci@2ndquadrant.it>

	added documentation for tablespaces relocation (#22)

2013-10-30  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Added TODO list

2013-09-05  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Bump version to 1.2.3

2013-08-29  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Updated README and man page

	Added stub of release notes

2013-08-26  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Initial Python 3 support

	Update setup.py to support py.test and recent setuptools

2013-08-24  Damon Snyder  <damonsnyder@users.sourceforge.net>

	27: Addresses potential corruption of WAL xlog.db files.
	In barman.lockfile.release() the file is unlinked (deleted). This effectively
	nullifies any future attempts to lock the file by a blocking process by deleting
	the open file table entry upon which the flock is based.

	This commit removes the unlink and instead unlocks the file and then closes the file
	descriptor leaving the lock file and open file table entry intact.

2013-08-22  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add support for restore target name (PostgreSQL 9.1+)

2013-08-21  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	PostgreSQL version in backup.info file is an integer

	Make WAL sequence calculation compatible with PostgreSQL 9.3
	With PostgreSQL 9.3 WAL files are written in a continuous stream,
	rather than skipping the last 16MB segment every 4GB, meaning WAL
	filenames may end in FF.

2013-06-24  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Fix config file parser tests

	Bump version to 1.2.2

	Fix python 2.6 compatibility

	Fix history in spec file

2013-06-17  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update RPM spec file

2013-06-13  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Fix remote recovery with bwlimit on a tablespace

2013-06-07  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Added the "tablespace_bandwidth_limit" option

2013-06-12  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Updated docs and man pages for 1.2.1

	Prepared NEWS file for 1.2.1 release

2013-04-26  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Added the "bandwidth_limit" global/server option which allows to limit the I/O bandwidth (in KBPS) for backup and recovery operations

	Added /etc/barman/barman.conf as default location

2013-03-13  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Removed duplicate message for previous backup in show command

2013-03-07  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Cosmetic change in message for "all" reserved section

2013-02-08  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Avoid triggering the minimum_redundancy check on FAILED backups

	Add BARMAN_VERSION to hook script environment

2013-01-31  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Update RPM's spec files

2013-01-30  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Finalised files for version 1.2.0

2013-01-28  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Forbid the usage of 'all' word as server name

2013-01-11  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Added basic support for Nagios plugin output for check command through the --nagios option

2013-01-28  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add @expects_obj decorator to cli function as required by the upcoming Argh 1.0 API

2013-01-11  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Migratte to new argh api.
	Now barman requires arg => 0.21.2 and argcomplete-

2013-01-11  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Prepared release notes

2012-12-18  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix typo in doc/barman.conf

2012-12-14  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Return failure exit code if backup command fails in any way

2012-12-14  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Prepared copyright lines for 2013

	Updated documentation and man pages

	Added retention policy examples in configuration file

2012-12-13  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Q/A on retention policy code

2012-12-12  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix configuration parser unit tests

	Exit with error if an invalid server name is passed in any command which takes a list of server

2012-12-08  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add retention status to show-backup and list-backup commands

	Auto-management of retention policies for base backups
	Using the report() method for retention policies, enforce retention
	policy through cron (if policy mode is 'auto'), by deleting OBSOLETE
	backups.

	Retention status and report() method for retention policies
	Created the following states for retention policies:
	VALID, OBSOLETE, NONE and POTENTIALLY_OBSOLETE (an
	object which is OBSOLETE but cannot be removed
	automatically due to minimum_redundancy requirements).
	Created the report() method for the retention policy
	base class, which exected the _backup_report() method
	for base backups and the _wal_report() method for WAL
	retention policies (currently not enforced).
	The report method iterates through the DONE backups
	and according to the retention policy, classifies
	the backup. RedundancyRetentionPolicy uses the number
	of backups, RecoveryWindowRetentionPolicy uses the
	time window and the recoverability point concept.

	Integrated minimum_redundancy with "barman check"

	Initialisation of retention policies for a server
	Added the _init_retention_policies() method in the
	Server class constructor, which integrates with
	the new RetentionPolicy classes and performs
	syntax checking.
	Integrated retention policies with log, 'barman check'
	and 'barman status'.

	String representation conforms to retention syntax
	The string representation produces now a syntax-valid
	retention policy configuration string.
	The previous __str__ method has been renamed into debug()
	SimpleWALRetentionPolicy objects are now created from
	the server's main retention policy by the factory class.

2012-12-07  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Add the global/server option minimum_redundancy. Check it is >= 0. Guarantees that when delete is performed (or retention policies are enforced), this is the minimum number of backups to be kept for that server.

	Add support for retention_policy_mode global/server option which defines the method for enforcing retention policies (currently only "auto", in future versions "manual" will be allowed)

	Added first stub of retention policy classes

	Started version 1.2.0

2012-12-04  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix unit config tests

	Update the ChangeLog file

	Add ssl_*_file and unix_socket_directory to dangerous options list

	Display tablespace's oid in show-backup output

	Alphabetically sort servers in all commands output

	Don't give up on first error in 'barman check all' command

2012-12-03  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Added sorting of files in configuration directory

2012-11-29  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix regression in barman check command when configuration_files_directory is None

	Update rpm files to 1.1.2 release

2012-11-29  Carlo Ascani  <carlo.ascani@2ndquadrant.it>

	Update README

2012-11-29  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Prepared files for release

2012-11-28  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Add the configuration_files_directory option which allows to include multiple files from a directory

2012-11-29  Carlo Ascani  <carlo.ascani@2ndquadrant.it>

	Update README

2012-11-28  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update NEWS file

2012-11-05  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Added support for list-backup all

2012-11-04  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Added latest/oldest for show-backup, delete, list-files and recover commands

	Added get_first_backup and get_last_backup functions to Server class

	Added application_name management for PostgreSQL >= 9.0

2012-11-13  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Switched to version 1.1.2

	Continue if a WAL file is not found during delete (bug #18)

2012-11-04  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Includes version 90200 for tablespace new function

2012-10-16  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Update NEWS file and rpm package

	Bump version to 1.1.1

	Add more information about the failing line in xlogdb_parse_line errors

2012-10-15  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fix two bug on recover command

2012-10-12  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Update the ChangeLog file

	Update rpm changelog

	Make recover fail if an invalid tablespace relocation rule is given

	Remove unused imports from cli.py

2012-10-11  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Updated version to 1.1.0

	Fixes bug #12

2012-10-11  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Fail fast on recover command if the destination directory contains the ':' character (Closes: #4)

	Fix typo in recovery messages

	Report an informative message when pg_start_backup() invocation fails because an exclusive backup is already running (Closes: #8)

	Make current_action an attribute of BackupManager class

2012-10-08  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Added ticket #10 to NEWS

	Add pg_config_detect_possible_issues function for issue #10

2012-10-04  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Updated NEWS file with bug fixing #9

	Fixes issue #9 on pg_tablespace_location() for 9.2

2012-08-31  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add BARMAN_PREVIOUS_ID variable to hooks environment

2012-08-20  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Merge spec changes from Devrim

	Add BARMAN_ERROR and BARMAN_STATUS variables to hook's environment

	Added backup all documentation to README

2012-08-20  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Updated release notes

	Set version to 1.0.1

2012-08-20  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Document {pre,post}_backup_script in README

	Document {pre,post}_backup_script in configuration man-page

2012-08-17  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add pre/post backup hook scripts definition (Closes: #7)
	Add the possibility to manage hook scripts before and after a base
	backup. Add the global (overridden per server) configuration options
	called:

	* pre_backup_script: executed before a backup
	* post_backup_script: executed after a backup

	Use the environment to pass at least the following variabiles:

	* BARMAN_BACKUP_DIR: backup destination directory
	* BARMAN_BACKUP_ID: ID of the backup
	* BARMAN_CONFIGURATION: configuration file used by barman
	* BARMAN_PHASE: 'pre' or 'post'
	* BARMAN_SERVER: name of the server

	The script definition is passed to the shell and can return any exit code.

	Barman won't perform any exit code check. It will simply log the result in the log file.

	To test it you can try adding

	pre_backup_script = env | grep ^BARMAN
	post_backup_script = env | grep ^BARMAN

	in your barman config and you'll see the variables on console.

2012-08-16  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Add documentation for 'backup all' command.

2012-07-19  Gabriele Bartolini  <gabriele.bartolini@2ndquadrant.it>

	Add 'backup all' shortcut and, in general, multiple servers specification (issue #1)

	Add 'backup all' shortcut and, in general, multiple servers specification (issue #1)

2012-07-16  Gabriele Bartolini  <gabriele.bartolini@2ndQuadrant.it>

	Fixed typo (thanks to Daymel Bonne Solís)

2012-07-06  Marco Nenciarini  <marco.nenciarini@2ndquadrant.it>

	Initial commit