The following log levels are supported:

The following log levels are supported:

Specifies which log levels will output to stderr rather than stdout (specified by log-level-console). The timestamp and process will not be output to stderr.

The following log levels are supported:

The log path provides a location for to store log files. Note that if log-level-file=off then no log path is required.

Enables the timestamp in console and file logging. This option is disabled in special situations such as generating documentation.

Enable file logging for any subprocesses created by this process using the log level specified by log-level-file.

By default only the restore command may be run as the root user since it is designed to carefully manage file ownership. Running other commands as root risks creating files (e.g. in the repository) that are owned by root and therefore inaccessible to the PostgreSQL user, causing later commands to fail.

Enable this option to run a command as root anyway. However, it is far better to run as the user that owns the repository and PostgreSQL cluster.

Allow beta features to be enabled but do not actually enable them. Beta features need to be separately enabled using related options.

Buffer size used for copy, compress, encrypt, and other operations. The number of buffers used depends on options and each operation may use additional memory, e.g. gz compression may use an additional 256KiB of memory.

Allowed values are 16KiB, 32KiB, 64KiB, 128KiB, 256KiB, 512KiB, 1MiB, 2MiB, 4MiB, 8MiB, and 16MiB.

may generate a command string, e.g. when the restore command generates the restore_command setting. The command used to run the process will be used in this case unless the cmd option is provided.

Use a specific SSH client command when an alternate is desired or the ssh command is not in $PATH.

Backup files are compatible with command-line compression tools.

This option is now deprecated. The compress-type option should be used instead.

The following compression types are supported:

Sets the level to be used for file compression when compress-type does not equal none or compress=y (deprecated).

Sets the network compression level when compress-type=none and the command is not run on the same host as the repository. Compression is used to reduce network traffic. When compress-type does not equal none the compress-level-network setting is ignored and compress-level is used instead so that the file is only compressed once.

Sets the timeout, in seconds, for queries against the database. This includes the backup start/stop functions which can each take a substantial amount of time. Because of this the timeout should be kept high unless you know that these functions will return quickly (i.e. if you have set start-fast=y and you know that the database cluster will not generate many WAL segments during the backup).

During a restore, by default the data and tablespace directories are expected to be present but empty. This option performs a delta restore using checksums.

During a backup, this option will use checksums instead of the timestamps to determine if files will be copied.

Timeout, in seconds, used for connections and read/write operations.

Note that the entire read/write operation does not need to complete within this timeout but some progress must be made, even if it is only a single byte.

The number of retries for jobs run in the local process, e.g. file backup.

The first retry is always performed immediately. This option specifies the interval to wait (in seconds) before performing each additional retry.

The lock path provides a location for to create lock files to prevent conflicting operations from being run concurrently.

Sets the umask to 0000 so modes in the repository are created in a sensible way. The default directory mode is 0750 and default file mode is 0640.

To use the executing user's umask instead specify neutral-umask=n in the config file or --no-neutral-umask on the command line.

Defines how much priority (i.e. niceness) will be given to the process by the kernel scheduler. Positive values decrease priority and negative values increase priority. In most case processes do not have permission to increase their priority.

This path is used to store data for the asynchronous archive-push and archive-get command.

The asynchronous archive-push command writes acknowledgements into the spool path when it has successfully stored WAL in the archive (and errors on failure) so the foreground process can quickly notify . Acknowledgement files are very small (zero on success and a few hundred bytes on error).

The asynchronous archive-get command queues WAL in the spool path so it can be provided very quickly when requests it. Moving files to is most efficient when the spool path is on the same filesystem as pg_xlog/pg_wal. However, it is not recommended to place the spool path within the pg_xlog/pg_wal directory as this may cause issues for utilities such as pg_rewind.

The data stored in the spool path is not strictly temporary since it can and should survive a reboot. However, loss of the data in the spool path is not a problem. will simply recheck each WAL segment to ensure it is safely archived for archive-push and rebuild the queue for archive-get.

The spool path is intended to be located on a local Posix-compatible filesystem, not a remote filesystem such as NFS or CIFS.

Each process will perform compression and transfer to make the command run faster, but don't set process-max so high that it impacts database performance.

Sets the timeout, in seconds, that the local or remote process will wait for a new message to be received on the protocol layer. This prevents processes from waiting indefinitely for a message.

Enable blocking on socket connections.

Enables keep-alive messages on socket connections.

Specifies the number of TCP keep-alive messages that can be lost before the connection is considered dead.

This option is available on systems that support the TCP_KEEPCNT socket option.

Specifies the amount of time (in seconds) with no network activity after which the operating system should send a TCP keep-alive message.

This option is available on systems that support the TCP_KEEPIDLE socket option.

Specifies the amount of time (in seconds) after which a TCP keep-alive message that has not been acknowledged should be retransmitted.

This option is available on systems that support the TCP_KEEPINTVL socket option.

All TLS connections between the client and server are encrypted. By default, connections to objects stores (e.g. S3) are also encrypted.

The accepted cipher suites can be adjusted if need arises. The example is reasonable choice unless you have specific security requirements. If unset (the default), the default of the underlying OpenSSL library applies.

All TLS connections between the client and server are encrypted. By default, connections to objects stores (e.g. S3) are also encrypted.

The accepted cipher suites can be adjusted if need arises. If unset (the default), the default of the underlying OpenSSL library applies.

Enabled by default, this option checks the WAL header against the version and system identifier to ensure that the WAL is being copied to the correct stanza. This is in addition to checking pg_control against the stanza and verifying that WAL is being copied from the same data directory where pg_control is located.

Therefore, disabling this check is fairly safe but should only be done when needed, e.g. if the WAL is encrypted.

Enabled by default, this option adds page header checks.

Disabling this option should be avoided except when necessary, e.g. if pages are encrypted.

The specified version will be used instead of the version automatically detected by reading pg_control or WAL headers. This is mainly useful for forks or development versions where those values are different from the release version. The version reported by via `server_version_num` must match the forced version.

Azure account used to store the repository.

Azure container used to store the repository.

repositories can be stored in the container root by setting repo-path=/ but it is usually best to specify a prefix, such as /repo, so logs and other Azure-generated content can also be stored in the container.

Endpoint used to connect to the blob service. The default is generally correct unless using Azure Government.

For custom/test configurations the repo-storage-ca-file, repo-storage-ca-path, repo-storage-host, repo-storage-port, and repo-storage-verify-tls options may be useful.

A shared key or shared access signature depending on the repo-azure-key-type option.

The following types are supported for authorization:

The following URI styles are supported:

Block incremental allows for more granular backups by splitting files into blocks that can be backed up independently. This saves space in the repository and can improve delta restore performance because individual blocks can be fetched without reading the entire file from the repository.

The block size for a file is determined based on the file size and age. Generally, older/larger files will get larger block sizes. If a file is old enough, it will not be backed up using block incremental.

Block incremental is most efficient when enabled for all backup types, including full. This makes the full a bit larger but subsequent differential and incremental backups can make use of the block maps generated by the full backup to save space.

Map file age (in days) to a block multiplier. Files that have not been modified recently are less likely to be modified in the future, so the block size is multiplied to reduce the map size. By default, if the file is old enough it will not be stored as a block incremental.

Map block size to checksum size. Smaller checksums save space in the map but may not be able to reliably detect changes in the block.

Map file size to block size. Block size is the minimum unit that will be stored in the block incremental. Smaller sizes allow for more granular incremental backups but larger sizes mean smaller maps.

A super block contains multiple blocks to improve compression efficiency. The disadvantage is that block reads must read from the beginning of the super block.

Super block size for full backups. This is generally larger than repo-block-size-super because it is likely that many blocks will need to be restored from the full backup so the larger super block size is not a disadvantage.

Bundle (combine) smaller files to reduce the total number of files written to the repository. Writing fewer files is generally more efficient, especially on object stores such as S3. In addition, zero-length files are not stored (except in the manifest), which saves time and space.

Defines the target size for files that will be added to a single bundle. The uncompressed bundle size may be as large as repo-bundle-size + repo-bundle-limit, so do not set this option to the maximum size that your file system allows.

In general, it is not a good idea to set this option too high because retries will need to redo the entire bundle.

Size limit for files that will be included in bundles. Files larger than this size will be stored separately.

Bundled files cannot be reused when a backup is resumed, so this option controls the files that can be resumed, i.e. higher values result in fewer resumable files.

GCS bucket used to store the repository.

repositories can be stored in the bucket root by setting repo-path=/ but it is usually best to specify a prefix, such as /repo, so logs and other GCS-generated content can also be stored in the bucket.

Endpoint used to connect to the storage service. May be updated to use a local GCS server or alternate endpoint.

A token or service key file depending on the repo-gcs-key-type option.

The following types are supported for authorization:

When repo-gcs-key-type=service the credentials will be reloaded when the authentication token is renewed.

GCS project ID used to determine request billing.

The following protocol types are supported:

When backing up and archiving to a locally mounted filesystem this setting is not required.

Use a CA file other than the system default for connecting to the repository host.

Use a CA path other than the system default for connecting to the repository host.

Sent to repository host to prove client identity.

Proves client certificate was sent by owner.

Required only if the path to the command is different on the local and repository hosts. If not defined, the repository host command will be set the same as the local command.

Sets the location of the configuration file on the repository host. This is only required if the repository host configuration file is in a different location than the local configuration file.

Sets the location of the configuration include path on the repository host. This is only required if the repository host configuration include path is in a different location than the local configuration include path.

Sets the location of the configuration path on the repository host. This is only required if the repository host configuration path is in a different location than the local configuration path.

Defines the user that will be used for operations on the repository host. Preferably this is not the postgres user but rather some other user like pgbackrest. If runs on the repository host the postgres user can be placed in the pgbackrest group so it has read permissions on the repository without being able to damage the contents accidentally.

Use this option to specify a non-default port for the repository host protocol.

Enable hard-linking of files in differential and incremental backups to their full backups. This gives the appearance that each backup is a full backup at the file-system level. Be careful, though, because modifying files that are hard-linked can affect all the backups in the set.

Indicates that the repository for the specified index is local and any repository host settings for that index (e.g. repo1-host) should be ignored.

The repository is where stores backups and archives WAL segments.

It may be difficult to estimate in advance how much space you'll need. The best thing to do is take some backups then record the size of different types of backups (full/incr/diff) and measure the amount of WAL generated per day. This will give you a general idea of how much space you'll need, though of course requirements will likely change over time as your database evolves.

When a full backup expires, all differential and incremental backups associated with the full backup will also expire. When the option is not defined a warning will be issued. If indefinite retention is desired then set the option to the max value.

Determines whether the repo-retention-full setting represents a time period (days) or count of full backups to keep.

If set to time then full backups older than repo-retention-full will be removed from the repository if there is at least one other backup that is equal to or greater than the repo-retention-full setting. For example, if repo-retention-full is 30 (days) and there are 2 full backups: one 25 days old and one 35 days old, no full backups will be expired because expiring the 35 day old backup would leave only the 25 day old backup, which would violate the 30 day retention policy of having at least one backup 30 days old before an older one can be expired. Archived WAL older than the oldest full backup remaining will be automatically expired unless repo-retention-archive-type and repo-retention-archive are explicitly set.

If set to count then full backups that exceed repo-retention-full will be expired. For example, if repo-retention-full is 4 and a fifth full backup is completed, then the oldest full backup will be expired to keep the count at 4.

Note that a backup must be successfully completed before it will be considered for retention. For example, if repo-retention-full-type is count and repo-retention-full is 2, then there must be 3 complete full backups before the oldest will be expired.

When a differential backup expires, all incremental backups associated with the differential backup will also expire. When not defined all differential backups will be kept until the full backups they depend on expire.

Note that full backups are included in the count of differential backups for the purpose of expiration. This slightly reduces the number of differential backups that need to be retained in most cases.

If set to full will keep archive logs for the number of full backups defined by repo-retention-archive. If set to diff (differential) will keep archive logs for the number of full and differential backups defined by repo-retention-archive, meaning if the last backup taken was a full backup, it will be counted as a differential for the purpose of repo-retention. If set to incr (incremental) will keep archive logs for the number of full, differential, and incremental backups defined by repo-retention-archive. It is recommended that this setting not be changed from the default which will only expire WAL in conjunction with expiring full backups.

If this value is not set and repo-retention-full-type is count (default), then the archive to expire will default to the repo-retention-full (or repo-retention-diff) value corresponding to the repo-retention-archive-type if set to full (or diff). This will ensure that WAL is only expired for backups that are already expired. If repo-retention-full-type is time, then this value will default to removing archives that are earlier than the oldest full backup retained after satisfying the repo-retention-full setting.

This option must be set if repo-retention-archive-type is set to incr. If disk space is at a premium, then this setting, in conjunction with repo-retention-archive-type, can be used to aggressively expire WAL segments. However, doing so negates the ability to perform PITR from the backups with expired WAL and is therefore not recommended.

A copy of the backup manifest is stored in the backup.history path when a backup completes. By default these files are never expired since they are useful for data mining, e.g. measuring backup and WAL growth over time.

Set repo-retention-history to define the number of days of backup history manifests to retain. Unexpired backups are always kept in the backup history. Specify repo-retention-history=0 to retain the backup history only for unexpired backups.

When a full backup history manifest is expired, all differential and incremental backup history manifests associated with the full backup also expire.

AWS key used to access this bucket.

AWS secret key used to access this bucket.

The following types are supported:

AWS security token used with temporary credentials.

Enables S3 server-side encryption using the specified AWS key management service key.

S3 bucket used to store the repository.

repositories can be stored in the bucket root by setting repo-path=/ but it is usually best to specify a prefix, such as /repo, so logs and other AWS generated content can also be stored in the bucket.

The AWS endpoint should be valid for the selected region.

For custom/test configurations the repo-storage-ca-file, repo-storage-ca-path, repo-storage-host, repo-storage-port, and repo-storage-verify-tls options may be useful.

Command (and optional arguments) to execute for retrieving temporary S3 credentials. The first list entry is the command and the remaining entries are passed as parameters.

The process must output JSON containing AccessKeyId, SecretAccessKey, SessionToken, and Expiration fields. Credentials will be automatically refreshed before the expiration time. See Process Credential Provider for format details.

The AWS role name (not the full ARN) used to retrieve temporary credentials when repo-s3-key-type=auto.

The AWS region where the bucket was created.

Enables S3 server-side encryption using the specified customer key.

Enables S3 requester pays.

The S3 signing service used in SigV4 authentication. Defaults to s3 for standard S3 endpoints. Set to s3-outposts when using an S3 Outposts endpoint.

The STS endpoint used to retrieve temporary credentials when repo-s3-key-type=web-id is configured. Set to a regional endpoint (e.g. sts.us-east-1.amazonaws.com) to use regional STS, which may be required for GovCloud, China regions, or to reduce latency.

The following URI styles are supported:

The SFTP host containing the repository.

SFTP repository host fingerprint generation should match the repo-sftp-host-key-hash-type. Generate the fingerprint via awk '{print $2}' ssh_host_xxx_key.pub | base64 -d | (md5sum or sha1sum) -b. The ssh host keys are normally found in the /etc/ssh directory.

The following SFTP host key check types are supported:

SFTP repository host key hash type. Declares the hash type to be used to compute the digest of the remote system's host key on SSH startup. Newer versions of libssh2 support sha256 in addition to md5 and sha1.

SFTP repository host port.

User on the host used to store the repository.

A known hosts file to search for an SFTP host match during authentication. When unspecified, will default to searching ~/.ssh/known_hosts, ~/.ssh/known_hosts2, /etc/ssh/ssh_known_hosts, and /etc/ssh/ssh_known_hosts2. If configured with one or more file paths, will search those for a match. File paths must be full or leading tilde paths. The repo-sftp-known-host option can be passed multiple times to specify more than one known hosts file to search. To utilize known hosts file checking repo-sftp-host-fingerprint must not be specified. See also repo-sftp-host-check-type option.

SFTP private key file used for authentication.

Passphrase used to access the private key. This is an optional feature when creating an SSH public/private key pair.

SFTP public key file used for authentication. Optional if compiled against OpenSSL, required if compiled against a different library.

Use a CA file other than the system default for storage (e.g. S3, Azure) certificates.

Use a CA path other than the system default for storage (e.g. S3, Azure) certificates.

Determines the number of requests to prefetch when reading from object stores, e.g. S3. This is not true concurrency but it reduces the latency involved in submitting requests.

Note that the number of open requests per read will be one greater than prefetch, since the initial read cannot be prefetched.

Connect to a host other than the storage (e.g. S3, Azure) endpoint. This is typically used for testing.

Port to use when connecting to the storage (e.g. S3, Azure) endpoint (or host if specified).

Bundles may contain data that is not needed during restore, for example files that already exist. This option skips over unneeded data without starting a new read request on object stores. This is generally faster since the latency on new read requests is higher than continuing a read already in progress. It is also less expensive for most storage classes.

Specify tags that will be added to objects when the repository is an object store (e.g. S3). The option can be repeated to add multiple tags.

There is no provision in to modify these tags so be sure to set them correctly before running stanza-create to ensure uniform tags across the entire repository.

Object stores such as S3 allow files to be uploaded in chunks when the file is too large to be stored in memory. Even if the file can be stored in memory, it is more memory efficient to limit the amount of memory used for uploads.

A larger chunk size will generally lead to better performance because it will minimize upload requests and allow more files to be uploaded in a single request rather than in chunks. The disadvantage is that memory usage will be higher and because the chunk buffer must be allocated per process, larger process-max values will lead to more memory being consumed overall.

Note that valid chunk sizes vary by storage type and by platform. For example, AWS S3 has a minimum chunk size of 5MiB. Terminology for chunk size varies by storage type, so when searching min/max values use part size for AWS S3, chunk size for GCS, and block size for Azure.

If a file is larger than 1GiB (the maximum size will create by default) then the chunk size will be increased incrementally up to the maximum allowed in order to complete the file upload.

This option provides the ability to enable/disable verification of the storage (e.g. S3, Azure) server TLS certificate. Disabling should only be used for testing or other scenarios where a certificate has been self-signed.

Enable creation of the lastest and tablespace symlinks. These symlinks are most useful when using snapshots to do in-place recovery in the repository, which is an uncommon use case.

While this feature is likely not useful for the vast majority of users it remains on by default for legacy purposes. However, it may be useful to disable symlinks for Posix-like storage that does not support them.

The target time defines the time that commands use to read a repository on versioned storage. This allows the command to read the repository as it was at a point-in-time in order to recover data that has been deleted or corrupted by user accident or malware.

Versioned storage is supported by S3, GCS, and Azure but is generally not enabled by default. In addition to enabling versioning, it may be useful to enable object locking for S3 and soft delete for GCS or Azure.

When the repo-target-time option is specified then the repo option must also be provided. It is likely that not all repository types will support versioning and in general it makes sense to target a single repository for recovery.

Note that comparisons to the storage timestamp are <= the timestamp provided and milliseconds are truncated from the timestamp when provided.

The following repository types are supported:

When an NFS mount is used as a posix repository, the same rules apply to as described in the documentation: Creating a Database Cluster - File Systems.

The following cipher types are supported:

Note that encryption is always performed client-side even if the repository type (e.g. S3) supports encryption.

Passphrase used to encrypt/decrypt files of the repository.

Users can attach informative key/value pairs to the backup. This option may be used multiple times to attach multiple annotations.

Annotations are output by the info command text output when a backup is specified with --set and always appear in the JSON output.

Checks that all WAL segments required to make the backup consistent are present in the WAL archive. It's a good idea to leave this as the default unless you are using another method for archiving.

This option must be enabled if archive-copy is enabled.

This slightly paranoid option protects against corruption in the WAL segment archive by storing the WAL segments required for consistency directly in the backup. WAL segments are still stored in the archive so this option will use additional space.

It is best if the archive-push and backup commands have the same compress-type (e.g. lz4) when using this option. Otherwise, the WAL segments will need to be recompressed with the compress-type used by the backup, which can be fairly expensive depending on how much WAL was generated during the backup.

On restore, the WAL segments will be present in pg_xlog/pg_wal and will use them in preference to calling the restore_command.

The archive-check option must be enabled if archive-copy is enabled.

Enabled by default, this option disallows archive_mode=always.

WAL segments pushed from a standby server might be logically the same as WAL segments pushed from the primary but have different checksums. Disabling archiving from multiple sources is recommended to avoid conflicts.

Enable backup from standby to reduce load on the primary cluster. This option requires that both the primary and standby hosts be configured.

The following modes are supported:

Directs to validate all data page checksums while backing up a cluster. This option is automatically enabled when data page checksums are enabled on the cluster.

Failures in checksum validation will not abort a backup. Rather, warnings will be emitted in the log (and to the console with default settings) and the list of invalid pages will be stored in the backup manifest.

All exclusions are relative to $PGDATA. If the exclusion ends with / then only files in the specified directory will be excluded, e.g. --exclude=junk/ will exclude all files in the $PGDATA/junk directory but include the directory itself. If the exclusion does not end with / then the file may match the exclusion exactly or match with / appended to the exclusion, e.g. --exclude=junk will exclude the $PGDATA/junk directory and all the files it contains.

Be careful using this feature -- it is very easy to exclude something critical that will make the backup inconsistent. Be sure to test your restores!

All excluded files will be logged at info level along with the exclusion rule. Be sure to audit the list of excluded files to ensure nothing unexpected is being excluded.

This option should not be used to exclude logs from a backup. Logs can be moved out of the PGDATA directory using the log_directory setting, which has the benefit of allowing logs to be preserved after a restore.

Multiple exclusions may be specified on the command-line or in a configuration file.

Defines how often the manifest will be saved during a backup. Saving the manifest is important because it stores the checksums and allows the resume function to work efficiently. The actual threshold used is 1% of the backup size or manifest-save-threshold, whichever is greater.

The setting is enabled by default. Use caution when disabling this option as doing so will result in retaining all backups and archives indefinitely, which could cause your repository to run out of space. The expire command will need to be run regularly to prevent this from happening.

When expire is run automatically after a successful backup it uses the configuration of the backup command, so options set only in an expire command section (e.g. [global:expire]) are not applied. To apply expire-specific configuration, disable this option and run the expire command separately.

Defines whether the resume feature is enabled. Resume can greatly reduce the amount of time required to run a backup after a previous backup of the same type has failed. It adds complexity, however, so it may be desirable to disable in environments that do not require the feature.

Forces a checkpoint (by passing y to the fast parameter of the backup start function) so the backup begins immediately. Otherwise the backup will start after the next regular checkpoint.

This will only be done if an exclusive advisory lock can be acquired to demonstrate that the prior failed backup process has really stopped.

This feature is not supported for >= 9.6 since backups are run in non-exclusive mode.

The setting is disabled by default because it assumes that is the only process doing exclusive online backups. It depends on an advisory lock that only sets so it may abort other processes that do exclusive online backups. Note that base_backup and pg_dump are safe to use with this setting because they do not call pg_start_backup() so are not exclusive.

Enables asynchronous operation for the archive-push and archive-get commands.

Asynchronous operation is more efficient because it can reuse connections and take advantage of parallelism. See the spool-path, archive-get-queue-max, and archive-push-queue-max options for more information.

Specifies the maximum size of the archive-get queue when archive-async is enabled. The queue is stored in the spool-path and is used to speed providing WAL to .

Retry a WAL segment that was previously reported as missing by the archive-get command when in asynchronous mode. This prevents notifications in the spool path from a prior restore from being used and possibly causing a recovery failure if consistency has not been reached.

Disabling this option allows to more reliably recognize when the end of the WAL in the archive has been reached, which permits it to switch over to streaming from the primary. With retries enabled, a steady stream of WAL being archived will cause to continue getting WAL from the archive rather than switch to streaming.

When disabling this option it is important to ensure that the spool path for the stanza is empty. The restore command does this automatically if the spool path is configured at restore time. Otherwise, it is up to the user to ensure the spool path is empty.

In asynchronous mode the archive-push process pushes all the WAL segments that are ready in a single run. Since archive-push-queue-max is only checked at the start of each run, a run that processes a very large number of segments can let the queue grow well beyond the limit before it is rechecked.

This option limits the amount of WAL processed per run so the process exits and is spawned again by the next archive-push, which rechecks the queue. Lower values recheck the queue more often at the cost of spawning the asynchronous process more frequently. The value is rounded down to a whole number of WAL segments but at least one segment is always processed.

After the limit is reached, the following will happen:

If this occurs then the archive log stream will be interrupted and PITR will not be possible past that point. A new backup will be required to regain full restore capability.

In asynchronous mode the entire queue will be dropped to prevent spurts of WAL getting through before the queue limit is exceeded again.

In asynchronous mode this limit is only checked at the start of each archive-push run, so the queue can grow beyond it within a single run. Reduce archive-push-batch-size to check the queue more frequently.

The purpose of this feature is to prevent the log volume from filling up at which point will stop completely. Better to lose the backup than have go down.

Set maximum time, in seconds, to wait for each WAL segment to reach the archive repository. The timeout applies to the check and backup commands when waiting for WAL segments required for backup consistency to be archived.

This option allows archiving to be preserved or disabled on a restored cluster. This is useful when the cluster must be promoted to do some work but is not intended to become the new primary. In this case it is not a good idea to push WAL from the cluster into the repository.

The following modes are supported:

NOTE: This option is not available on < 12.

Databases excluded will be restored as sparse, zeroed files to save space but still allow to perform recovery. After recovery, those databases will not be accessible but can be removed with the drop database command. The {[dash]}-db-exclude option can be passed multiple times to specify more than one database to exclude.

When used in combination with the {[dash]}-db-include option, {[dash]}-db-exclude will only apply to standard system databases (template0, template1, and postgres).

This feature allows only selected databases to be restored. Databases not specifically included will be restored as sparse, zeroed files to save space but still allow to perform recovery. After recovery, the databases that were not included will not be accessible but can be removed with the drop database command.

The {[dash]}-db-include option can be passed multiple times to specify more than one database to include.

See Restore Selected Databases for additional information and caveats.

By default symlinked directories and files are restored as normal directories and files in $PGDATA. This is because it may not be safe to restore symlinks to their original destinations on a system other than where the original backup was performed. This option restores all the symlinks just as they were on the original system where the backup was performed.

Allows the destination file or path of a symlink to be changed on restore. This is useful for restoring to systems that have a different storage layout than the original system where the backup was generated.

See Server Configuration for details on postgresql.auto.conf or recovery.conf options (be sure to select your version). This option can be used multiple times.

For >= 12, options will be written into postgresql.auto.conf. For all other versions, options will be written into recovery.conf.

Since does not start after writing the postgresql.auto.conf or recovery.conf file, it is always possible to edit/check postgresql.auto.conf or recovery.conf before manually restarting.

Moves a tablespace to a new location during the restore. This is useful when tablespace locations are not the same on a replica, or an upgraded system has different mount points.

Tablespace locations are not stored in pg_tablespace so moving tablespaces can be done with impunity. However, moving a tablespace to the data_directory is not recommended and may cause problems. For more information on moving tablespaces http://www.databasesoup.com/2013/11/moving-tablespaces.html is a good resource.

Tablespaces are restored into their original locations by default. This behavior can be modified for each tablespace with the tablespace-map option, but it is sometimes preferable to remap all tablespaces to a new directory all at once. This is particularly useful for development or staging systems that may not have the same storage layout as the original system where the backup was generated.

The path specified will be the parent path used to create all the tablespaces in the backup.

Clients are authorized on the server by verifying their certificate and checking their certificate CN (Common Name) against a list on the server configured with the tls-server-auth option.

A client CN can be authorized for as many stanzas as needed by providing a comma-separated list to the tls-server-auth option or for all stanzas by specifying tls-server-auth=client-cn=*. Wildcards may not be specified for the client CN.

Checks that client certificates are signed by a trusted certificate authority.

Sent to the client to show the server identity.

Proves server certificate was sent by the owner.

IP address the server will listen on for client requests.

Port the server will listen on for client requests.

Required only if the path to the command is different on the local and hosts. If not defined, the host command will be set the same as the local command.

Sets the location of the configuration file on the host. This is only required if the host configuration file is in a different location than the local configuration file.

Sets the location of the configuration include path on the host. This is only required if the host configuration include path is in a different location than the local configuration include path.

Sets the location of the configuration path on the host. This is only required if the host configuration path is in a different location than the local configuration path.

Used for backups where the host is different from the repository host.

Use a CA file other than the system default for connecting to the host.

Use a CA path other than the system default for connecting to the host.

Sent to host to prove client identity.

Proves client certificate was sent by owner.

The following protocol types are supported:

This user will also own the remote process and will initiate connections to . For this to work correctly the user should be the database cluster owner which is generally postgres, the default.

The database name used when connecting to . The default is usually best but some installations may not contain this database.

Note that for legacy reasons the setting of the PGDATABASE environment variable will be ignored.

Indicates that for the specified index is local and any host settings for that index (e.g. pg1-host) should be ignored.

This should be the same as the data_directory reported by . Even though this value can be read from various places, it is prudent to set it in case those resources are not available during a restore or offline backup scenario.

The pg-path option is tested against the value reported by on every online backup so it should always be current.

Port that is running on. This usually does not need to be specified as most clusters run on the default port.

The unix socket directory that was specified when was started. will automatically look in the standard location for your OS so there is usually no need to specify this setting unless the socket directory was explicitly modified with the unix_socket_directories setting in postgresql.conf.

The database user name used when connecting to . If not specified will connect with the local OS user or PGUSER.

Use this option to specify a non-default port for the host protocol.