patroni.postgresql package

Subpackages

patroni.postgresql.available_parameters package
- Module contents
patroni.postgresql.mpp package
- Submodules
  - patroni.postgresql.mpp.citus module
- Module contents

Submodules

Module contents

class patroni.postgresql.PgIsReadyStatus(*values)

Bases: IntEnum

Possible PostgreSQL connection status pg_isready utility can report.

Variables:

RUNNING – return code 0. PostgreSQL is accepting connections normally.
REJECT – return code 1. PostgreSQL is rejecting connections.
NO_RESPONSE – return code 2. There was no response to the connection attempt.
UNKNOWN – Return code 3. No connection attempt was made, something went wrong.

NO_RESPONSE = 2

REJECT = 1

RUNNING = 0

UNKNOWN = 3

class patroni.postgresql.Postgresql(config: Dict[str, Any], mpp: AbstractMPP)

Bases: object

POSTMASTER_START_TIME = 'pg_catalog.pg_postmaster_start_time()'

TL_LSN = "CASE WHEN pg_catalog.pg_is_in_recovery() THEN 0 ELSE ('x' || pg_catalog.substr(pg_catalog.pg_{0}file_name(pg_catalog.pg_current_{0}_{1}()), 1, 8))::bit(32)::int END, CASE WHEN pg_catalog.pg_is_in_recovery() THEN 0 ELSE pg_catalog.pg_{0}_{1}_diff(pg_catalog.pg_current_{0}{2}_{1}(), '0/0')::bigint END, pg_catalog.pg_{0}_{1}_diff(pg_catalog.pg_last_{0}_replay_{1}(), '0/0')::bigint, pg_catalog.pg_{0}_{1}_diff(COALESCE(pg_catalog.pg_last_{0}_receive_{1}(), '0/0'), '0/0')::bigint, pg_catalog.pg_is_in_recovery() AND pg_catalog.pg_is_{0}_replay_paused()"

__init__(config: Dict[str, Any], mpp: AbstractMPP) → None

_before_stop() → None: Synchronously run a script prior to stopping postgres.

_cluster_info_state_get(name: str) → Any | None

_do_before_stop(cmd: str) → None

_do_stop(mode: str, block_callbacks: bool, checkpoint: bool, on_safepoint: Callable[[...], Any] | None, on_shutdown: Callable[[int, int], Any] | None, before_shutdown: Callable[[...], Any] | None, stop_timeout: int | None) → Tuple[bool, bool]

_get_gucs() → CaseInsensitiveSet

Get all available GUCs based on postgres --describe-config output.

Returns:: all available GUCs in the local Postgres server.

_pre_promote() → bool: Runs a fencing script after the leader lock is acquired but before the replica is promoted. If the script exits with a non-zero code, promotion does not happen and the leader key is removed from DCS.

_query(sql: str, *params: Any) → List[Tuple[Any, ...]]

Execute sql query with params and optionally return results.

Parameters:

sql – SQL statement to execute.
params – parameters to pass.

Returns:

a query response as a list of tuples if there is any.

Raises:

Error if had issues while executing sql.

PostgresConnectionException: if had issues while connecting to the database.

RetryFailedError: if it was detected that connection/query failed due to PostgreSQL restart.

_version_file_exists() → bool

_wait_for_connection_close(postmaster: PostmasterProcess) → None

_wait_promote(wait_seconds: int) → bool | None

static _wal_position(is_primary: bool, wal_position: int, receive_lsn: int | None, replay_lsn: int | None) → int

property available_gucs: CaseInsensitiveSet: GUCs available in this Postgres server.

call_nowait(cb_type: CallbackAction) → None: pick a callback command and call it without waiting for it to finish

property callback: Dict[str, str]

property can_advance_slots: bool: True if :attr:major_version is greater than 110000.

can_create_replica_without_replication_connection(replica_methods: List[str] | None) → bool: go through the replication methods to see if there are ones that does not require a working replication connection.

property cb_called: bool

check_for_startup() → bool: Checks PostgreSQL status and returns if PostgreSQL is in the middle of startup.

check_startup_state_changed() → bool

Checks if PostgreSQL has completed starting up or failed or still starting.

Should only be called when state == ‘starting [after custom bootstrap]’

Returns:: True if state was changed from ‘starting’

checkpoint(connect_kwargs: Dict[str, Any] | None = None, timeout: float | None = None) → str | None

property cluster_info_query: str

Returns the monitoring query with a fixed number of fields.

The query text is constructed based on current state in DCS and PostgreSQL version:

function names depend on version. wal/lsn for v10+ and xlog/location for pre v10.
for primary we query timeline_id (extracted from pg_walfile_name()) and pg_current_wal_lsn()
for replicas we query pg_last_wal_receive_lsn(), pg_last_wal_replay_lsn(), and pg_is_wal_replay_paused()
for v9.6+ we query primary_slot_name and primary_conninfo from pg_stat_get_wal_receiver()
for v11+ with permanent logical slots we query from pg_replication_slots and aggregate the result
for standby_leader node running v9.6+ we also query pg_control_checkpoint to fetch timeline_id
if sync replication is enabled we query pg_stat_replication and aggregate the result. In addition to that we get current values of synchronous_commit and synchronous_standby_names GUCs.

If some conditions are not satisfied we simply put static values instead. E.g., NULL, 0, ‘’, and so on.

configure_server_parameters() → None

connection() → connection3 | Connection3[Any]

controldata() → Dict[str, str]: return the contents of pg_controldata, or non-True value if pg_controldata call failed

property create_replica_methods: List[str]

property data_dir: str

data_directory_empty() → bool

property database: str

property enforce_hot_standby_feedback: bool

ensure_major_version_is_known() → bool

Calls configure_server_parameters() if _major_version is not known

Returns:: True if _major_version is set, otherwise False

follow(member: Leader | Member | None, role: PostgresqlRole = replica, timeout: float | None = None, do_reload: bool = False) → bool | None

Reconfigure postgres to follow a new member or use different recovery parameters.

Method may call on_role_change callback if role is changing.

Parameters:

member – The member to follow
role – The desired role, one of PostgresqlRole values, normally REPLICA, but could also be a STANDBY_LEADER
timeout – start timeout, how long should the start() method wait for postgres accepting connections
do_reload – indicates that after updating postgresql.conf we just need to do a reload instead of restart

Returns:

True - if restart/reload were successfully performed, False - if restart/reload failed None - if nothing was done or if Postgres is still in starting state after timeout seconds.

get_guc_value(name: str) → str | None

get_history(timeline: int) → List[Tuple[int, int, str] | Tuple[int, int, str, str, str]]

get_major_version() → int

Reads major version from PG_VERSION file

Returns:: major PostgreSQL version in integer format or 0 in case of missing file or errors

get_postgres_role_from_data_directory() → PostgresqlRole

get_primary_timeline() → int

Returns:: current timeline if postgres is running as a primary or 0.

get_replica_timeline() → int | None

get_replication_connection_cursor(host: str | None = None, port: int | str = 5432, **kwargs: Any) → Iterator[cursor | Cursor[Any]]

handle_parameter_change() → None

initdb(*args: str, **kwargs: Any) → bool

Builds and executes the initdb command.

Parameters:

args – List of arguments to be joined into the initdb command.
kwargs – Keyword arguments to pass to subprocess.call.

Returns:

True if the result of subprocess.call`, the exit code, is ``0.

is_healthy() → bool

is_primary() → bool

is_running() → PostmasterProcess | None: Returns PostmasterProcess if one is running on the data directory or None. If most recently seen process is running updates the cached process based on pid file.

is_starting() → bool

last_operation() → int

latest_checkpoint_locations(data: Dict[str, str] | None = None) → Tuple[int | None, int | None]

Get shutdown checkpoint location.

Parameters:: data – dict object with values returned by pg_controldata tool.
Returns:: a tuple of checkpoint LSN for the cleanly shut down primary, and LSN of prev wal record (SWITCH) if we know that the checkpoint was written to the new WAL file due to the archive_mode=on.

property lsn_name: str

property major_version: int

move_data_directory() → None

parse_wal_record(timeline: str, lsn: str) → Tuple[str, str, str, str] | Tuple[None, None, None, None]

property pending_restart_reason: CaseInsensitiveDict

Get _pending_restart_reason value.

_pending_restart_reason is a CaseInsensitiveDict object of the PG parameters that are causing pending restart state. Every key is a parameter name, value - a dictionary containing the old and the new value (see get_param_diff()).

pg_control_exists() → bool

pg_control_timeline() → int | None

pg_ctl(cmd: str, *args: str, **kwargs: Any) → bool

Builds and executes pg_ctl command

Returns:: !True when return_code == 0, otherwise !False

pg_isready() → PgIsReadyStatus

Runs pg_isready to see if PostgreSQL is accepting connections.

Returns:: one of PgIsReadyStatus values.

pg_stat_replication() → List[Dict[str, Any]]

Returns:: a result set of ‘SELECT * FROM pg_stat_replication’.

pg_tblspc_realpaths() → Dict[str, str]: Returns a dict containing the symlink (key) and target (values) for the tablespaces

pg_wal_realpath() → Dict[str, str]: Returns a dict containing the symlink (key) and target (value) for the wal directory

pgcommand(cmd: str) → str

Return path to the specified PostgreSQL command.

Note

If postgresql.bin_name.*cmd* was configured by the user then that binary name is used, otherwise the default binary name cmd is used.

Parameters:: cmd – the Postgres binary name to get path to.
Returns:: path to Postgres binary named cmd.

postmaster_start_time() → str | None

primary_conninfo() → str | None

primary_slot_name() → str | None

promote(wait_seconds: int, task: CriticalTask, before_promote: Callable[[...], Any] | None = None) → bool | None

query(sql: str, *params: Any, retry: bool = True) → List[Tuple[Any, ...]]

Execute sql query with params and optionally return results.

Parameters:

sql – SQL statement to execute.
params – parameters to pass.
retry – whether the query should be retried upon failure or given up immediately.

Returns:

a query response as a list of tuples if there is any.

Raises:

Error if had issues while executing sql.

PostgresConnectionException: if had issues while connecting to the database.

RetryFailedError: if it was detected that connection/query failed due to PostgreSQL restart or if retry deadline was exceeded.

receive_lsn() → int | None

received_timeline() → int | None

reload(block_callbacks: bool = False) → bool

reload_config(config: Dict[str, Any], sighup: bool = False) → None

remove_data_directory() → None

replay_lsn() → int | None

replay_paused() → bool

replica_cached_timeline(primary_timeline: int | None) → int | None

replica_method_can_work_without_replication_connection(method: str) → bool

replica_method_options(method: str) → Dict[str, Any]

replication_state() → str | None

Checks replication state from pg_stat_get_wal_receiver().

Note

Available only since 9.6

Returns:: streaming, in archive recovery, or None

replication_state_from_parameters(is_primary: bool, receiver_state: str | None, restore_command: str | None) → str | None

Figure out the replication state from input parameters.

Note

This method could be only called when Postgres is up, running and queries are successfully executed.

Is_primary:

True is postgres is not running in recovery

Receiver_state:

value from pg_stat_get_wal_receiver.state or None if Postgres is older than 9.6

Restore_command:

value of restore_command GUC for PostgreSQL 12+ or postgresql.recovery_conf.restore_command if it is set in Patroni configuration

Returns:

None for the primary and for Postgres older than 9.6;
’streaming’ if replica is streaming according to the pg_stat_wal_receiver view;
’in archive recovery’ if replica isn’t streaming and there is a restore_command

reset_cluster_info_state(cluster: Cluster | None, tags: Tags | None = None) → None

Reset monitoring query cache.

Note

It happens in the beginning of heart-beat loop and on change of synchronous_standby_names.

Parameters:

cluster – currently known cluster state from DCS
tags – reference to an object implementing Tags interface.

Restarts PostgreSQL.

When timeout parameter is set the call will block either until PostgreSQL has started, failed to start or timeout arrives.

Returns:: True when restart was successful and timeout did not expire when waiting.

resume_wal_replay() → None

property role: PostgresqlRole

schedule_sanity_checks_after_pause() → None: After coming out of pause we have to: 1. configure server parameters if necessary 2. sync replication slots, because it might happen that slots were removed 3. get new ‘Database system identifier’ to make sure that it wasn’t changed

property server_version: int

set_enforce_hot_standby_feedback(value: bool) → None

set_pending_restart_reason(diff_dict: CaseInsensitiveDict) → None

Set new or update current _pending_restart_reason.

Parameters:: diff_dict – :class:CaseInsensitiveDict object with the parameters that are causing pending restart state with the diff of their values. Used to reset/update the _pending_restart_reason.

set_role(value: PostgresqlRole) → None

set_state(value: PostgresqlState) → None

slots() → Dict[str, int]

Get replication slots state.

..note::: Since this methods is supposed to be used only by the leader and only to publish state of replication slots to DCS so that other nodes can advance LSN on respective replication slots, we are also adding our own name to the list. All slots that shouldn’t be published to DCS later will be filtered out by maybe_filter_permanent_slots() method.

Returns:: A dict object with replication slot names and LSNs as absolute values.

Start PostgreSQL

Waits for postmaster to open ports or terminate so pg_isready can be used to check startup completion or failure.

Returns:: True if start was initiated and postmaster ports are open, False if start failed, and None if postgres is still starting up

property state: PostgresqlState

stop(mode: str = 'fast', block_callbacks: bool = False, checkpoint: bool | None = None, on_safepoint: Callable[[...], Any] | None = None, on_shutdown: Callable[[int, int], Any] | None = None, before_shutdown: Callable[[...], Any] | None = None, stop_timeout: int | None = None) → bool

Stop PostgreSQL

Supports a callback when a safepoint is reached. A safepoint is when no user backend can return a successful commit to users. Currently this means we wait for user backends to close. But in the future alternate mechanisms could be added.

Parameters:

on_safepoint – This callback is called when no user backends are running.
on_shutdown – is called when pg_controldata starts reporting Database cluster state: shut down
before_shutdown – is called after running optional CHECKPOINT and before running pg_ctl stop

property supports_multiple_sync: bool

True if Postgres version supports more than one synchronous node.

Type:: returns

property supports_quorum_commit: bool: True if quorum commit is supported by Postgres.

synchronous_commit() → str

Returns:: “synchronous_commit” GUC value.

synchronous_standby_names() → str

Returns:: “synchronous_standby_names” GUC value.

property sysid: str

terminate_postmaster(postmaster: PostmasterProcess, mode: str, stop_timeout: int | None) → bool | None

terminate_starting_postmaster(postmaster: PostmasterProcess) → None: Terminates a postmaster that has not yet opened ports or possibly even written a pid file. Blocks until the process goes away.

time_in_state() → float

timeline_wal_position() → Tuple[int, int, int | None, int | None, int | None]

Get timeline and various wal positions.

Returns:: a tuple composed of 5 integers representing timeline, pg_current_wal_lsn() position on primary/the biggest value among receive and replay LSN on replicas, pg_control_checkpoint() value on a standby leader, receive and replay LSN on replicas (if available).

wait_for_port_open(postmaster: PostmasterProcess, timeout: float) → bool: Waits until PostgreSQL opens ports.

wait_for_startup(timeout: float = 0) → bool | None

Waits for PostgreSQL startup to complete or fail.

Returns:: True if start was successful, False otherwise

property wal_dir: str

property wal_flush: str: For PostgreSQL 9.6 onwards we want to use pg_current_wal_flush_lsn()/pg_current_xlog_flush_location().

property wal_name: str

waldump(timeline: int | str, lsn: str, limit: int) → Tuple[bytes | None, bytes | None]

patroni.postgresql.null_context()