patroni.postgresql.mpp package¶
Submodules¶
Module contents¶
Abstract classes for MPP handler.
MPP stands for Massively Parallel Processing, and Citus belongs to this architecture. Currently, Citus is the only supported MPP cluster. However, we may consider adapting other databases such as TimescaleDB, GPDB, etc. into Patroni.
-
class
patroni.postgresql.mpp.
AbstractMPP
(config: Dict[str, Union[str, int]])¶ Bases:
abc.ABC
An abstract class which should be passed to
AbstractDCS
.Note
We create
AbstractMPP
andAbstractMPPHandler
to solve the chicken-egg initialization problem. When initializing DCS, we dynamically create an object implementingAbstractMPP
, later this object is used to instantiate an object implementingAbstractMPPHandler
.-
__init__
(config: Dict[str, Union[str, int]]) → None¶ Init method for
AbstractMPP
.Parameters: config – configuration of MPP section.
-
_abc_impl
= <_abc_data object>¶
-
_get_handler_cls
() → Iterator[Type[patroni.postgresql.mpp.AbstractMPPHandler]]¶ Find Handler classes inherited from a class type of this object.
Yields: handler classes for this object.
-
coordinator_group_id
¶ The group id of the coordinator PostgreSQL cluster.
-
get_handler_impl
(postgresql: Postgresql) → AbstractMPPHandler¶ Find and instantiate Handler implementation of this object.
Parameters: postgresql – a reference to Postgresql
object.Raises: PatroniException
: if the Handler class haven’t been found.Returns: an instantiated class that implements Handler for this object.
-
group
¶ The group for a given MPP implementation.
-
is_coordinator
() → bool¶ Check whether this node is running in the coordinator PostgreSQL cluster.
Returns: True
if MPP is enabled and the group id of this node matches with thecoordinator_group_id
, otherwiseFalse
.
-
is_enabled
() → bool¶ Check if MPP is enabled for a given MPP.
Note
We just check that the
_config
object isn’t empty and expect it to be empty only in case ofNull
.Returns: True
if MPP is enabled, otherwiseFalse
.
-
is_worker
() → bool¶ Check whether this node is running as a MPP worker PostgreSQL cluster.
Returns: True
if MPP is enabled and this node is known to be not running as the coordinator PostgreSQL cluster, otherwiseFalse
.
-
k8s_group_label
¶ Group label used for kubernetes DCS of the MPP cluster.
Returns: A string representation of the k8s group label of a given MPP implementation.
-
type
¶ The type of the MPP cluster.
Returns: A string representation of the type of a given MPP implementation.
-
static
validate_config
(config: Any) → bool¶ Check whether provided config is good for a given MPP.
Parameters: config – configuration of MPP section. Returns: True
is config passes validation, otherwiseFalse
.
-
-
class
patroni.postgresql.mpp.
AbstractMPPHandler
(postgresql: Postgresql, config: Dict[str, Union[str, int]])¶ Bases:
patroni.postgresql.mpp.AbstractMPP
An abstract class which defines interfaces that should be implemented by real handlers.
-
__init__
(postgresql: Postgresql, config: Dict[str, Union[str, int]]) → None¶ Init method for
AbstractMPPHandler
.Parameters: - postgresql – a reference to
Postgresql
object. - config – configuration of MPP section.
- postgresql – a reference to
-
_abc_impl
= <_abc_data object>¶
-
adjust_postgres_gucs
(parameters: Dict[str, Any]) → None¶ Adjust GUCs in the current PostgreSQL configuration.
Parameters: parameters – dictionary of GUCs, with key as GUC name and the corresponding value as current GUC value.
-
bootstrap
() → None¶ Bootstrap handler.
Is called when the new cluster is initialized (through
initdb
or a custom bootstrap method).
-
handle_event
(cluster: patroni.dcs.Cluster, event: Dict[str, Any]) → None¶ Handle an event sent from a worker node.
Parameters: - cluster – the currently known cluster state from DCS.
- event – the event to be handled.
-
ignore_replication_slot
(slot: Dict[str, str]) → bool¶ Check whether provided replication slot existing in the database should not be removed.
Note
MPP database may create replication slots for its own use, for example to migrate data between workers using logical replication, and we don’t want to suddenly drop them.
Parameters: slot – dictionary containing the replication slot settings, like name
,database
,type
, andplugin
.Returns: True
if the replication slots should not be removed, otherwiseFalse
.
-
on_demote
() → None¶ On demote handler.
Is called when the primary was demoted.
-
schedule_cache_rebuild
() → None¶ Cache rebuild handler.
Is called to notify handler that it has to refresh its metadata cache from the database.
-
sync_meta_data
(cluster: patroni.dcs.Cluster) → None¶ Sync meta data on the coordinator.
Parameters: cluster – the currently known cluster state from DCS.
-
-
class
patroni.postgresql.mpp.
Null
¶ Bases:
patroni.postgresql.mpp.AbstractMPP
Dummy implementation of
AbstractMPP
.-
_abc_impl
= <_abc_data object>¶
-
coordinator_group_id
¶ The group id of the coordinator PostgreSQL cluster.
Returns: always None
.
-
-
class
patroni.postgresql.mpp.
NullHandler
(postgresql: Postgresql, config: Dict[str, Union[str, int]])¶ Bases:
patroni.postgresql.mpp.Null
,patroni.postgresql.mpp.AbstractMPPHandler
Dummy implementation of
AbstractMPPHandler
.-
__init__
(postgresql: Postgresql, config: Dict[str, Union[str, int]]) → None¶ Init method for
NullHandler
.Parameters: - postgresql – a reference to
Postgresql
object. - config – configuration of MPP section.
- postgresql – a reference to
-
_abc_impl
= <_abc_data object>¶
-
adjust_postgres_gucs
(parameters: Dict[str, Any]) → None¶ Adjust GUCs in the current PostgreSQL configuration.
Parameters: parameters – dictionary of GUCs, with key as GUC name and corresponding value as current GUC value.
-
bootstrap
() → None¶ Bootstrap handler.
Is called when the new cluster is initialized (through
initdb
or a custom bootstrap method).
-
handle_event
(cluster: patroni.dcs.Cluster, event: Dict[str, Any]) → None¶ Handle an event sent from a worker node.
Parameters: - cluster – the currently known cluster state from DCS.
- event – the event to be handled.
-
ignore_replication_slot
(slot: Dict[str, str]) → bool¶ Check whether provided replication slot existing in the database should not be removed.
Note
MPP database may create replication slots for its own use, for example to migrate data between workers using logical replication, and we don’t want to suddenly drop them.
Parameters: slot – dictionary containing the replication slot settings, like name
,database
,type
, andplugin
.Returns: always False
.
-
on_demote
() → None¶ On demote handler.
Is called when the primary was demoted.
-
schedule_cache_rebuild
() → None¶ Cache rebuild handler.
Is called to notify handler that it has to refresh its metadata cache from the database.
-
sync_meta_data
(cluster: patroni.dcs.Cluster) → None¶ Sync meta data on the coordinator.
Parameters: cluster – the currently known cluster state from DCS.
-
-
patroni.postgresql.mpp.
get_mpp
(config: Union[Config, Dict[str, Any]]) → patroni.postgresql.mpp.AbstractMPP¶ Attempt to load and instantiate a MPP module from known available implementations.
Parameters: config – object or dictionary with Patroni configuration. Returns: The successfully loaded MPP or fallback to Null
.
-
patroni.postgresql.mpp.
iter_mpp_classes
(config: Union[Config, Dict[str, Any], None] = None) → Iterator[Tuple[str, Type[patroni.postgresql.mpp.AbstractMPP]]]¶ Attempt to import MPP modules that are present in the given configuration.
Parameters: config – configuration information with possible MPP names as keys. If given, only attempt to import MPP modules defined in the configuration. Else, if None
, attempt to import any supported MPP module.Yields: tuples, each containing the module name
and the imported MPP class object.