Skip to main content
Version: 2.x

Extension Modules Overview

Apache Cloudberry ships with a set of optional modules that add data types, functions, operators, foreign-data wrappers, storage formats, and diagnostic tools to the database. The modules live in two directories of the source tree:

  • contrib/ holds the modules that come from PostgreSQL, plus several modules that Apache Cloudberry adds.
  • gpcontrib/ holds the modules that are specific to Apache Cloudberry.

This page lists every module in both directories as of Apache Cloudberry 2.2, with its version, its history across the 2.x releases, and how it is built.

Use a module​

Most modules are packaged as extensions. An extension is registered once in every database in which you want to use it:

CREATE EXTENSION <extension_name>;

To list the extensions that are installed on the cluster and the ones that are registered in the current database:

SELECT name, default_version, installed_version FROM pg_available_extensions ORDER BY name;

Some modules are plain shared libraries that do not have SQL objects. You load such a module with LOAD, or by adding it to shared_preload_libraries or session_preload_libraries. The page of each module tells you which method applies.

How to read the tables​

Version is the extension version (default_version in the control file of the module) in Apache Cloudberry 2.2. A dash means that the module is a library or a program without an extension version.

2.x history compares the 2.0, 2.1, and 2.2 releases with each other, and 2.0 with the 1.6 release:

  • New in 2.0, New in 2.1, New in 2.2: the release that first includes the module.
  • Updated in ...: the extension version changed in that release.
  • A dash means that the module is present in 2.0, 2.1, and 2.2 with the same version.

Build tells you how the module gets built when you build Apache Cloudberry from source:

  • Default: make and make install at the top of the source tree build and install the module.
  • On request: the module is in the source tree, but you need to build it explicitly, for example with make -C contrib/<module> install, or by building the world target. The build scripts in devops/build/automation/cloudberry/scripts of the source tree also build and install contrib/, so a cluster built with them has these modules, provided that the required configure options are set.
  • A configure option: the module is built only when you pass that option to configure. Where the option is on by default, the table says so.
  • Not built: the source is in the tree, but no makefile target of Apache Cloudberry builds it.

The module files must be installed on the coordinator and on every segment host.

note

A module name that links to a page in this section has been reviewed against the Apache Cloudberry 2.x source code. For the other PostgreSQL modules, the link points to the PostgreSQL documentation. Those modules follow the PostgreSQL behavior, and some of them might have limitations in a distributed cluster.

Modules in contrib/​

PostgreSQL modules​

ModuleDescriptionVersion2.x historyBuild
adminpackSupport functions for administration tools such as pgAdmin.2.1—Not built
amcheckVerifies the logical consistency of relations and indexes.1.3—On request
auth_delayPauses before reporting an authentication failure.——On request
auto_explainLogs execution plans of slow statements automatically.——Default
bloomIndex access method based on Bloom filters.1.0—On request
bool_plperlTransform between bool and PL/Perl.1.0—On request, with --with-perl
btree_ginGIN operator classes with B-tree behavior.1.3—Default
btree_gistGiST operator classes with B-tree behavior.1.6—On request
citextCase-insensitive character string type.1.6—Default
cubeData type for multidimensional cubes.1.5—Not built
dblinkConnects to other databases from within a session.1.2—Default
dict_intText search dictionary template for integers.1.0—On request
dict_xsynText search dictionary template for extended synonym processing.1.0—On request
earthdistanceCalculates great-circle distances on the surface of the Earth.1.1—Not built
file_fdwForeign-data wrapper for files in the server file system.1.0—Default
fuzzystrmatchDetermines similarities and distance between strings.1.1—Default
hstoreStores sets of key/value pairs in a single value.1.8—Default
hstore_plperlTransform between hstore and PL/Perl.1.0—On request, with --with-perl
hstore_plpythonTransform between hstore and PL/Python.1.0—On request, with --with-python (on by default)
intaggInteger aggregator and enumerator (obsolete).1.1—On request
intarrayFunctions, operators, and index support for arrays of integers.1.5—On request
isnData types for international product numbering standards.1.2—On request
jsonb_plperlTransform between jsonb and PL/Perl.1.0—On request, with --with-perl
jsonb_plpythonTransform between jsonb and PL/Python.1.0—On request, with --with-python (on by default)
loLarge object maintenance.1.1—Not built
ltreeData type for hierarchical tree-like labels.1.2—On request
ltree_plpythonTransform between ltree and PL/Python.1.0—On request, with --with-python (on by default)
oid2nameUtility program that resolves OIDs and file nodes to names.——On request
old_snapshotShows the XID/time mapping that old_snapshot_threshold uses.1.0—On request
pageinspectInspects the contents of database pages at a low level.1.9—Default
passwordcheckChecks the strength of passwords.——Default
pg_buffercacheExamines the shared buffer cache.1.4.1Updated in 2.0 (1.3 to 1.4.1)Default
pg_freespacemapExamines the free space map.1.2—On request
pg_prewarmLoads relation data into the buffer cache.1.2—On request
pg_standbyUtility program for warm standby servers.——Not built
pg_stat_statementsTracks planning and execution statistics of SQL statements.1.9—On request
pg_surgeryPerforms surgery on a damaged relation.1.0—On request
pg_trgmText similarity measurement and index searching based on trigrams.1.6—Default
pg_visibilityExamines the visibility map and page-level visibility information.1.2—On request
pgcryptoCryptographic functions.1.3—Default
pgrowlocksShows row-level locking information.1.2—On request
pgstattupleShows tuple-level statistics.1.5—On request
postgres_fdwForeign-data wrapper for data stored in external PostgreSQL servers.1.1—On request
segData type for line segments or floating-point intervals.1.4—On request
sepgsqlLabel-based mandatory access control based on SELinux.——On request, with --with-selinux
spiExample trigger functions: autoinc, insert_username, moddatetime, and refint.1.0—On request
sslinfoInformation about the SSL certificate of the current client.1.2—On request, requires OpenSSL
start-scriptsExample scripts for starting the server at boot time.——Not built
tablefuncFunctions that return tables, including crosstab.1.0—Default
tcnTrigger function that notifies listeners of changes to a table.1.0—On request
test_decodingExample output plugin for logical decoding.——On request
tsm_system_rowsSYSTEM_ROWS table sampling method.1.0—On request
tsm_system_timeSYSTEM_TIME table sampling method.1.0—On request
unaccentText search dictionary that removes accents.1.1—On request
uuid-osspFunctions that generate universally unique identifiers (UUIDs).1.1—On request, with --with-uuid=LIB
vacuumloUtility program that removes orphaned large objects.——On request
xml2XPath querying and XSLT functions (obsolete).1.1—On request, with --with-libxml

Apache Cloudberry modules​

ModuleDescriptionVersion2.x historyBuild
extprotocolExample of a custom protocol for external tables.——Default
formatterExample of a custom data formatter for external tables.——Default
formatter_fixedwidthCustom formatter that reads and writes fixed-width data files.——Default
indexscanThe readindex() function, which reads the entries of an index for testing and diagnosis.——Default
interconnectThe interconnect implementations (UDP, TCP, and proxy) as a loadable module. The interconnect extension adds views with interconnect statistics.1.0Updated in 2.2 (packaged as an extension)Default
pax_storageThe PAX table format. See PAX Storage Format.—New in 2.0--enable-pax
sasdemoExample programs and formatter functions for exchanging data with SAS.——Not built
try_convertThe try_convert() function for type casts that return a default value instead of an error.1.0New in 2.2Default
udp2A reworked implementation of the UDP interconnect that is separated from the database kernel.—New in 2.1--enable-ic-udp2

Modules in gpcontrib/​

ModuleDescriptionVersion2.x historyBuild
diskquotaLimits the disk space that schemas, roles, and tablespaces can use.2.3New in 2.2--with-diskquota
gp_debug_numsegmentsGets and sets the default number of segments that new tables are created on. For development and testing.1.0—Default (debug extensions)
gp_distribution_policyChecks whether the rows of a table are stored on the segments that the distribution policy expects.1.0—Default
gp_exttable_fdwForeign-data wrapper that implements external tables.1.0—Default
gp_inject_faultFault injection framework. For development and testing.1.0—Default (debug extensions)
gp_internal_toolsDiagnostic views and functions for troubleshooting.1.0.0—Default
gp_legacy_string_aggThe single-argument string_agg() aggregate.1.0.0—Default
gp_relsizes_statsTracks the on-disk sizes of tables.1.0New in 2.2Default
gp_replica_checkCompares primary and mirror segments. For development and testing.0.0.1—Default (debug extensions)
gp_sparse_vectorSparse vector data type and its operators.1.0.1—Default
gp_stats_collectorCollects query and plan execution metrics and reports them to a monitoring agent.1.2New in 2.2--with-gp-stats-collector
gp_toolkitAdministrative views and functions.1.7Updated in 2.0 (1.4 to 1.6), Updated in 2.2 (1.6 to 1.7)Default
gpcloudThe s3 protocol for external tables and the gpcheckcloud utility. See Load Data from Amazon S3.——--enable-gpcloud
gpmapreduceThe gpmapreduce utility.——--enable-mapreduce
orafceOracle compatibility functions and packages.4.9—--enable-orafce
pg_hint_planControls execution plans with hints in SQL comments. See Query Plan Hints.1.3.9New in 2.0Default
pxf_fdwForeign-data wrapper for the Platform Extension Framework (PXF). See Load Data Using PXF.1.0—--enable-pxf (on by default)
reject_partition_fullscanRejects queries that scan all partitions of a table without pruning.1.0New in 2.2Default (debug extensions)
yezzeyOffloads the data of append-optimized tables to S3-compatible object storage. The source is a Git submodule.2.0New in 2.2--with-yezzey
zstdZstandard compression for append-optimized tables.——--with-zstd (on by default)

The modules that are marked as debug extensions are built unless you pass --disable-debug-extensions to configure. Do not use gp_debug_numsegments, gp_inject_fault, or gp_replica_check in production clusters.

Removed modules​

ModuleDescriptionRemoved in
quicklzQuickLZ compression for append-optimized tables.2.1