Add experimental chunk replication view

A new view in the experimental schema shows information related to
chunk replication. The view can be used to learn the replication
status of a chunk while also providing a way to easily find nodes to
move or copy chunks between in order to ensure a fully replicated
multi-node cluster.

Tests have been added to illustrate the potential usage.

cmake/ScriptFiles.cmake

@@ -54,6 +54,7 @@ set(SOURCE_FILES
     metadata.sql
     dist_internal.sql
     views.sql
+    views_experimental.sql
     gapfill.sql
     maintenance_utils.sql
     partialize_finalize.sql

sql/updates/reverse-dev.sql

@@ -1,4 +1,3 @@
-DROP SCHEMA IF EXISTS timescaledb_experimental CASCADE;
 DROP FUNCTION IF EXISTS _timescaledb_internal.block_new_chunks;
 DROP FUNCTION IF EXISTS _timescaledb_internal.allow_new_chunks;
 DROP FUNCTION IF EXISTS _timescaledb_internal.refresh_continuous_aggregate;
@@ -12,6 +11,8 @@ DROP PROCEDURE IF EXISTS timescaledb_experimental.move_chunk;
 DROP PROCEDURE IF EXISTS timescaledb_experimental.copy_chunk;
 DROP TABLE IF EXISTS _timescaledb_catalog.chunk_copy_activity;
 DROP SEQUENCE IF EXISTS _timescaledb_catalog.chunk_copy_activity_id_seq;
+DROP VIEW IF EXISTS timescaledb_experimental.chunk_replication_status;
+DROP SCHEMA IF EXISTS timescaledb_experimental CASCADE;
 
 -- We need to rewrite all continuous aggregates to make sure that the
 -- queries do not contain qualification. They will be re-written in

sql/views_experimental.sql

@@ -0,0 +1,28 @@
+-- This file and its contents are licensed under the Apache License 2.0.
+-- Please see the included NOTICE for copyright information and
+-- LICENSE-APACHE for a copy of the license.
+
+CREATE VIEW timescaledb_experimental.chunk_replication_status AS
+SELECT
+    h.schema_name AS hypertable_schema,
+    h.table_name AS hypertable_name,
+    c.schema_name AS chunk_schema,
+    c.table_name AS chunk_name,
+    h.replication_factor AS desired_num_replicas,
+    count(cdn.chunk_id) AS num_replicas,
+    array_agg(cdn.node_name) AS replica_nodes,
+    -- compute the set of data nodes that doesn't have the chunk
+    (SELECT array_agg(node_name) FROM
+            (SELECT node_name FROM _timescaledb_catalog.hypertable_data_node hdn
+             WHERE hdn.hypertable_id = h.id
+             EXCEPT
+             SELECT node_name FROM _timescaledb_catalog.chunk_data_node cdn
+             WHERE cdn.chunk_id = c.id
+             ORDER BY node_name) nodes) AS non_replica_nodes
+FROM _timescaledb_catalog.chunk c
+INNER JOIN _timescaledb_catalog.chunk_data_node cdn ON (cdn.chunk_id = c.id)
+INNER JOIN _timescaledb_catalog.hypertable h ON (h.id = c.hypertable_id)
+GROUP BY h.id, c.id, hypertable_schema, hypertable_name, chunk_schema, chunk_name
+ORDER BY h.id, c.id, hypertable_schema, hypertable_name, chunk_schema, chunk_name;
+
+GRANT SELECT ON ALL TABLES IN SCHEMA timescaledb_experimental TO PUBLIC;

test/expected/pg_dump.out

@@ -545,6 +545,7 @@ WHERE   refclassid = 'pg_catalog.pg_extension'::pg_catalog.regclass AND
         ORDER BY objid::text DESC;
                        objid                       
 ---------------------------------------------------
+ timescaledb_experimental.chunk_replication_status
  timescaledb_information.compression_settings
  timescaledb_information.dimensions
  timescaledb_information.chunks
@@ -561,7 +562,7 @@ WHERE   refclassid = 'pg_catalog.pg_extension'::pg_catalog.regclass AND
  _timescaledb_internal.bgw_policy_chunk_stats
  _timescaledb_internal.bgw_job_stat
  _timescaledb_catalog.tablespace_id_seq
-(16 rows)
+(17 rows)
 
 -- Make sure we can't run our restoring functions as a normal perm user as that would disable functionality for the whole db
 \c :TEST_DBNAME :ROLE_DEFAULT_PERM_USER

tsl/test/expected/dist_views.out

@@ -164,3 +164,49 @@ SELECT * FROM  hypertable_index_size( 'dist_table_time_idx') ;
                 114688
 (1 row)
 
+-- Test chunk_replication_status view
+SELECT * FROM timescaledb_experimental.chunk_replication_status
+ORDER BY chunk_schema, chunk_name
+LIMIT 4;
+ hypertable_schema | hypertable_name |     chunk_schema      |      chunk_name       | desired_num_replicas | num_replicas |       replica_nodes       | non_replica_nodes 
+-------------------+-----------------+-----------------------+-----------------------+----------------------+--------------+---------------------------+-------------------
+ public            | dist_table      | _timescaledb_internal | _dist_hyper_1_1_chunk |                    2 |            2 | {view_node_1,view_node_2} | {view_node_3}
+ public            | dist_table      | _timescaledb_internal | _dist_hyper_1_2_chunk |                    2 |            2 | {view_node_2,view_node_3} | {view_node_1}
+ public            | dist_table      | _timescaledb_internal | _dist_hyper_1_3_chunk |                    2 |            2 | {view_node_3,view_node_1} | {view_node_2}
+ public            | quote'tab       | _timescaledb_internal | _dist_hyper_2_4_chunk |                    2 |            2 | {view_node_1,view_node_2} | {view_node_3}
+(4 rows)
+
+-- drop one chunk replica
+SELECT _timescaledb_internal.chunk_drop_replica(format('%I.%I', chunk_schema, chunk_name)::regclass, replica_nodes[1])
+FROM timescaledb_experimental.chunk_replication_status
+ORDER BY chunk_schema, chunk_name
+LIMIT 1;
+ chunk_drop_replica 
+--------------------
+ 
+(1 row)
+
+SELECT * FROM timescaledb_experimental.chunk_replication_status
+WHERE num_replicas < desired_num_replicas
+ORDER BY chunk_schema, chunk_name;
+ hypertable_schema | hypertable_name |     chunk_schema      |      chunk_name       | desired_num_replicas | num_replicas | replica_nodes |     non_replica_nodes     
+-------------------+-----------------+-----------------------+-----------------------+----------------------+--------------+---------------+---------------------------
+ public            | dist_table      | _timescaledb_internal | _dist_hyper_1_1_chunk |                    2 |            1 | {view_node_2} | {view_node_1,view_node_3}
+(1 row)
+
+-- Example usage of finding data nodes to copy/move chunks between
+SELECT
+    format('%I.%I', chunk_schema, chunk_name)::regclass AS chunk,
+    replica_nodes[1] AS copy_from_node,
+    non_replica_nodes[1] AS copy_to_node
+FROM
+    timescaledb_experimental.chunk_replication_status
+WHERE
+    num_replicas < desired_num_replicas
+ORDER BY
+    chunk_schema, chunk_name;
+                    chunk                    | copy_from_node | copy_to_node 
+---------------------------------------------+----------------+--------------
+ _timescaledb_internal._dist_hyper_1_1_chunk | view_node_2    | view_node_1
+(1 row)
+

tsl/test/sql/dist_views.sql

@@ -60,3 +60,30 @@ SELECT create_hypertable( 'special#tab', 'a', 'b', replication_factor=>2, chunk_
 INSERT into "special#tab" select generate_series( '2020-02-02 10:00', '2020-02-05 10:00' , '1 day'::interval), 10;
 SELECT * FROM  chunks_detailed_size( '"special#tab"') ORDER BY chunk_name, node_name;
 SELECT * FROM  hypertable_index_size( 'dist_table_time_idx') ;
+
+-- Test chunk_replication_status view
+SELECT * FROM timescaledb_experimental.chunk_replication_status
+ORDER BY chunk_schema, chunk_name
+LIMIT 4;
+
+-- drop one chunk replica
+SELECT _timescaledb_internal.chunk_drop_replica(format('%I.%I', chunk_schema, chunk_name)::regclass, replica_nodes[1])
+FROM timescaledb_experimental.chunk_replication_status
+ORDER BY chunk_schema, chunk_name
+LIMIT 1;
+
+SELECT * FROM timescaledb_experimental.chunk_replication_status
+WHERE num_replicas < desired_num_replicas
+ORDER BY chunk_schema, chunk_name;
+
+-- Example usage of finding data nodes to copy/move chunks between
+SELECT
+    format('%I.%I', chunk_schema, chunk_name)::regclass AS chunk,
+    replica_nodes[1] AS copy_from_node,
+    non_replica_nodes[1] AS copy_to_node
+FROM
+    timescaledb_experimental.chunk_replication_status
+WHERE
+    num_replicas < desired_num_replicas
+ORDER BY
+    chunk_schema, chunk_name;