mirror of https://github.com/citusdata/citus.git
157 lines
5.7 KiB
Plaintext
157 lines
5.7 KiB
Plaintext
--
|
|
-- MULTI_AGG_APPROXIMATE_DISTINCT
|
|
--
|
|
-- Create HLL extension if present, print false result otherwise
|
|
SELECT CASE WHEN COUNT(*) > 0 THEN
|
|
'CREATE EXTENSION HLL'
|
|
ELSE 'SELECT false AS hll_present' END
|
|
AS create_cmd FROM pg_available_extensions()
|
|
WHERE name = 'hll'
|
|
\gset
|
|
:create_cmd;
|
|
hll_present
|
|
-------------
|
|
f
|
|
(1 row)
|
|
|
|
\c - - - :worker_1_port
|
|
:create_cmd;
|
|
hll_present
|
|
-------------
|
|
f
|
|
(1 row)
|
|
|
|
\c - - - :worker_2_port
|
|
:create_cmd;
|
|
hll_present
|
|
-------------
|
|
f
|
|
(1 row)
|
|
|
|
\c - - - :master_port
|
|
-- Try to execute count(distinct) when approximate distincts aren't enabled
|
|
SELECT count(distinct l_orderkey) FROM lineitem;
|
|
count
|
|
-------
|
|
2985
|
|
(1 row)
|
|
|
|
-- Check approximate count(distinct) at different precisions / error rates
|
|
SET citus.count_distinct_error_rate = 0.1;
|
|
SELECT count(distinct l_orderkey) FROM lineitem;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
SET citus.count_distinct_error_rate = 0.01;
|
|
SELECT count(distinct l_orderkey) FROM lineitem;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
-- Check approximate count(distinct) for different data types
|
|
SELECT count(distinct l_partkey) FROM lineitem;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
SELECT count(distinct l_extendedprice) FROM lineitem;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
SELECT count(distinct l_shipdate) FROM lineitem;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
SELECT count(distinct l_comment) FROM lineitem;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
-- Check that we can execute approximate count(distinct) on complex expressions
|
|
SELECT count(distinct (l_orderkey * 2 + 1)) FROM lineitem;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
SELECT count(distinct extract(month from l_shipdate)) AS my_month FROM lineitem;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
SELECT count(distinct l_partkey) / count(distinct l_orderkey) FROM lineitem;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
-- Check that we can execute approximate count(distinct) on select queries that
|
|
-- contain different filter, join, sort and limit clauses
|
|
SELECT count(distinct l_orderkey) FROM lineitem
|
|
WHERE octet_length(l_comment) + octet_length('randomtext'::text) > 40;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
SELECT count(DISTINCT l_orderkey) FROM lineitem, orders
|
|
WHERE l_orderkey = o_orderkey AND l_quantity < 5;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
SELECT count(DISTINCT l_orderkey) as distinct_order_count, l_quantity FROM lineitem
|
|
WHERE l_quantity < 32.0
|
|
GROUP BY l_quantity
|
|
ORDER BY distinct_order_count ASC, l_quantity ASC
|
|
LIMIT 10;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
-- Check that approximate count(distinct) works at a table in a schema other than public
|
|
-- create necessary objects
|
|
CREATE SCHEMA test_count_distinct_schema;
|
|
CREATE TABLE test_count_distinct_schema.nation_hash(
|
|
n_nationkey integer not null,
|
|
n_name char(25) not null,
|
|
n_regionkey integer not null,
|
|
n_comment varchar(152)
|
|
);
|
|
SELECT master_create_distributed_table('test_count_distinct_schema.nation_hash', 'n_nationkey', 'hash');
|
|
master_create_distributed_table
|
|
---------------------------------
|
|
|
|
(1 row)
|
|
|
|
SELECT master_create_worker_shards('test_count_distinct_schema.nation_hash', 4, 2);
|
|
master_create_worker_shards
|
|
-----------------------------
|
|
|
|
(1 row)
|
|
|
|
\copy test_count_distinct_schema.nation_hash FROM STDIN with delimiter '|';
|
|
SET search_path TO public;
|
|
SET citus.count_distinct_error_rate TO 0.01;
|
|
SELECT COUNT (DISTINCT n_regionkey) FROM test_count_distinct_schema.nation_hash;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
-- test with search_path is set
|
|
SET search_path TO test_count_distinct_schema;
|
|
SELECT COUNT (DISTINCT n_regionkey) FROM nation_hash;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
SET search_path TO public;
|
|
-- If we have an order by on count(distinct) that we intend to push down to
|
|
-- worker nodes, we need to error out. Otherwise, we are fine.
|
|
SET citus.limit_clause_row_fetch_count = 1000;
|
|
SELECT l_returnflag, count(DISTINCT l_shipdate) as count_distinct, count(*) as total
|
|
FROM lineitem
|
|
GROUP BY l_returnflag
|
|
ORDER BY count_distinct
|
|
LIMIT 10;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
SELECT l_returnflag, count(DISTINCT l_shipdate) as count_distinct, count(*) as total
|
|
FROM lineitem
|
|
GROUP BY l_returnflag
|
|
ORDER BY total
|
|
LIMIT 10;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
SELECT
|
|
l_orderkey,
|
|
count(l_partkey) FILTER (WHERE l_shipmode = 'AIR'),
|
|
count(DISTINCT l_partkey) FILTER (WHERE l_shipmode = 'AIR'),
|
|
count(DISTINCT CASE WHEN l_shipmode = 'AIR' THEN l_partkey ELSE NULL END)
|
|
FROM lineitem
|
|
GROUP BY l_orderkey
|
|
ORDER BY 2 DESC, 1 DESC
|
|
LIMIT 10;
|
|
ERROR: cannot compute count (distinct) approximation
|
|
HINT: You need to have the hll extension loaded.
|
|
-- Check that we can revert config and disable count(distinct) approximations
|
|
SET citus.count_distinct_error_rate = 0.0;
|
|
SELECT count(distinct l_orderkey) FROM lineitem;
|
|
count
|
|
-------
|
|
2985
|
|
(1 row)
|
|
|