enable_vector_index

enable_experimental_vector

sql
ADMIN SET FRONTEND CONFIG ("enable_experimental_vector" = "true");

enable_experimental_vector = true

fe.conf

INDEX ... USING VECTOR

CREATE TABLE

ALTER TABLE

efsearch

ann_params

sql
    SELECT /_+ SET_VAR(ann_params='{efsearch=256}') _/ ...

CREATE TABLE

USING VECTOR

"index_type" = "hnsw"

dim

cosine_similarity

cosine

ef_construction

efconstruction

sql
CREATE TABLE db_rag.rag_table_store (
  table_id VARCHAR(64) NOT NULL,
  summary_text TEXT,
  summary_vector ARRAY<FLOAT> NOT NULL, 
  raw_table_markdown TEXT,

  -- Corrected Vector Index Definition
  INDEX idx_vector (summary_vector) USING VECTOR (
    "index_type" = "hnsw",        -- Required: Specify HNSW here
    "dim" = "1536",               -- Required: Match your vector length
    "metric_type" = "cosine_similarity", -- Use 'cosine_similarity' or 'l2_distance'
    "M" = "16",
    "efconstruction" = "40"       -- Parameter name has no underscore
  )
)
ENGINE=OLAP
DUPLICATE KEY(table_id)
DISTRIBUTED BY HASH(table_id);

ADMIN SET FRONTEND CONFIG ("enable_experimental_vector" = "true");

ARRAY<FLOAT>

approx_cosine_similarity

approx_l2_distance

metric_type = "cosine_similarity"

approx_cosine_similarity()

approx_dot_product

knowlegedb.idx_vector

db_rag.rag_table_store

AS similarity

SELECT

WHERE

DESC

LIMIT

WHERE

ORDER BY ... LIMIT

sql
SELECT 
    *, 
    approx_cosine_similarity(summary_vector, [0.015, -0.023, 0.412, 0.009]) AS similarity
FROM db_rag.rag_table_store
ORDER BY similarity DESC
LIMIT 3;

approx_cosine_similarity

approx_l2_distance

DESC

ASC

ORDER BY

LIMIT

0.8

sql
SELECT * FROM (
    SELECT 
        *, 
        approx_cosine_similarity(summary_vector, [0.015, -0.023, 0.412, 0.009]) AS similarity
    FROM db_rag.rag_table_store
) t
WHERE similarity > 0.8
ORDER BY similarity DESC
LIMIT 3;

approx_cosine_similarity

cosine_similarity

sql
approx_cosine_similarity(vector_column, constant_array)

vector_column

constant_array

[0.1, 0.5, -0.2]

ARRAY<FLOAT>

dim

"metric_type" = "cosine_similarity"

ORDER BY

DESC

LIMIT

sql
SELECT 
    id, 
    approx_cosine_similarity(summary_vector, [0.12, 0.45, 0.88]) AS score
FROM db_rag.rag_table_store
ORDER BY score DESC
LIMIT 5;

approx_cosine_similarity

cosine_similarity

LIMIT

"is_vector_normed" = "true"

US

CN

M

VARCHAR(65533)

VARCHAR(1048576)

VARCHAR(65533)

STRING

VARCHAR(M)

VARCHAR(N)

CHAR

VARCHAR

STRING

dt

(symbol, recorded_at)

dt

dt

sql
CREATE TABLE starrocks_audit_db__.fundingrate (
    symbol                  VARCHAR(32)  NOT NULL,
    recorded_at             BIGINT       NOT NULL COMMENT 'unix ms, thời điểm script chạy',
    dt                      DATETIME     NOT NULL DEFAULT (NOW()) COMMENT 'partition key',
    last_funding_rate       DOUBLE       NOT NULL,
    mark_price              DOUBLE,
    index_price             DOUBLE,
    estimated_settle_price  DOUBLE,
    interest_rate           DOUBLE,
    next_funding_time       BIGINT,
    time                    BIGINT,
    id                      BIGINT       NOT NULL
) ENGINE=OLAP
PRIMARY KEY(symbol, recorded_at, dt) -- Added dt here
PARTITION BY date_trunc('month', dt)
DISTRIBUTED BY HASH(symbol) BUCKETS AUTO -- Recommended: use AUTO buckets
PROPERTIES (
    "compression" = "LZ4",
    "enable_persistent_index" = "true",
    "fast_schema_evolution" = "true",
    "replication_num" = "3" -- Recommended: 3 for production HA
);

dt

PRIMARY KEY

BUCKETS 10

BUCKETS AUTO

AUTO

"replication_num" = "2"

3

1

2

starrocks_audit_db__

funding_db

dt

dt

recorded_at

dt

recorded_at

DEFAULT (NOW())

dt

Unexpected input 'AUTO'

AUTO

BUCKETS AUTO

10

dt

sql
CREATE TABLE starrocks_audit_db__.fundingrate (
    symbol                  VARCHAR(32)  NOT NULL,
    recorded_at             BIGINT       NOT NULL COMMENT 'unix ms, thời điểm script chạy',
    dt                      DATETIME     NOT NULL DEFAULT (NOW()) COMMENT 'partition key',
    last_funding_rate       DOUBLE       NOT NULL,
    mark_price              DOUBLE,
    index_price             DOUBLE,
    estimated_settle_price  DOUBLE,
    interest_rate           DOUBLE,
    next_funding_time       BIGINT,
    time                    BIGINT,
    id                      BIGINT       NOT NULL
) ENGINE=OLAP
PRIMARY KEY(symbol, recorded_at, dt)
PARTITION BY date_trunc('month', dt)
DISTRIBUTED BY HASH(symbol) BUCKETS 10 -- Use an integer here
PROPERTIES (
    "compression" = "LZ4",
    "enable_persistent_index" = "true",
    "fast_schema_evolution" = "true",
    "replication_num" = "3" 
);

BUCKETS 10

replication_num

"replication_num" = "3"

"2"

"1"

starrocks_audit_db__

starrocks_monitor

trading_audit

funding_data

LIKE

=

information_schema

=

ILIKE

sql
    WHERE TASK_NAME ILIKE '%KBV%'
    -- OR
    WHERE lower(TASK_NAME) LIKE '%kbv%'

information_schema.task_runs

LIKE

=

LIKE

information_schema

LIKE

_

KBV_CE

LIKE '%KBV%'

KBV_CE

LIKE '%KBV_%'

KBVACE

KBVBCE

LIKE '%KBV\_%'

='KBV_CE'

information_schema

utf8_bin

utf8_general_ci

LIKE

REGEXP

LIKE

sql
    WHERE TASK_NAME REGEXP 'KBV'

SELECT DISTINCT TASK_NAME FROM information_schema.task_runs

ILIKE

REGEXP

task_runs

task_runs table have this issue

information_schema.task_runs

LIKE

=

task_runs

tasks

TASK_NAME = 'KBV_CE'

LIKE

LIKE

REGEXP

TaskRun

tasks

information_schema.tasks

task_runs

=

LIKE

WHERE TASK_NAME ILIKE '%KBV%'

task_runs

SchemaScan

task_runs

task_runs

=

be_http_port

http://<be_ip>:8040

http://<be_ip>:8040/metrics

be_http_port

be_port

brpc_port

heartbeat_service_port

JSON

JSON

JSON

2026-08-03T14:40:12.253819

DATETIME

T

str_to_date()

"product_type_id": []

ARRAY<INT>

JSON

"ext_number": null

NULL

bash
-H "format: json" -H "strip_outer_array: true"

strip_outer_array: true

[]

logcall_crm_tbl

id

queue_id

phone_number

ext_answer

call_status

id_snowflake

id_data

campaign_id

start_time

ringing_time

waiting_time

talking_time

created_at

data

update_at

is_send

destination_caller

recording_url

ext_no_answer

data

JSON

STRING

VARCHAR

JSON

STRING

VARCHAR(length)

STRING

VARCHAR(65533)

sql
CREATE TABLE `logcall_crm_tbl` (
  ...
  `data` STRING NULL COMMENT "",  -- Changed from JSON to STRING
  ...
);

JSON

STRING

ALTER TABLE MODIFY COLUMN

sql
    ALTER TABLE logcall_crm_tbl ADD COLUMN data_new STRING;

CAST

sql
    UPDATE logcall_crm_tbl SET data_new = CAST(data AS STRING);

sql
    ALTER TABLE logcall_crm_tbl DROP COLUMN data;

sql
    -- Supported in v3.3.2+
    ALTER TABLE logcall_crm_tbl RENAME COLUMN data_new TO data;

STRING

VARCHAR

JSON

STRING

data->'$.id'

get_json_string(data, '$.id')

ALTER TABLE

sql
    SHOW ALTER TABLE COLUMN;

DATA

CREATE TABLE

sql
CREATE TABLE `logcall_crm_tbl` (
  ...
  `data` STRING NULL COMMENT "",  -- Using backticks is recommended
  ...
);

column_name

data

order

key

time

_

SELECT

TABLE

JOIN

PRIMARY

DATA

DB

STATS

debt_collection_cus_action_hist_raw

knowlegedb

a

ID

knowlegedb

a

SYS_RUN_DATE

INSERT OVERWRITE

SELECT

INSERT OVERWRITE

INSERT OVERWRITE

INSERT OVERWRITE

INSERT OVERWRITE

INSERT OVERWRITE

INSERT OVERWRITE

sql
    CREATE MATERIALIZED VIEW my_mv
    REFRESH ASYNC [EVERY INTERVAL]
    AS SELECT ... FROM source_table;

INSERT OVERWRITE

sql
INSERT OVERWRITE `debt_collection_cus_action_hist_raw` 
PARTITION (p20231027)
SELECT ... FROM ... WHERE date = '2023-10-27';

INSERT OVERWRITE

sql
SELECT * FROM information_schema.task_runs 
WHERE task_name = 'your_task_name' 
ORDER BY create_time DESC;