AnalyzeMgr

OP_ADD_EXTERNAL_BASIC_STATS_META

ANALYZE

_statistics_.external_column_statistics

EXPLAIN

statistic_cache_columns

ANALYZE

sql
    ANALYZE TABLE <table_name> [WITH SYNC MODE];

fe.log

fe.warn.log

AnalyzeMgr

Statistics

sql
    SELECT * FROM _statistics_.external_column_statistics WHERE table_name = 'your_table';

SHOW ANALYZE STATUS

SHOW ANALYZE JOB

ANALYZE

ANALYZE TABLE

sql
ANALYZE TABLE <table_name> PARTITION (p20231001, p20231002);

FULL

SAMPLE

sql
-- 테이블의 일부 데이터만 샘플링하여 분석
ANALYZE SAMPLE TABLE <table_name>;

ANALYZE SAMPLE TABLE <table_name> PROPERTIES("statistic_sample_collect_rows" = "1000000");

sql
-- 병렬도 설정 (기본값 1, 더 높은 값으로 설정 가능)
SET statistic_collect_parallel = 4; 
ANALYZE TABLE <table_name>;

sql
-- 특정 컬럼만 분석
ANALYZE TABLE <table_name>(col1, col2);

-- (v3.5 이상) 쿼리에 사용된 적이 있는 조건 컬럼만 자동으로 선택하여 분석
ANALYZE TABLE <table_name> PREDICATE COLUMNS;

ANALYZE

statistic_collect_parallel

statistic_collect_parallel

SHOW VARIABLES LIKE 'statistic_collect_parallel';

SET statistic_collect_parallel = 4;

SET GLOBAL statistic_collect_parallel = 4;

ANALYZE SAMPLE TABLE

sql
ANALYZE SAMPLE TABLE <table_name>;

statistic_collect_max_row_count_per_query

ADMIN SET FRONTEND CONFIG ("statistic_collect_max_row_count_per_query" = "1000000000");

ANALYZE TABLE <table_name>(col1, col2);

ANALYZE TABLE <table_name> PARTITION (p202401, p202402);

ADMIN SET FRONTEND CONFIG ("connector_table_query_trigger_analyze_max_running_task_num" = "4");

SAMPLE

statistic_collect_max_row_count_per_query

EXPLAIN COSTS

SCAN

SCAN

EXPLAIN COSTS

Estimates: {row: ..., cpu: ..., memory: ..., network: ..., cost: ...}

WHERE

cost

WHERE

SCAN

row

sql
-- 프로파일 수집 활성화 (세션 수준)
SET enable_profile = true;

-- 쿼리 실행
SELECT ... FROM ... ;

-- 실행한 쿼리의 ID 확인
SELECT last_query_id();

-- 프로파일 출력 (위에서 확인한 ID 입력)
SELECT get_query_profile('019b364f-10c4-704c-b79a-af2cc3a77b89')\G

EXCHANGE_SOURCE_OPERATOR

NetworkTime

WaitTime

RpcAvgTime

RpcCount

NetworkBandwidth

ClosureBlockTime

DeserializeChunkTime

DecompressChunkTime

Execution Overview

QueryCumulativeNetworkTime

EXPLAIN COSTS

ANALYZE

WHERE

col = 'value'

sql
-- 특정 테이블의 통계 수집
ANALYZE TABLE <table_name>;

-- 수집 후 다시 확인
EXPLAIN COSTS <query>;

(MBE)

(FULL)

(SAMPLE)

row

information_schema.tables

sql
SELECT 
    TABLE_SCHEMA, 
    TABLE_NAME, 
    TABLE_TYPE 
FROM information_schema.tables 
WHERE TABLE_NAME = '조회할_이름';

BASE TABLE

VIEW

SYSTEM VIEW

SHOW CREATE

sql
    SHOW CREATE VIEW 테이블_혹은_뷰_이름;

sql
    SHOW CREATE TABLE 테이블_혹은_뷰_이름;

CREATE TABLE ...

sql
-- 비동기 MV의 정의를 확인하여 기반이 되는 뷰/테이블 파악
SHOW CREATE MATERIALIZED VIEW MV_이름;

CREATE MATERIALIZED VIEW ... AS SELECT ...

FROM

SELECT TABLE_TYPE FROM information_schema.tables WHERE TABLE_NAME = '...'

SHOW CREATE VIEW ...

SHOW CREATE TABLE ...

enable_get_stats_from_external_metadata

ANALYZE

ANALYZE

enable_get_stats_from_external_metadata = true

enable_get_stats_from_external_metadata = false

ANALYZE

cn.warn

txn_log of txn: [ID] not found and can not find the tablet_meta

CANCEL LOAD

transaction_clean_interval

sql
    -- 트랜잭션 상태 확인
    SELECT * FROM information_schema.be_txns WHERE TXN_ID = [에러로그의_ID];

    -- 최근 취소된 로드 작업 확인
    SHOW LOAD WHERE STATE = 'CANCELLED';

Transaction cancel

WARN

starrocks_fe_txn_publish_ack_latency_ms

ready-to-finish

VISIBLE

finish

Publish

publish_ack_latency

SELECT

summary

ANALYZE

sql
    ANALYZE TABLE 테이블명;

sql
    ANALYZE TABLE 테이블명 (컬럼1, 컬럼2);

sql
    ANALYZE SAMPLE TABLE 테이블명;

CREATE ANALYZE

sql
CREATE ANALYZE [FULL | SAMPLE] TABLE 테이블명 (컬럼1, 컬럼2)
PROPERTIES (
    "statistic_auto_collect_ratio" = "0.2",        -- 데이터의 20%가 변경되면 수집 트리거
    "statistic_auto_collect_interval" = "3600"    -- 수집 주기 (단위: 초, 예: 1시간마다 체크)
);

statistic_auto_collect_ratio

statistic_auto_collect_interval

statistic_sample_collect_rows

sql
    SHOW ANALYZE STATUS [WHERE TABLE_NAME = '테이블명'];

sql
    SHOW ANALYZE JOB;

sql
    DROP ANALYZE JOB [Job_ID];

sql
    SHOW STATS META [WHERE TABLE_NAME = '테이블명'];

statistic_collect_interval_sec

ANALYZE

ratio

interval

EXPLAIN ANALYZE

IcebergScanNode

HdfsScanNode

NumFiles

TotalRawSize

NumFiles

TotalRawSize

WaitRowsTime

ScanTime

FileOpenTime

MetadataFetchTime

IOBufferFullWaitTime

sql
-- 테이블 내 파일들의 평균 크기 및 개수 확인
SELECT 
    count(*) as total_file_count,
    avg(file_size_in_bytes) / 1024 / 1024 as avg_file_size_mb,
    sum(file_size_in_bytes) / 1024 / 1024 / 1024 as total_gb
FROM <catalog_name>.<db_name>."<table_name>$files";

starrocks_fe_external_catalog_metadata_request_duration

starrocks_be_scan_bytes_per_read

rewrite_data_files

EXPLAIN

EXPLAIN COSTS <쿼리>

EXPLAIN VERBOSE <쿼리>

EXPLAIN ANALYZE <쿼리>

sql
    SET show_statistics_item = true;
    EXPLAIN COSTS <쿼리>;

EXPLAIN COSTS

sql
    -- 기본 EXPLAIN만 쳐도 COSTS 수준으로 나오게 설정 (v3.3+ 기본값)
    SET query_detail_explain_level = 'COSTS';

{row: 100, cpu: 50.0, memory: 10.0, network: 0.0, cost: 200.5}

[min, max, null_count, avg_size, ndv]

SHOW

SHOW PROCESSLIST

sql
-- 실행 중인 쿼리 목록 확인
SHOW PROCESSLIST;

-- 전체 SQL 문장을 포함하여 확인
SHOW FULL PROCESSLIST;

SHOW RUNNING QUERIES

sql
SHOW RUNNING QUERIES;

SHOW ANALYZE STATUS

ANALYZE

sql
SHOW ANALYZE STATUS [WHERE TABLE_NAME = '테이블명'];

PENDING

RUNNING

FINISHED

FAILED

SHOW LOAD

STREAM LOAD

sql
SHOW LOAD [FROM 데이터베이스] ORDER BY createtime DESC LIMIT 1;

ETL:100%; LOAD:100%

SHOW

Queries

get_query_profile

SHOW PROCESSLIST

SHOW RUNNING QUERIES

Running

sql
    SELECT get_query_profile('복사한_QueryId');

IcebergScanNode

http://<FE_IP>:8030

sql
SHOW FULL PROCESSLIST;

sql
SHOW BACKENDS;

FileOpenTime

RowsReturned

jdbc_connection_pool_size

jdbc_minimum_idle_connections

jdbc_connection_timeout_ms

jdbc_connection_max_lifetime_ms

wait_timeout

jdbc_connection_idle_timeout_ms

be.conf

shell
    # be.conf 예시
    jdbc_connection_pool_size = 16

fe.conf

sql
    -- 동적 변경 예시 (관리자 권한 필요)
    ADMIN SET FRONTEND CONFIG ("jdbc_connection_pool_size" = "16");

Connection is not available

jdbc_connection_pool_size

BE 노드 수 * jdbc_connection_pool_size

SESSIONS

PROCESSES

jdbc_connection_max_lifetime_ms

IDLE_TIMEOUT

jdbc_connection_idle_timeout_ms

concurrency_limit

sql
    CREATE RESOURCE GROUP oracle_protection_group
    TO (db='oracle_catalog_db') -- 특정 DB(Catalog) 접속 시 적용
    WITH (
        "concurrency_limit" = "5", -- Oracle로 동시에 나가는 쿼리를 5개로 제한
        "cpu_weight" = "10",
        "mem_limit" = "20%"
    );

is_unpartitioned_column_with_range

jdbc_uri

defaultRowPrefetch

Running

SHOW PROC '/global_current_queries'

sql
SHOW PROC '/global_current_queries';

sql
    ANALYZE PROFILE FROM '복사한_QueryId';

IcebergScanNode

WHERE

JOIN

GROUP BY

DISTINCT

merge_condition

merge_condition

version

_statistics_.predicate_columns

enable_collect_full_statistic

ResultSink

ResultSink

ANALYZE PROFILE FROM 'QueryID';

RowsRead

RowsReturned

SerializeChunkTime

RpcAvgTime

RpcCount

query_timeout

query_timeout

brpc_connection_type

RPC error

connection closed

ResultSink

WHERE

enable_statistic_collect = false

sql
    -- FE 설정 변경 (ADMIN SET CONFIG)
    ADMIN SET FRONTEND CONFIG ("enable_predicate_columns_collection" = "false");

INSERT INTO ... SELECT

sql
    ADMIN SET FRONTEND CONFIG ("enable_statistic_collect_on_first_load" = "false");

enable_statistic_collect

predicate_columns

NDV

NULL COUNT

MIN/MAX

_statistics_

predicate_columns

sql
-- 1. 전역 자동 통계 수집 중지
ADMIN SET FRONTEND CONFIG ("enable_statistic_collect" = "false");
-- 2. 쿼리 기반 자동 수집(v3.4+) 중지
SET GLOBAL enable_query_trigger_analyze = false;
-- 3. 불필요한 메타 기록 중지
ADMIN SET FRONTEND CONFIG ("enable_predicate_columns_collection" = "false");

enable_statistic_collect = false

predicate_columns

sql
-- 1. 전역 자동 통계 수집 중지
ADMIN SET FRONTEND CONFIG ("enable_statistic_collect" = "false");
-- 2. 쿼리 기반 자동 수집(v3.4+) 중지
SET GLOBAL enable_query_trigger_analyze = false;
-- 3. 불필요한 메타 기록 중지
ADMIN SET FRONTEND CONFIG ("enable_predicate_columns_collection" = "false");

Analyze

loads_history_sync_interval_second

sql
-- 최근에 실행된 통계 수집 작업이 있는지 확인 (Status가 RUNNING인 것이 없어야 함)
SHOW ANALYZE STATUS;

-- 예약된 자동 수집 작업이 있는지 확인
SHOW ANALYZE JOB;

User: root

sql
    ALTER SYSTEM ADD FOLLOWER "새노드IP:9010";

sql
    ALTER SYSTEM ADD OBSERVER "새노드IP:9010";

meta

bash
# fe/conf/fe.conf 에 정의된 meta_dir 경로 확인 (기본값은 fe/meta)
rm -rf /path/to/fe/meta/*

--helper

bash
./bin/start_fe.sh --helper 리더IP:9010 --daemon

ALTER SYSTEM ADD

fe.conf

priority_networks

SHOW FRONTENDS;

Alive

priority_networks

FE with the same host : already exists

ALTER SYSTEM ADD FOLLOWER/OBSERVER

sql
    SHOW FRONTENDS; -- 또는 SHOW PROC '/frontends';

ALTER SYSTEM ADD

bash
    # (이미 meta가 생성되었다면 삭제 후)
    ./bin/start_fe.sh --helper <리더IP>:9010 --daemon

DROP

DROP

sql
    -- 1. 기존 잘못된 노드 정보 삭제 (리더 FE에서 실행)
    ALTER SYSTEM DROP FRONTEND "해당IP:9010";

    -- 2. 다시 추가
    ALTER SYSTEM ADD FOLLOWER "해당IP:9010";

SHOW FRONTENDS;

IP

priority_networks

ADD

ADD

bash
ps -ef | grep StarRocksFE

bash
./bin/stop_fe.sh

v3.3

-g

bash
./bin/stop_fe.sh -g

kill

bash
# 위 1번에서 확인한 PID를 입력 (예: PID가 1234인 경우)
kill -9 1234

fe.pid

fe/conf/fe.conf

meta_dir

fe/meta

image

fe.pid

bash
# 예시 경로 (실제 meta_dir 경로에 따라 다를 수 있음)
rm /path/to/fe/meta/image/fe.pid

bash
./bin/start_fe.sh --daemon

trace times all select * ;

TRACE

sql
TRACE { TIMES | VALUES | LOGS | ALL } [ <module> ] <query_statement>

BASE

MV

OPTIMIZER

SCHEDULE

EXTERNAL

TRACE TIMES SELECT _ FROM table;

TRACE ALL SELECT _ FROM table;

TRACE TIMES OPTIMIZER SELECT _ FROM table;

TRACE

SELECT aurora_version();

SHOW FRONTENDS;

TRACE

sql
-- 기본(BASE) 모듈의 시간 분석
TRACE TIMES SELECT * FROM your_table LIMIT 10;

-- 옵티마이저 단계만 상세 분석
TRACE TIMES OPTIMIZER SELECT * FROM your_table LIMIT 10;

TRACE

sql
EXPLAIN ANALYZE SELECT * FROM your_table LIMIT 10;

_statistics_

external_column_statistics

sql
SHOW DATA FROM _statistics_;

Size

1.2 GB

500 MB

sql
SHOW PARTITIONS FROM _statistics_.external_column_statistics;

fe.conf

DELETE

ANALYZE

_statistics_.external_column_statistics

ANALYZE

WITH HISTOGRAM

-Xmx

statistics_max_full_collect_data_size

SET GLOBAL

fe.audit.log

fe/log/fe.audit.log

bash
    grep "변수명" fe.audit.log | grep "SET"

sql
    SELECT timestamp, user, clientIp, stmt
    FROM starrocks_audit_db*.starrocks_audit_tbl*
    WHERE stmt LIKE '%SET GLOBAL 변수명%';

SYSTEM-level OPERATE

ALTER USER

ALTER USER ... SET PROPERTIES ('session.변수명' = '값')

REFRESH CONNECTIONS;

fe.conf

fe.conf

free

sql
-- 현재 실행 중인 쿼리 목록
SHOW PROCESSLIST;
-- 혹은 더 상세하게
SHOW PROC '/current_queries';

QueryContext

sql
SHOW PROC '/mem_info';

bash
# CN/BE 노드 터미널에서 실행
curl -X POST http://<CN_IP>:<HTTP_PORT>/mem_release

cn.out

mem_info

sql
    -- 'jdbc_catalog' 내의 'sales_db'를 조회하는 모든 쿼리를 rg1에 할당
    CREATE RESOURCE GROUP rg1
    TO (db='sales_db') -- JDBC Catalog 내부의 DB명을 지정
    WITH ("concurrency_limit" = "5", "cpu_weight" = "10");

db

sql
    CREATE RESOURCE GROUP rg_jdbc
    TO (user='jdbc_user')
    WITH ("concurrency_limit" = "10");

sql
    SET resource_group = 'rg_jdbc';
    SELECT * FROM jdbc_catalog.db.table;

concurrency_limit

jdbc_connection_pool_size

enable_statistic_collect

enable_collect_full_statistic

enable_predicate_columns_collection

WHERE

JOIN

GROUP BY

statistic_auto_collect_predicate_columns_threshold

enable_statistic_collect_on_first_load

INSERT

enable_statistic_collect_on_update

UPDATE

enable_query_trigger_analyze = false

SELECT

PUBLISH

shared_data

enable_statistic_collect

SELECT

_statistics_.column_statistics

INSERT

PUBLISH

shared_data

PUBLISH

shared_data

sql
    -- 어떤 테이블에서 트랜잭션이 발생하는지 확인 가능
    SHOW PROC '/transactions';

analyze_...

sql
    -- 현재 돌아가고 있는 통계 수집 작업 확인
    SHOW ANALYZE STATUS;

SELECT

PUBLISH

PUBLISH

_statistics_

sql
-- 어떤 데이터베이스/테이블에서 컴팩션이 도는지 리스트가 나옵니다.
SHOW PROC '/compactions';

Partition

_statistics_

sql
-- 최근에 완료되거나 진행 중인 트랜잭션의 레이블과 테이블 명을 확인합니다.
SHOW PROC '/transactions';

enable_statistic_collect = false

PUBLISH

_statistics_

shared_data

task_run_history

information_schema.task_runs

DELETE

loads_history

predicate_columns

WHERE

JOIN

GROUP BY

SELECT

column_statistics

enable_statistic_collect = false

external_column_statistics

ANALYZE

enable_query_trigger_analyze = false

enable_statistic_collect = false

enable_predicate_columns_collection = false

SELECT

sql
    -- 설정이 ON으로 되어 있는지 확인
    ADMIN SHOW AUTOMATED CLUSTER SNAPSHOT;
    -- 또는 설정 변수 확인
    SHOW VARIABLES LIKE '%automated_cluster_snapshot%';

sql
SELECT * FROM information_schema.cluster_snapshot_jobs 
ORDER BY created_time DESC;

state

ERROR

error_message

checkpoint

image

automated_cluster_snapshot_interval_seconds

sql
    SHOW STORAGE VOLUMES;

fe.log

ClusterSnapshot

automated_cluster_snapshot

SnapshotManager

sql
-- Leader FE에서 메타데이터 이미지(Checkpoint) 생성을 강제로 시작합니다.
ALTER SYSTEM CREATE IMAGE;

triggering a new checkpoint manually...

finished save image...

automated_cluster_snapshot_{timestamp}

manual

sql
-- 수동으로 특정 이름을 가진 클러스터 스냅샷 생성
CREATE CLUSTER SNAPSHOT <snapshot_name> 
[STORAGE VOLUME <storage_volume_name>];

sql
-- 현재 진행 중인 스냅샷 작업의 상태(SNAPSHOTING, FINISHED, ERROR 등) 확인
SELECT * FROM information_schema.cluster_snapshot_jobs 
ORDER BY created_time DESC;

-- 생성 완료된 스냅샷 목록 및 경로 확인
SELECT * FROM information_schema.cluster_snapshots;

_statistics_.external_column_statistics

ANALYZE

enable_get_stats_from_external_metadata = true

EXPLAIN COSTS

sql
-- 옵션 OFF 상태에서 실행
EXPLAIN COSTS SELECT ... ;

-- 옵션 ON 상태에서 실행
EXPLAIN COSTS SELECT ... ;

cardinality

column stats

range

SET enable_profile = true;

Query Planning

sql
SELECT * FROM _statistics_.external_column_statistics 
WHERE table_name = 'your_iceberg_table_name';

DELETE

sql
-- Iceberg 메타데이터 테이블 조회 (v3.4.1+)
SELECT * FROM your_catalog.your_db.your_table$files;

column_sizes

value_counts

nan_value_counts

ndv

enable_get_stats_from_external_metadata = true

EXPLAIN COSTS

sql
-- 옵션 OFF/ON 각각 수행 후 결과 비교
EXPLAIN COSTS <당신의_쿼리>;

column statistics

ndv

ndv

ndv

enable_metadata_profile

sql
SET enable_metadata_profile = true;
-- 쿼리 실행 후
SHOW PROFILE;

Query Planning

sql
-- 테이블명 뒤에 $files를 붙여 조회
SELECT column_sizes, value_counts, null_value_counts, lower_bounds, upper_bounds 
FROM <catalog>.<db>.<table_name>$files;

lower_bounds

upper_bounds

cardinality

_statistics_.loads_history

ALTER TABLE ... DROP PARTITION

DROP PARTITION

sql
    SHOW PARTITIONS FROM _statistics_.loads_history;

PartitionName

sql
-- 올바른 문법 예시
ALTER TABLE _statistics_.loads_history DROP PARTITION p20240925;

loads_history_retained_days

sql
    ADMIN SET FRONTEND CONFIG ("loads_history_retained_days" = "14");

SHOW PARTITIONS FROM _statistics_.loads_history;

DROP PARTITION

ANALYZE

sql
    SHOW STATS <iceberg_table_name>;

sql
    SELECT * FROM _statistics_.external_column_statistics
    WHERE table_name = '테이블명';

sql
    SELECT * FROM <catalog>.<db>.<table_name>$files;

lower_bounds

sql
    EXPLAIN COSTS
    SELECT * FROM <iceberg_table_name> WHERE <컬럼명> = '값';

Estimates

cardinality

1

100

column statistics

ndv

range

enable_iceberg_column_statistics

snapshots

sql
SELECT * FROM <catalog>.<db>.<table_name>$snapshots;

summary

$files