risingwave: bug: database memory surges when running a lot of small batch queries

Describe the bug

When starting a cluster service through docker-compose, create a simple table (only four very simple fields) in the database, and then write data, and find that the memory grows rapidly. When 300,000 data are written, the memory has grown 10G or so. And there is no response to the query on the table at this time.(select count(*) from t).

create table t (id int,v1 int,v2 int,s1 varchar);

sql = "insert into t1 values({},{},{},'test_value');"
for i in range(1,1000000):
    sql = sql.format(i,i+1,i+2)
    cursor = conn.cursor()
    cursor.execute(sql)

To Reproduce

1. docker compose -f docker-compose.yml up -d
2. psql -h localhost -p 4566 -d dev -U root
3. create table t (id int,v1 int,v2 int,s1 varchar);
4. insert into t values({},{},{},‘test_value’);

this is docker-componse.yml

version: "3"
services:
  compactor-0:
    image: "ghcr.io/risingwavelabs/risingwave:v0.1.12"
    command:
      - compactor-node
      - "--host"
      - "0.0.0.0:6660"
      - "--state-store"
      - "hummock+minio://hummockadmin:hummockadmin@minio-0:9308/hummock001"
      - "--meta-address"
      - "http://meta-node-0:5690"
    expose:
      - "6660"
    ports: []
    depends_on:
      - meta-node-0
      - minio-0
    environment:
      RUST_BACKTRACE: "1"
    container_name: compactor-0
    
  compute-node-0:
    image: "ghcr.io/risingwavelabs/risingwave:v0.1.12"
    command:
      - compute-node
      - "--host"
      - "0.0.0.0:5688"
      - "--client-address"
      - "compute-node-0:5688"
      - "--state-store"
      - "hummock+minio://hummockadmin:hummockadmin@minio-0:9308/hummock001"
      - "--meta-address"
      - "http://meta-node-0:5690"
    expose:
      - "5688"
    ports: []
    depends_on:
      - meta-node-0
      - minio-0
    environment:
      RUST_BACKTRACE: "1"
    container_name: compute-node-0
  etcd-0:
    image: "quay.io/coreos/etcd:latest"
    command:
      - /usr/local/bin/etcd
      - "--listen-client-urls"
      - "http://0.0.0.0:2388"
      - "--advertise-client-urls"
      - "http://etcd-0:2388"
      - "--listen-peer-urls"
      - "http://0.0.0.0:2389"
      - "--initial-advertise-peer-urls"
      - "http://etcd-0:2389"
      - "--listen-metrics-urls"
      - "http://0.0.0.0:2379"
      - "--name"
      - risedev-meta
      - "--max-txn-ops"
      - "999999"
      - "--auto-compaction-mode"
      - periodic
      - "--auto-compaction-retention"
      - 1m
      - "--snapshot-count"
      - "10000"
    expose:
      - "2388"
    ports:
      - "2388:2388"
      - "2389:2389"
    depends_on: []
    volumes:
      - "/Users/wangdexin/mnt/etcd/etcd-0:/etcd-data"
    environment: {}
    container_name: etcd-0
  frontend-node-0:
    image: "ghcr.io/risingwavelabs/risingwave:v0.1.12"
    command:
      - frontend-node
      - "--host"
      - "0.0.0.0:4566"
      - "--meta-addr"
      - "http://meta-node-0:5690"
    expose:
      - "4566"
    ports:
      - "4566:4566"
    depends_on:
      - meta-node-0
    environment:
      RUST_BACKTRACE: "1"
    container_name: frontend-node-0
    
  
  meta-node-0:
    image: "ghcr.io/risingwavelabs/risingwave:v0.1.12"
    command:
      - meta-node
      - "--listen-addr"
      - "0.0.0.0:5690"
      - "--host"
      - meta-node-0
      - "--backend"
      - etcd
      - "--etcd-endpoints"
      - "etcd-0:2388"
    expose:
      - "5690"
      - "1250"
      - "5691"
    ports: []
    depends_on: []
    environment:
      RUST_BACKTRACE: "1"
    container_name: meta-node-0
  
  minio-0:
    image: "quay.io/minio/minio:latest"
    command:
      - server
      - "--address"
      - "0.0.0.0:9308"
      - "--console-address"
      - "0.0.0.0:9400"
      - /data
    expose:
      - "9308"
      - "9400"
    ports:
      - "9308:9308"
      - "9400:9400"
    depends_on: []
    volumes:
      - "/Users/wangdexin/mnt/minio/minio-0:/data"
    entrypoint: "\n/bin/sh -c '\nset -e\nmkdir -p \"/data/hummock001\"\n/usr/bin/docker-entrypoint.sh \"$$0\" \"$$@\"\n'"
    environment:
      MINIO_CI_CD: "1"
      MINIO_ROOT_PASSWORD: hummockadmin
      MINIO_ROOT_USER: hummockadmin
    container_name: minio-0

  zookeeper:
    image: confluentinc/cp-zookeeper:7.0.1
    environment:
      ZOOKEEPER_CLIENT_PORT: 2181
      ZOOKEEPER_TICK_TIME: 2000
      KAFKA_OPTS: "-Dzookeeper.4lw.commands.whitelist=ruok"
    expose:
      - "2181"
    ports:
      - "2181:2181"
    healthcheck:
      {
        test: echo ruok | nc localhost 2181 || exit -1,
        interval: 1s,
        timeout: 5s,
        retries: 3,
        start_period: 120s
      }

  kafka:
    image: wurstmeister/kafka
    environment:
      #KAFKA_ZOOKEEPER_CONNECT: zookeeper:2181
      #KAFKA_ADVERTISED_LISTENERS: PLAINTEXT://127.0.0.1:9092
      #KAFKA_OFFSETS_TOPIC_REPLICATION_FACTOR: 1
      #KAFKA_COMPRESSION_TYPE: lz4
      #KAFKA_LOG_RETENTION_MS: 31104000000
      KAFKA_ZOOKEEPER_CONNECT: zookeeper:2181
      KAFKA_LISTENER_SECURITY_PROTOCOL_MAP: INSIDE:PLAINTEXT,OUTSIDE:PLAINTEXT
      KAFKA_LISTENERS: INSIDE://:9092,OUTSIDE://127.0.0.1:9091
      KAFKA_ADVERTISED_LISTENERS: INSIDE://:9092,OUTSIDE://127.0.0.1:9091
      KAFKA_INTER_BROKER_LISTENER_NAME: INSIDE
    depends_on:
      zookeeper: { condition: service_healthy }
    expose:
      - "9092"
    ports:
      - "9091:9091"
    healthcheck:
      {
        test: nc -z localhost 9092,
        interval: 5s,
        start_period: 120s
      }



volumes:
  etcd-0:
    external: false
  minio-0:
    external: false

name: risingwave-compose

Expected behavior

The memory grows quickly, and when more than 300,000 data is inserted, the query does not respond. select count(*) from t;

Additional context

This is the result whether I create a materialized view based on the table or not.

Oh, I glued it with indentation, and it was gone when the submission was completed. You can click edit and you can copy the indented yml.