Check mysql connection status

[aura6u83]

Move to sql to mariadb due to performance issue while accessing history, but discover unix_socket provide more performance. But how to confirm homeassistant is connected to unix socket instead of tcp.

Running both mariadb and homeassistant in same instance.

recorder:
  db_url: mysql://homeassistant:[email protected]/homeassistant_data?unix_socket=/var/run/mysqld/mysqld.sock&charset=utf8mb4
  purge_keep_days: 14

version: '3'
services:
  mariadb:
    image: mariadb
    container_name: mariadb
    restart: always
    environment:
      MARIADB_ROOT_PASSWORD: mariadb@LGKn8iSy
      MARIADB_USER: homeassistant
      MARIADB_PASSWORD: SJ9BX7W9
      MARIADB_DATABASE: homeassistant_data
    volumes:
      - /external/appdata/mysql:/var/lib/mysql
      - /var/run/mysqld:/var/run/mysqld
    ports:
      - 127.0.0.1:3306:3306

[YuryT]

I’m in between of moving my HA OS supervised setup to docker and actually decided to quick test linux sockets vs TCP. I got TCP to be ~3 times faster
My setup:

• Rockchip 3588 based SBC with 16G RAM
• Armbian Linux 6.1.43-vendor-rk35xx
• Docker 26

I followed example from stackoverflow on this compose setup (ipvlan, L2)

services:
  netshoot1:
    image: 'nicolaka/netshoot:latest'
    container_name: netshoot1
    tty: true
    shm_size: '2gb'
    volumes:
      - redis-socket:/run/redis-socket
    networks:
     e0lan:
       ipv4_address: 192.168.50.221
      
  netshoot2:
    image: 'nicolaka/netshoot:latest'
    container_name: netshoot2
    tty: true
    shm_size: '2gb'
    depends_on:
      - "netshoot1"
    volumes:
      - redis-socket:/run/redis-socket
    networks:
     e0lan:
       ipv4_address: 192.168.50.222

volumes:
  redis-socket:
networks:
  e0lan:
    driver: ipvlan
    driver_opts:
      parent: end0
      ipvlan_mode: l2
    ipam:
      driver: default
      config:
        - subnet: 192.168.50.0/24
          ip_range: 192.168.50.128/25
          gateway: 192.168.50.1

Anything I missed on socat part? Also - I did not see any spikes on IO during socat test