How correctly use ConnectionPool for mssql?

Question

How version ConnectionPool for mssql is better and why? I try understand, both work. I want a correct connection, I don't want create redundant connections.

All the examples I found use .then and version A, maybe is another better option?

Should I close connection?

version A

// db.js
import sql from "mssql"
const config = {...}
export const mssqlPool = new sql.ConnectionPool(config).connect()


// n endpoint.js
import { mssqlPool } from "../lib/mssql/db"
try {
  const pool = await mssqlPool
  const result = await pool.request().query(`SELECT * FROM table`)
  console.log(result)
} catch (err) {
  console.log('error')
}

version B

// db.js
import sql from "mssql"
const config = {...}
export const mssqlPool = new sql.ConnectionPool(config)


// n endpoint.js
import { mssqlPool } from "../lib/mssql/db"
try {
  const pool = await mssqlPool.connect();
  const result = await pool.request().query(`SELECT * FROM table`)
  console.log(result)
} catch (err) {
  console.log('error')
}

Edit: answer github