nodeJs MariaDB 3.2.0 -> MariaDB 3.2.1 (and subsequent) connector hanging issue on AWS lambda with RDS (MariaDB)

[Leigh-M]

When connector is installed with npm i mariadb@3.2.0 AWS lambda permits connection to RDS (MariaDB) with minimal config:

    const pool = mariadb.createPool({
        host: 'someHostName',
        user: 'someuser',
        database: 'someDB',
        password: 12345
        connectionLimit: 5,
        port: 3306
    });

Yet when installed with npm i mariadb@3.2.1 (and any subsequent later release) the connection just hangs with:

{
  "name": "SqlError",
  "sqlMessage": "retrieve connection from pool timeout after 10000ms\n    (pool connections: active=0 idle=0 limit=5)",
  "sql": null,
  "fatal": false,
  "errno": 45028,
  "sqlState": "HY000",
  "code": "ER_GET_CONNECTION_TIMEOUT"
}

The release notes mention only:

• Binary Result-Set Parsing Improvements (CONJS-262):
• New Option for infileStreamFactory (CONJS-266):

Neither of these appear to constitute such a breaking change? I assume there is another undocumented (unintended?) breaking change that needs to be understood and documented?

[rusher]

Not sure about the change in 3.2.0, i'll look at the changes, but meanwhile, could you try to set connectTimeout to a bigger value than de default 1000ms ?

Next version will have an improved pooling error that will show detailed error, like :

(pool connections: active=0 idle=0 limit=1)
  at module.exports.createError (C:\projects\mariadb-connector-nodejs\lib\misc\errors.js:65:21)
  at Pool._rejectTimedOutRequest (C:\projects\mariadb-connector-nodejs\lib\pool.js:326:33)
  at Pool._checkRequestTimeouts (C:\projects\mariadb-connector-nodejs\lib\pool.js:294:14)
  at listOnTimeout (node:internal/timers:581:17)
  at process.processTimers (node:internal/timers:519:7) {
sqlMessage: 'pool timeout: failed to retrieve a connection from pool after 205ms\n' +
'    (pool connections: active=0 idle=0 limit=1)',
sql: null,
fatal: false,
errno: 45028,
sqlState: 'HY000',
code: 'ER_GET_CONNECTION_TIMEOUT',
  [cause]: SqlError: (conn:-1, no: 45012, SQLState: 08S01) Connection timeout: failed to create socket after 1ms
    at module.exports.createFatalError (C:\projects\mariadb-connector-nodejs\lib\misc\errors.js:90:10)
    at Connection.connectTimeoutReached (C:\projects\mariadb-connector-nodejs\lib\connection.js:1099:24)
    at listOnTimeout (node:internal/timers:583:11)
    at process.processTimers (node:internal/timers:519:7)
  From event:
    at C:\projects\mariadb-connector-nodejs\lib\connection.js:142:13
    at new Promise (<anonymous>)
    at Connection.connect (C:\projects\mariadb-connector-nodejs\lib\connection.js:131:12)
    at Pool._createPoolConnection (C:\projects\mariadb-connector-nodejs\lib\pool.js:144:8)
    at Timeout._onTimeout (C:\projects\mariadb-connector-nodejs\lib\pool.js:278:18)
    at listOnTimeout (node:internal/timers:581:17)
    at process.processTimers (node:internal/timers:519:7) {
  sqlMessage: 'Connection timeout: failed to create socket after 1ms',
  sql: null,
  fatal: true,
  errno: 45012,
  sqlState: '08S01',
  code: 'ER_CONNECTION_TIMEOUT'

[rusher]

i suppose it's due to e2a7e16, connectTimeout taking in account the time for retrieving a connection, not just establishing socket to server

[crushton-qg]

@rusher Hi, I am currently running version 3.3.1 and it is being used in an enterprise level application, and over the last 3/4 weeks (we think), we have started to see sporadic errors relating to socket not being obtained

2025-06-29T18:36:21.842Z	bf3d2e0c-62be-4451-987f-9b49962b27ff	ERROR	SqlError: (conn:-1, no: 45012, SQLState: 08S01) Connection timeout: failed to create socket after 1001ms
    at module.exports.createFatalError (/opt/nodejs/node_modules/mariadb/lib/misc/errors.js:90:10)
    at Connection.connectTimeoutReached (/opt/nodejs/node_modules/mariadb/lib/connection.js:1080:24)
    at listOnTimeout (node:internal/timers:583:11)
    at process.processTimers (node:internal/timers:519:7)
  sql: null,
  fatal: true,
  errno: 45012,
  sqlState: '08S01',
  code: 'ER_CONNECTION_TIMEOUT'
}

In terms of how sporadic, at roughly 5pm UK time yesterday, we checked one Lambda function which had been invoked 34210 times in the past 24 hours, and we saw this error 4 times, so very minimal, but we need to understand the issue better and fix

we have checked the DB monitoring in AWS and cannot see anything out of the ordinary, just wondering if this is related to the above issue is related to this and whether this is a known issue and/or a fix has been made with a new version?

Also worth noting that we do not explicitly set a socket or connection timeout in our connection configuration so default values will be used here

Thanks in advance,
Charlie

[rusher]

Lambda are nice because not up when not used, but first response time when down is usually long, more than a second. have you set a connectTimeout to a bigger value than default 1s. When using lamdba, it's better to have connection timeout set to a bigger value