AWS Lambda connect via PG.js to RDS Postgres database (connection made, no timeout, but no database?)

Question

I have my lambda function is trying to connect to an RDS PostGreSQL DB. Since I use https://serverless.com/ to deploy the function (sets up my cloudfront) it puts the LF in a separate VPC from the RDS DB.

Not a big issue. If you read: https://docs.aws.amazon.com/lambda/latest/dg/services-rds-tutorial.html you see you can setup the serverless.yml file (as below) with the subnet, and security Group IDs, and then give a role to the Lambda Function that has AWSLambdaVPCAccessExecutionRole (I gave it full for the VPC and for Lambda). If you don't do this you will get a ECONNREFUESED.

But even after doing this I get an 3D00 error, which says that the db with name "ysgdb" is not found. But in RDS I see it is there and is public.

The code works fine when the DB is set to a local PostGreSQL.

Any ideas where to go next?

# serverless.yml

service: lambdadb
provider:
  name: aws
  stage: dev
  region: us-east-2
  runtime: nodejs10.x
functions:
  dbConn:
    # this is formatted as <FILENAME>.<HANDLER>
    handler: main.handler
    vpc:
      securityGroupIds:
        - sg-a1e6f4c3
      subnetIds:
        - subnet-53b45038
        - subnet-4a2a7830
        - subnet-1469d358
    events:
    - http:
        path: lambdadb
        method: post
        cors: true
    - http:
        path: lambdadb
        method: get
        cors: true

---

REPLY

{
  "statusCode": 200,
  "headers": {
    "Content-Type": "application/json"
  },
  "body": {
    "name": "error",
    "length": 90,
    "severity": "FATAL",
    "code": "3D000",
    "file": "postinit.c",
    "line": "880",
    "routine": "InitPostgres"
  },
  "isBase64Encoded": false
}

---

server.js

const aws = {
    user: "postgres",
    host: "ysgdb.cxeokcheapqj.us-east-2.rds.amazonaws.com",
    database: "ysgdb",
    password: "***",
    port: 5432
}

console.log(`PostgreSQL GET Function`)

const { Client } = require('pg')

const local = {
    user: "postgres",
    host: "localhost",
    database: "m3_db",
    password: "xxxx",
    port: 5432
}

const aws = {
    user: "postgres",
    host: "ysgdb.cxeokcheapqj.us-east-2.rds.amazonaws.com",
    database: "ysgdb",
    password: "xxxx",
    port: 5432
}

let response = {
    "statusCode": 200,
    "headers": {
        "Content-Type": "application/json"
    },
    "body": 'none',
    "isBase64Encoded": false
}

exports.handler = async (event, context, callback) => {
    const c = aws // switch to local for the local DB
    console.log(`aws credentials: ${JSON.stringify(c)}`)
    const client = new Client({
        user: c.user,
        host: c.host,
        database: c.database,
        password: c.password,
        port: c.port
    })
    try {
        await client.connect();
        console.log(`DB connected`)
    } catch (err) {
        console.error(`DB Connect Failed: ${JSON.stringify(err)}`)
        response.body = err
        callback(null, response)
    }

    client.query('SELECT NOW()', (err, res) => {
        if (err) {
            console.log('Database ' + err)
            response.body = err
            callback(null, response)
        } else {
            response.body = res
            callback(null, response)
        }
        client.end()
    })
}


if (process.env.USERNAME == 'ysg4206') {
    this.handler(null, null, (_, txt) => {console.log(`callback: ${JSON.stringify(txt)}`)})
} 

Answer 1

I found the answer. While it shows my stupidity, I wanted to post it here in case others have the same issue.

When I created the RDS (custom) PostGreSQL I specified I database name. Little did I know that about 20 fields down in the form, there is a second location you specify the dbname. If you do not, it says it does not create the db and assigns no name (the console shows "-" as the name.

The first name is for the first part of the endpoint. The second time you specify the name is for the database itself.

I am using the PostGreSQL for the security/login when I setup the RDS.

Hope this helps someone.