Saving data to Postgres from AWS Lambda - node.js

Question

I'm building a lambda function that is supposed to save a game feedback, like a performance grade, into my Postgres database, which is in AWS RDS.
I'm using NodeJS typescript and the function is kinda working, but in a strange way.
I made an API Gateway so I can POST data to the URL to the lambda process it and save it, the thing is, when I POST the data the function seems to process it until it reaches a max limit of connected clients, and than it seems to lose the other clients'data.
Another problem is that every time I POST data I get a response saying that there was a Internal Server Error and with a 'X-Cache→Error from cloudfront' header. For a GET request I figured it out that it was giving me this response because the format of the response was incorrect, but in this case I fixed the response format and still get this problem...
Sometimes I get a timeout response.
My function's code:
import { APIGatewayEvent, Callback, Context, Handler } from "aws-lambda";
import { QueryConfig, Client, Pool, PoolConfig } from "pg";
export const insert: Handler = async (
event: APIGatewayEvent,
context: Context,
cb: Callback
) => {
// context.callbackWaitsForEmptyEventLoop = false;
const config: PoolConfig = {
user: process.env.PG_USER,
host: process.env.PG_HOST,
database: process.env.PG_DB,
password: process.env.PG_PASS,
port: parseInt(process.env.PG_PORT),
idleTimeoutMillis: 0,
max: 10000
};
const pool = new Pool(config);
let postdata = event.body || event;
console.log("POST DATA:", postdata);
if (typeof postdata == "string") {
postdata = JSON.parse(postdata);
}
let query: QueryConfig = <QueryConfig>{
name: "get_all_questions",
text:
"INSERT INTO gamefeedback (gameid, userid, presenterstars, gamestars) VALUES ($1, $2, $3, $4);",
values: [
parseInt(postdata["game_id"]),
postdata["user_id"],
parseInt(postdata["presenter_stars"]),
parseInt(postdata["game_stars"])
]
};
console.log("Before Connect");
let con = await pool.connect();
let res = await con.query(query);
console.log("res.rowCount:", res.rowCount);
if (res.rowCount != 1) {
cb(new Error("Error saving the feedback."), {
statusCode: 400,
body: JSON.stringify({
message: "Error saving data!"
})
});
}
cb(null, {
statusCode: 200,
body: JSON.stringify({
message: "Saved successfully!"
})
});
console.log("The End");
};
Than the log from CloudWatch error with max number of clients connected looks like this:
2018-08-03T15:56:04.326Z b6307573-9735-11e8-a541-950f760c0aa5 (node:1) UnhandledPromiseRejectionWarning: error: sorry, too many clients already
at u.parseE (/var/task/webpack:/node_modules/pg/lib/connection.js:553:1)
at u.parseMessage (/var/task/webpack:/node_modules/pg/lib/connection.js:378:1)
at Socket.<anonymous> (/var/task/webpack:/node_modules/pg/lib/connection.js:119:1)
at emitOne (events.js:116:13)
at Socket.emit (events.js:211:7)
at addChunk (_stream_readable.js:263:12)
at readableAddChunk (_stream_readable.js:250:11)
at Socket.Readable.push (_stream_readable.js:208:10)
at TCP.onread (net.js:607:20)
Can any of you guys help me with this strange problem?
Thanks

Answer 1

Well for one thing you need to put creating a pool above the handler, like so:
const config: PoolConfig = {
user: process.env.PG_USER,
...
};
const pool = new Pool(config);
export const insert: Handler = async (
event: APIGatewayEvent,
context: Context,
cb: Callback
) => {
..etc
The way you have it you are creating a pool on every invocation. If you create the pool outside the handler it gives Lambda a chance to share the pool between invocations.