When serializing lambda functions is it possible t...
# typescript
m
When serializing lambda functions is it possible to run a specific method and keep additional code inside the function but not in the handler?
Copy code
const lambda = async (event, context) => {
  const knex = Knex(...);
  knex.query(...)
}
this establishes a connection inside of the function invocation and as a result can't be used between separate invocations with lambda's container use but also takes time to start up each time as a result. The best practices suggest initialising sdks and connections outside of the handler https://docs.aws.amazon.com/lambda/latest/dg/best-practices.html is it possible to create lambdas a bit like in the below example?
Copy code
const knex = Knex(...);
const lambda = async (event, context) => {
  knex.query(...)
}
t
The factory will run once and return the lamba body back
m
ah nice one thanks i'll have a look at that
@tall-librarian-49374 works perfectly, thanks!
a
@miniature-arm-21874 can you share an example please?
m
Copy code
import { Callback, Context } from '@pulumi/aws/lambda';
import { BucketEvent } from '@pulumi/aws/s3';
import * as Knex from 'knex';

const conversationsToPostgresFactory = (): Callback<BucketEvent, void> => {
  const knex = Knex({
    client: 'pg',
    connection: {
      host: process.env.DB_HOST,
      user: process.env.DB_USER,
      port: Number(process.env.DB_PORT),
      password: process.env.DB_PASS,
      database: process.env.DB_NAME,
    },
    useNullAsDefault: true,
  });

  return async function conversationsToPostgres(
    event: BucketEvent,
    context: Context
  ): Promise<void> {
    context.callbackWaitsForEmptyEventLoop = false;

    const records = [{}];

    return await knex.transaction(async (transaction) => {
      await transaction('conversations').insert(records);
    });
  };
};

const sendToPostgresLambda = new aws.lambda.CallbackFunction(
  'conversation-data-to-postgres',
  {
    timeout: 150,
    memorySize: 256,
    callbackFactory: conversationsToPostgres,
    runtime: aws.lambda.NodeJS12dXRuntime,
    environment: {
      variables: {
        DB_HOST: rds.address,
        DB_USER: rds.username,
        DB_PASS: password.result,
        DB_PORT: rds.port.apply((port) => port.toString()),
        DB_NAME: rds.name,
      },
    },
    vpcConfig: {
      securityGroupIds: rds.vpcSecurityGroupIds,
      subnetIds: vpc.publicSubnetIds,
    },
    deadLetterConfig: {
      targetArn: sendToPostgresDLQueue.arn,
    },
  }
);
@alex
just noticed the @acceptable-stone-35112 didn't work in my previous message. did the above help?
a
@miniature-arm-21874 thanks a lot! So it closes over Callback and its async function conversationsToPostgres it returns is serialized as handler part of lambda, correct?
m
everything returned from
conversationsToPostgresFactory
becomes the lambda function code, and the handler is a function within that
👍 1