vimson/dynamodb-logger

A DynamoDB wrapper which logs all the entries to the table and an easy way to query the logs

DynamoDB wrapper for logging and scanning the logs in an easy manner

We all know the pain of finding the data from the logs especially when working with Lambda function in serverless environment. I agree cloudwatch is one of the great options for logging. But we need an easy solution to scan the logs without using the setup like Elasticsearch stack. That's why we developed this small library which is basically a DynamoDb wrapper, which easily do the basic insert and scan operation very easily.

Table name: LambdaLogs

Partition key	Sort key
PK (String)	SK (String)

Local secondary indexes

Name	Partition key	Sort key	Projected attributes
LSI-1-index	PK (String)	LSI-1-SK (String)	All
LSI-2-index	PK (String)	LSI-2-SK (String)	All
LSI-3-index	PK (String)	LSI-3-SK (String)	All

The above is the default Table & Index convention. But it is totally configurable. Deafult configuration is

const config: LoggerConfig = {
  region: 'eu-west-1',
  table: {
    name: 'LambdaLogs',
    partitionKey: 'PK',
    sortKey: 'SK',
    lsi1SortKey: 'LSI-1-SK',
    lsi2SortKey: 'LSI-2-SK',
    lsi1Index: 'LSI-1-index',
    lsi2Index: 'LSI-2-index',
  },
};

Type details are

export type LoggerTableConfig = {
  name: string;
  partitionKey?: string;
  sortKey?: string;
  lsi1SortKey?: string;
  lsi2SortKey?: string;
  lsi1Index?: string;
  lsi2Index?: string;
};

export type LoggerConfig = {
  region: string;
  credentials?: Credentials; //import { Credentials } from '@aws-sdk/types';
  table?: LoggerTableConfig;
};

An example implementation if you follow the same table name and index names are

import { logger, getLoggerConfig } from 'dynamodb-logger';
logger.initailize(getLoggerConfig()); // If you are using the same Table and index name conventions as above

An example if you are using diferent table name and index names are then

import { logger, getLoggerConfig } from 'dynamodb-logger';
logger.initailize({
  region: 'eu-west-1',
  table: {
    name: 'LambdaLogs',
    partitionKey: 'PK',
    sortKey: 'SK',
    lsi1SortKey: 'LSI-1-SK',
    lsi2SortKey: 'LSI-2-SK',
    lsi1Index: 'LSI-1-index',
    lsi2Index: 'LSI-2-index',
  },
  credentials: {
    accessKeyId: 'XXXXXXX',
    secretAccessKey: 'YYYYYYYYY',
  },
}); // If you are using API keys and different table name and index names

It is better you can use a logger helper module to put this initialzation and can use in your project

Logging example is

logger.log(LogModules.LambdaInvoker, LogLevels.Payload, JSON.stringify(logStack, null, 2)); // we added LogModules & LogLevels as fixed constants

Scanning the log table example is

import { logger, getLoggerConfig } from 'dynamodb-logger';
logger.initailize(getLoggerConfig());

const searchOptions: LogSearchConfig = {
  module: LogModules.PushNotifier,
  level: LogLevel.ERROR,
  startTime: '2022-10-03T07:01:46.037Z',
  endTime: '2022-12-03T07:01:46.037Z',
  limit: 1,
  sort: 'asc',
};

let logResult: LogResult;
do {
  logResult = await logger.search(searchOptions);
  if (logResult.lastEvaluatedKey) {
    searchOptions.lastEvaluatedKey = logResult.lastEvaluatedKey;
  }
  console.log(logResult);
} while (logResult.lastEvaluatedKey);