Skip to content

rpgeeganage/ifto

Repository files navigation

IFTO - If (Lambda) Timeout

License Language grade: JavaScript Codacy Badge Codacy Badge Build Status Known Vulnerabilities Maintainability

The purpose of this module is to work as a debugging tool during a timeout. It is hard to debug when a lambda exit exists with Task timed out after 2.00 seconds.

  • One solution is to increase the Timeout value. Although this fixed the problem (sometimes), there is no guarantee that, a timeout may not occur again until proper debugging has done.

  • Another solution is to a use a module like Debug and enable logging. The problem is this, will create log entries irrespective of the timeout.

How is this module work?

This package calculates possible timeout errors using getRemainingTimeInMillis() provided in the Context object. Context is passed during the executing of the lambda as mentioned below.

async handler(event: SQSEvent, context: Context) {
  ...
}

Configuration

(Environment variables)
  • ifto_start or IFTO_START

Please set the ifto_start as true in environment variables to allow the monitoring. (Monitoring will not start without setting the value as mentioned.)

  • ifto_flush_when or IFTO_FLUSH_WHEN

This indicates the minimum number of milliseconds remaining in context.getRemainingTimeInMillis(), before flushing the logs to the output. (default setting is to write using console.log). See the below.

(Default value is 50)

if context.getRemainingTimeInMillis() <= ifto_flush_when then
  flush the logs to standard output
end if

How to use it?

Assume that you lambda is structured as follows.

index.ts
lib/handler.ts
modifications in index.ts
/**
 * Just import the Ifto module.
 * This will, create a Ifto in the "global" name space.
 * So other files can use Ifto module without creating a object
 */
import 'Ifto';
import { Context, SQSEvent } from 'aws-lambda';
import { handler } from './lib/handler';

export async function myLambda(event: SQSEvent, context: Context) {
  Ifto
    .addLambdaContext(context) // Add the context module
    .init(process.env); // Add the process.env inorder to read the required configurations.

  // Use Ifto.log() to add log entry
  Ifto
    .log('My lambda execution started.');


  // Use Ifto.monitor() function to start monitoring, if configured properly.
  return Ifto.monitor(
    // Pass the handler function
    handler(event)
  );
}
modifications in lib/handler.ts
/**
 * below import statement will required, only if you write individual unit tests for this file
 */
import 'Ifto';
import { SQSEvent } from 'aws-lambda';

/*
 * You don't have to pass the context object here
 */
export async function handler(event: SQSEvent) {
  ...
  Ifto.log('log entry 1');

  ...
  Ifto.log('log entry 2');

  ...
  Ifto.log('log entry 3');

}

Methods

  • Ifto.addLambdaContext(context);

set Context object passed to the lambda function.

  • Ifto.init(process.env);

Pass the process.env to read the environment variables defined.

  • Ifto. log(logMessage);

    Adds the log message, so in the case of timeout, this message will be flushed to the standard output.

  • Ifto.monitor(Promise);

Accepts the handler function to in order to monitoring. (This won't change the return values or errors thrown buy the handler function).

Final output - (in case of Timeout error)

2019-01-20T11:27:26.578Z    ccbc1d49-3336-47c2-9d49-c1d47ffc23de
Expecting a possible lambda timeout.
Only 50 milliseconds remaining.
(If this is a false positive error, change value by setting up environment variable "ifto_flush_when").
Current log:
2019-0-0T11:27:24.629 0: My lambda execution started.
2019-0-0T11:27:24.919 1: log entry 1
2019-0-0T11:27:24.956 2: log entry 2
2019-0-0T11:27:25.236 3: log entry 3

Spy modules 👽

These modules are used to spy on operations which can take more execution time. (More modules will be added in the future 🚀 🚀).

HTTP / HTTPS spy module 👮

This module keeps track (most recent 10) and outputs the URL of an HTTP or HTTPS request which is not complete at the time of flushing the logs.

Output with spy modules 👉

START RequestId: 55ca052e-45ae-49c5-89ef-bd1e5bfb0abf Version: $LATEST
2019-02-10T16:24:45.737Z	55ca052e-45ae-49c5-89ef-bd1e5bfb0abf
Expecting a possible lambda timeout.
Only 50 milliseconds remaining.
(If this is a false positive error change the value by setting up the environment variable "ifto_flush_when").
Current log:
2019-02-10T17:24:43.41 0: Handler entry
2019-02-10T17:24:43.53 1: Start operation 1
2019-02-10T17:24:43.54 2: Start operation 2
2019-02-10T17:24:45.35 3: Run HTTP requests

******************
Spied module logs:
******************

http

Possible unfinished HTTP requests
2019-02-10T17:24:45.53 https://ifto-spy-testing.free.beeceptor.com/a-request-which-takes-long-time-to-process-004
2019-02-10T17:24:45.41 https://ifto-spy-testing.free.beeceptor.com/a-request-which-takes-long-time-to-process-003
2019-02-10T17:24:45.40 https://ifto-spy-testing.free.beeceptor.com/a-request-which-takes-long-time-to-process-002
2019-02-10T17:24:45.38 https://ifto-spy-testing.free.beeceptor.com/a-request-which-takes-long-time-to-process-001

END RequestId: 55ca052e-45ae-49c5-89ef-bd1e5bfb0abf
REPORT RequestId: 55ca052e-45ae-49c5-89ef-bd1e5bfb0abf	Duration: 3003.15 ms	Billed Duration: 3000 ms 	Memory Size: 512 MB	Max Memory Used: 226 MB
2019-02-10T16:24:45.789Z 55ca052e-45ae-49c5-89ef-bd1e5bfb0abf Task timed out after 3.00 seconds

Important note

💡 The default of ifto_flush_when is 50 and it was decided by running couple of lambdas and printing context.getRemainingTimeInMillis() value. In case of false positive change this value.

💡 I haven't run any analysis on memory usage.

💡 I would thankfull to hear thoughts and bugs.