Error events take too much time to emit when fluentd is not available

[Shantanu35]

Hey, We are using your library to send logs from our lambda functions to fluentd via Bunyan Logger.

// Instance style fluent-logger creation
const sender = require('fluent-logger').createFluentSender('tag-prefix', {
            enableReconnect: false,
            host: 'localhost',
            port: 24224,
            reconnectInterval: 3000,
            requireAckResponse: true,
            timeout: 3.0
        });

// Listener for detecting error events
sender.on('error', (error) => {
      console.log('Error found on stream here. ' + error);
      // Printing the logs on stdout using the buffer.
});

// bunyan logger instance for sending logs to fluentd.
logger = bunyan.createLogger({
            name: 'FluentdLogger',
            streams: [{ stream: sender.toStream('fluentd') }]
        });

logger.info('this log record is sent to fluent daemon');

1. Using the above mentioned approach, the logs are successfully being sent to fluentd from lambda.
2. However, when fluentd is not available, then the expected behavior is to catch the error event during the creation of bunyan instance and send the logs to stdout.

However, we have noticed that it takes a significant amount of time for the error event to be emitted when fluentd is not available, before which the lambda is completing its execution thus not reaching the error handler.

setTimeout(()=> sender.end(), 5000);

The above snippet seems to not work for me.

Therefore, is there any way to:

1. Avoid delaying the lambda execution in the case when fluentd is not available and error event needs to be caught?
2. If not, then what is the best way to perform a timeout so as to make the lambda quit after the error listener code has been executed?

It would be great if anyone could advise me on it.

[okkez]

Why you set timeout: 3.0 ?

Set the socket to timetout after timeout milliseconds of inactivity on the socket.

This library will reconnect to fluentd if fluentd is not available. If you want to shutdown your Lambda function when Fluentd is down, you can throw an error to re-run your Lambda.

Please check the documentation for AWS Lambda about error handling and retry.

[Shantanu35]

timeout: 3.0 is the default timeout and as far as I know, it just indicates the amount of time after which socket will timeout. I have set enableReconnect: false, so that it does not try reconnecting.

The aim is that, if fluentd is not available, then the 'error' event emitted takes a significant amount of time, which causes the lambda to wait for it. Is there any efficient or faster way to send error events from the library in case fluentd is down?