20

I have a Node application that uses Express and node_redis. I'm following the approach outlined in the Learning Node book and creating a single client for the life of the application. Given this approach, when do I call close() on the redis client? Do I even need to?

Relevant code

var express = require( 'express' ),
    redis = require( 'redis' );

var app = express(),
    config = require( './config/application' )[ app.get( 'env' ) ];

// create Redis client
var redisClient = redis.createClient();
redisClient.on( 'error', function( err ) {
  console.log( 'Error' + err );
} );
// select the database
redisClient.select( config.redis_database );

...
/* more setup, route defintions, etc. */
...

http.createServer( app ).listen( 4000, function() {
  console.log( 'Server started and ready for action!' );
});
Bart Jedrocha
  • 11,450
  • 5
  • 43
  • 53

2 Answers2

19

You have a couple options.

  1. Be lazy and set an idle timeout for all clients on the redis server. Then when a client is idle for too long the server would just kill their connection.
  2. Kill the connection when the node process exits.

.

process.on("exit", function(){
    redisClient.quit();
});
Community
  • 1
  • 1
aembke
  • 2,479
  • 1
  • 15
  • 15
  • 5
    You can only perform synchronous operations inside "exit" event handler. The process will die before any "quit" command is send to the redis server. – Maxim Apr 02 '15 at 12:17
  • 1
    close() function doesn't exists. Use redisClient.quit(); – gyss Jul 28 '15 at 17:26
  • 3
    If you are killing the node process anyway, why do you need to explicitly "quit" the redis client? When the node process dies, any connection resources associated with it will be released/disconnected anyway. – Phil Apr 26 '21 at 15:33
  • @Phil, u wanna say if I disconnect from my data source, there are no reasons to disconnect redis? – HackerMF Mar 22 '23 at 16:20
7

: The question was 3 years ago and you might get answer already, anyway this might be useful for some new people.

You can make nodejs process listens to SIGINT which is interrupt sent from keyboard via the following code

process.on('SIGINT', function() {
    redisClient.quit();
    console.log('redis client quit');
});

Then you can either Ctrl+C from the keyboard, or send such signal via kill -SIGINT <process id>. When that happens, that code will be executed.

This is when we don't know exactly when to quit redis client, or we want external control to cleanly quit/clear resource we was using.

haxpor
  • 2,431
  • 3
  • 27
  • 46
  • 1
    In addition to SIGINT, it might also help to also listen to SIGTERM to handle the `kill ` command – daparic Sep 15 '17 at 22:36
  • @ifelsemonkey that makes sense too. – haxpor Sep 15 '17 at 22:49
  • Ultimately, is there no reason to quit executing each request? I mean, for example, we have a lot of different requests. In some of these requests, we need to get some redis data. Are there no reasons for each request to make a connection and at the end of this request to make a disconnection? – HackerMF Mar 17 '23 at 22:28