CODEWITHSUNDEEP
node.jsredis
Chun
Chun

Reputation: 279

How does nodejs-redis(&connect-redis) deal with sync and async?

I used connect-redis for my session store, and when I use req.session, it seems all the operations on it are synchronized, it's like operating on ordinary Javascript variables, the code obey the order. but I check the source code, which uses the asynchronized way, so I wonder why the req.session acts like that.

Another question is that if I have multiple redis queries,

client.sadd('test', 1);
client.del('test');
client.sadd('test', 2);
client.sadd('test', 3);

no matter where I put the del operation, the results always the same. I thought these queries might be run in any order right? since they all asynchronized called, so the results I expected should be different every time.

Thanks for you help

Upvotes: 3

Views: 519

Answers (1)

Didier Spezia
Didier Spezia

Reputation: 73216

The fact that roundtrips to the Redis server are managed asynchronously does not mean the queries will be sent in random order.

Redis (and therefore most Redis client libraries) supports pipelining, generally used to optimize the number of roundtrips. The idea is to send multiple queries, and then wait for the replies. The order is critical, because it is used by the client to match queries and replies.

Node.js is very well suited to support this kind of mechanisms. Matt Ranney's node_redis client supports pipelining in a transparent way. Provided the same client object is used, all the queries will be serialized and executed in order.

In your example, it is normal the queries are always executed in the same order. You can check this point by using the monitor command to display the flow of queries sent to Redis.

Now, it is important the last query of the pipeline is associated with a callback, otherwise your program will never know when the last query is complete.

Upvotes: 2

Related Questions