Search code examples
redisredis-cliredis-serverredisclientclientcache

Redis client-side cache (tracking) test

I'm testing Redis client-side caching using redis-cli.

Redis server version: 7.0.11

redis-cli version: 7.0.11

redis-cli 1 I started with the monitor command:

redis-cli monitor

In redis-cli 2 I run the snippet:

HELLO: 3
CLIENT TRACKING ON PREFIX test: BCAST NOLOOP;
SET test:1 "a"
GET test:1
GET test:1

In the redis-cli 1 (monitor) every get command is shown.

1687010819.444931 [0 127.0.0.1:56346] "CLIENT" "TRACKING" "ON" "PREFIX" "test:" "BCAST" "NOLOOP"
1687010843.136642 [0 127.0.0.1:56346] "SET" "test:1" "a"
1687010854.656921 [0 127.0.0.1:56346] "GET" "test:1"
1687010856.557103 [0 127.0.0.1:56346] "GET" "test:1"

I wasn't expecting to see the second GET command on the monitor. As I understand it, the second get shouldn't call the server, it should get the data from the redis-cli local cache.

Do I need to implement caching somehow? Does redi-cli not implement it? The Redis client for Ruby(redis-5.0.6) has the same behaviour, with or without hiredis(hiredis-client-0.14.1)

Solution

  • Yes, you need to implement the local cache part at client side. Normally, a client library should implement it.

    However, AFAIK, both redis-cli and hiredis do not implement this feature.