Redis Client Pool - How does the redis connection pool work?

[gspaiva]

Hey Quarkus team, I got a question.

I'm setting up a quarkus application that uses the quarkus redis-client library, I would like to have a connection pool with ready connections to be used. Although, when starting the application I don't see ready connections being opened in the Redis server, the number of connections opened only changes when I request the endpoint that interacts with Redis. I'm using the ReactiveRedisDataSource and RedisDataSource to interact with Redis.

I was expecting that the client would open the number of connections I configured via the quarkus.redis.max-pool-size property, however, it seems like it's opening them lazily. Although that's not much of a problem, I checked that once I'm done sending commands to Redis for the request, the connection is closed. I connected to a Redis instance directly and saw that although I made a lot of requests to my endpoint, the Redis connection number always dropped afterward, which means the connections need to be reopened on every request. Isn't that a performance issue?

For checking the number of active clients in the Redis server, I used the redis-cli info command.

Thanks in advance. If this is not the right place to post these questions, let me know.

[quarkus-bot[bot]]

/cc @cescoffier (redis), @gsmet (redis), @machi1990 (redis)

[cescoffier]

The connections are created on demand (lazily) and recycled.

[cescoffier]

BTW, we you want to keep the same connection, use .withConnection.