ERR max number of clients reached on Heroku free tier

[jstorm31]

I'm hosting my server on Heroku free Dyno and I've been getting (Redis) ERR max number of clients reached recenty.

I've found out it could be solved either by setting a shorter timeout on Heroku or connection pooling. There was a method for pooling in Vapor 3 req.withPooledConnection(to: .redis), but I haven't found such a method in Vapor 4, so I use directly req.redis.anyCommand(...).

My questions:

1. Is this just a free dyno problem or a problem in my application?
2. Is there a pooling of connections for Redis in Vapor 4?

EDIT: When I checked the Redis Heroku dashboard, there were 16 - 20 clients connected and never disconnected. I wonder how is that possible. 🤔 Nevertheless, I could kill all connections from Redis site with Heroku cli heroku redis:cli and running CLIENT KILL TYPE normal. I think that won't fix the problem though.

[Mordil]

@jstorm31 Pooling is done under the hood in Vapor 4.

Each EventLoop is given an RedisConnectionPool, so if your configuration says to populate a pool with 4 connections, and you have 4 EventLoops, then you will have 16 total connections.

Try modifying the configuration with app.redis.configuration.pool.

In the next week or so, documentation will improve as Vapor/Redis moves to full 4.0 release

[jstorm31]

I see, thank you 🙂