logging details for connection pool reactor-netty

[gurudatta11]

1. logging as warn, if connection pool reached maximum limit might be helpful.
2. And also other loggers to indicate how many connections are being used at any given point of time.

Motivation

Currently it's difficult to debug if there is issue with connection pool in reactor-netty. Hence more loggers and monitoring might help when application is scaled.

Desired solution

• logging as warn, if connection pool reached maximum limit will be helpful and also advise on how to address this (adjust configuration or etc.,)
• And also other loggers to indicate how many connections are being used at any given point of time.
• During startup of application, log what kind of connection pool being used with the min and max limits.
• Also if possible, Monitoring endpoints (make it integrable) Spring Boot Actuator: Health check, Auditing, Metrics gathering and Monitoring

Considered alternatives

Additional context

Creating this feature request based on conversation in #907

[violetagg]

If you configured Micrometer in your Spring Boot application you should be able to see
reactor.netty.connection.provider.<name>.total.connection
reactor.netty.connection.provider.<name>.active.connection
reactor.netty.connection.provider.<name>.idle.connection
reactor.netty.connection.provider.<name>.pending.connection

Where <name> by default is http otherwise the name that you provided when creating the ConnectionProvider

About the logs - if you enable the log level for Reactor Netty to DEBUG you should be able to see:

21:18:22.388 [Test worker] DEBUG r.n.r.PooledConnectionProvider - Creating new client pool [test] for /0:0:0:0:0:0:0:1:60172
21:18:22.434 [reactor-http-nio-2] DEBUG r.n.r.PooledConnectionProvider - [id: 0x50025567] Created new pooled channel, now 1 active connections and 0 inactive connections
21:18:22.517 [reactor-http-nio-2] DEBUG r.n.r.PooledConnectionProvider - [id: 0x50025567, L:/0:0:0:0:0:0:0:1:60173 - R:/0:0:0:0:0:0:0:1:60172] Channel connected, now 1 active connections and 0 inactive connections
21:18:22.645 [reactor-http-nio-2] DEBUG r.n.r.PooledConnectionProvider - [id: 0x50025567, L:/0:0:0:0:0:0:0:1:60173 - R:/0:0:0:0:0:0:0:1:60172] Releasing channel
21:18:22.648 [reactor-http-nio-2] DEBUG r.n.r.PooledConnectionProvider - [id: 0x50025567, L:/0:0:0:0:0:0:0:1:60173 ! R:/0:0:0:0:0:0:0:1:60172] Channel closed, now 0 active connections and 0 inactive connections
21:18:22.650 [reactor-http-nio-2] DEBUG r.n.r.PooledConnectionProvider - [id: 0x50025567, L:/0:0:0:0:0:0:0:1:60173 ! R:/0:0:0:0:0:0:0:1:60172] Channel cleaned, now 0 active connections and 0 inactive connections

[violetagg]

In addition to the comment above

The log was extended

16:31:31.511 [Test worker] DEBUG r.n.r.PooledConnectionProvider - Creating a new fixed client pool with name [test] and max connections [1] for [/0:0:0:0:0:0:0:1:59978]

logging as warn, if connection pool reached maximum limit will be helpful and also advise on how to address this (adjust configuration or etc.,)

The fact that the pool reached the max connections limit still cannot be handled as a warning.

Having TimeoutException with a message Pool#acquire(Duration) has been pending for more than the configured timeout of 20ms is the point when a warning should be triggered.

In addition to the metrics mentioned above we will expose also an API #925