When high concurrent requests may cause connection leak

[semistone]

Redis version

7.0.11

Redisson version

3.30.0

Redisson configuration

    Config config = new Config();
    config.setCodec(new StringCodec());
    config.setNettyThreads(64);
    config.useSingleServer().setAddress("redis://127.0.0.1:6379").setTimeout(5000).setConnectionMinimumIdleSize(1)
            .setConnectTimeout(500)
            .setPingConnectionInterval(100)
            .setKeepAlive(true)
            .setRetryAttempts(3)
            .setIdleConnectionTimeout(100)
            .setConnectionPoolSize(1);

What is the Expected behavior?

do 1000 set/get request parallel
it should success or failure but
connection will recover after some time later

increase connection pool size will fix this issue, but why connection broken.
and I guess that's the reason why we see timeout error in our production server.

What is the Actual behavior?

connection never recover

Additional information

I had write a reproduce procedure in
https://github.com/semistone/redission-issue/

[semistone]

I found in current code,
when high load, queue is always >=0
if current code,
it will get Semaphore -> handle request1 -> get Semaphore -> handle request 2 ....
until it goes too deep
and at some point, that thread seem just disappear. ( I found the debug counter disappear in strange place )

I change to switch thread per 100 rquests
#5977

and that connection leak seems been fixed in my test code
not sure is it correct?
@mrniko