Connection not released when request is cancelled

Bug Report

Versions

• Driver: Driver: 0.8.8.RELEASE
• Database: Posgresql 12.7
• Java: 11
• OS: Windows

Current Behavior

This issue was first opened on r2dbc-postgresql, some discussion can be seen there https://github.com/pgjdbc/r2dbc-postgresql/issues/459

Then it was opened here: https://github.com/r2dbc/r2dbc-pool/issues/140

The result was a fix on the pool but it does not fix the problem.

We have created a new project without spring dependencies and the connection keeps getting stucked.

You can find the source code here: https://github.com/jfrossetto/r2dbc-stress-cancel

pid	datname	usename	application_name	backend_start	query_start	query
3065	postgres	postgres	sample	2021-11-01 10:02:41.308	2021-11-01 10:02:45.159	SELECT 1
3064	postgres	postgres	sample	2021-11-01 10:02:41.307	2021-11-01 10:02:46.358	SELECT 1
3066	postgres	postgres	sample	2021-11-01 10:02:45.640	2021-11-01 10:02:50.739	SELECT 1

LOGFILE: https://github.com/jfrossetto/r2dbc-stress-cancel/blob/master/log_stresscancel.txt

There was another issue opened with the same problem here: https://github.com/r2dbc/r2dbc-pool/issues/143

Stack trace

// your stack trace here

Table schema

Input Code

-- your SQL here;

Steps to reproduce

Input Code

// your code here;

Expected behavior/code

Possible Solution

Additional context

[mp911de]

I can confirm that upgrading to Reactor Core 3.4.14 snapshots fixes the issue.

[Ernir]

Commenting for increased visibility - looks like the reactor-core team has a potential solution, see https://github.com/reactor/reactor-core/pull/2859 .

If you can reproduce the issue locally, confirmation that the fix works is desired. See https://github.com/reactor/reactor-core/issues/2836#issuecomment-999797297 .