Memory usage did not decrease when disconnected

[ytxfate]

I wrote a socket.io template，client connect and join room, then start a background task to get data from redis channel and emit data to room. When the client is connected, the memory slowly grows and there is no sign of stopping, but after a period of disconnection, the memory does not decrease. I did not find where the memory is consumed. Here is my code: https://github.com/ytxfate/socketio_template.git.
The Python version is 3.9.7 and The Python env is as follows:

bidict==0.21.4
certifi==2021.10.8
dnspython==2.1.0
eventlet==0.32.0
greenlet==1.1.2
loguru==0.5.3
python-engineio==4.3.0
python-socketio==5.4.1
redis==3.5.3
six==1.16.0

The redis hash:

> HGETALL socketio_server_v1.0:proid_redchan_mapping.hash
ceshi01
["ceshi_room01","ceshi_room04"]

thanks a lot

[miguelgrinberg]

Use tracemalloc to see what memory your application holds on to. https://docs.python.org/3/library/tracemalloc.html

[schnabel]

How did you monitor memory? I am asking because top (linux) might not show freed memory.

[ytxfate]

pidstat -r -p {PID} 10

[xxthegonzxx]

@ytxfate did you ever figure this issue out? We're experiencing the same high memory usage even after disconnecting clients.

[ytxfate]

@xxthegonzxx Not solved

[xxthegonzxx]

@ytxfate I ran a memory profiler for my application using this setup:

flask-socketio = "^5"
eventlet = "^0.33"
greenlet = "^2"
flask-redis = "^0.4.0"

It looks like my profiler was stating that the issue might be due to eventlet library and dereferenced objects not getting cleaned up by the Garbage Collector. We're still working on a solution but feel like we're getting closer. Good luck!