Hello.
There is something to be said about the memory, tcp established metrics that arise while using MQTT.
I am using rabbitmq 3.8.14 installed on ubuntu16 version in a single vmware.
I am using mqtt, management plugin and follow the default settings.
I am not using it as a cluster.
An average of 1400 IoT devices are communicating based on mqtt
Each network is an internal network, but the device's IP is different.
There are two problems we are currently facing.
1. Increasing issue of tcp established
Currently, tcp established is constantly increasing.
Currently mqtt ttl is the default setting (24h)
Checking with netstat or management tends to increase by about 100 per day.
(attached a file : check file description)
Currently, it is assumed that the connection with a specific IoT device has been esatablished, and the connection has increased because the client (iot device) did not give a FIN.
2. Continuous memory increase issue
Attach the results of the current rabbitmq-diagnostics.
In the case of allocated_unsued, there is a trend of steadily increasing, which is
Could it be possible to cause problems with memory management in the future?
Can someone with similar experience related to me give you advice?
Thank you for helping us use excellent open source.