Just a follow up on this.
I have multiple WiFi networks around the house for different purposes. Including a couple of extenders to get to remote parts of the house.
The Evohome has been connected to one of the original ones in the middle of the house since the install about 18 months ago. and this had iffy signal qualities. Hence all the new Wifi and extenders.
While looking for the reason behind this I noticed this old connection in the controller's WiFi setup and, therefore, connected it to the nearest WiFi hub which had a 100% strength signal. Just for completeness I also rebooted the main Draytek Internet Router/AP to clear all it's internal tables.
Lo and behold the "problem" is gone. After days of running I no longer see multiple (let alone hundreds) of socket connections linked to the Evohome Controller IP.
This leads me to believe that there is a bug in the controller's software related to timeouts of the TCP connection which it thinks are dropped but in fact are being managed by TCP flow/reconnect management. The TCP socket remains "active" but the software thinks it's an error and creates another connection to the Server without fully terminating/destroying the original socket (been there with that problem in my own software life). Theory maybe, but fits the facts.
So "problem" closed, but something for the Evohome Controller software guys to read.
I have multiple WiFi networks around the house for different purposes. Including a couple of extenders to get to remote parts of the house.
The Evohome has been connected to one of the original ones in the middle of the house since the install about 18 months ago. and this had iffy signal qualities. Hence all the new Wifi and extenders.
While looking for the reason behind this I noticed this old connection in the controller's WiFi setup and, therefore, connected it to the nearest WiFi hub which had a 100% strength signal. Just for completeness I also rebooted the main Draytek Internet Router/AP to clear all it's internal tables.
Lo and behold the "problem" is gone. After days of running I no longer see multiple (let alone hundreds) of socket connections linked to the Evohome Controller IP.
This leads me to believe that there is a bug in the controller's software related to timeouts of the TCP connection which it thinks are dropped but in fact are being managed by TCP flow/reconnect management. The TCP socket remains "active" but the software thinks it's an error and creates another connection to the Server without fully terminating/destroying the original socket (been there with that problem in my own software life). Theory maybe, but fits the facts.
So "problem" closed, but something for the Evohome Controller software guys to read.
Comment