I’ve got an annoying situation where an ethernet connected MQTT client is disconnecting at random. The node didn’t do it at all the first day I installed it, the second day it was hopping on and off the broker either every couple of minutes to up to circa 1 hour or so intervals.
I went in on day three and replaced the network switch it was connected to and that seemed to do the trick as it settled right down after that to maybe one reconnection per day at most. Then today I had cause to restart the main router for the building. Since then the bloody client has been back at it, reconnecting every few minutes. So annoying!
It’s clearly not the code as obviously it can go days very reliably. Overnight I’ve never known it disconnect. On bad days I tend to get the first disconnect round around 8am and the last one around 6pm. So I think it must somehow be linked to network traffic (?) even though it’s been disconnecting like crazy this afternoon following the router reboot and there’s no extra ‘load’ on the network. I’m just wondering
- if I upgrade the Eclipse HA add-on, might I get new options to set a different log level for the broker that might help me diagnose WHY a client is publishing its LWT or reconnecting? There doesn’t seem to be an option to change the log level with v2.
At the moment I’m still running the ancient version 2 because I do not want any of the ACL, forced logins and security hysteria that seems to come with the later versions of the add-on. I know version 2 is inherently stable because I run it on another HA instance and it’s rock solid there so somehow I doubt upgrading to 5.1 would actually help but still. - Anyone have any idea what I might be able to do to diagnose what’s going on? So confused. I know wireshark is a potential tool…but it seems like potentially overkill for this problem and I really don’t know what I’m doing with wireshark.
Any help much appreciated