No, … the issues I’m talking about are from a while ago. Basically the connection needs a keep alive. It needs fresh mdns update, if you will. It seems it’s bad at keeping the mdns records. They seem to expire to fast.
I think the issue is with esphome. only emporia has like 60 entities while most have like 6 and that’s probably why I do notice it in emporia and not in the other devices. Sometimes when i open esphome. It says that the device is offline. but when i click it I can connect to it. As if there isn’t anything wrong with it.
Posted you both log and capture log in a private message, then you can see on what time it happens and see through the capture log. Filtered on mac address so it should include all traffic related to that client.
I don’t think anyone should post pcap files on a public forum. Hope that’s ok.
Packet 13172 is the first packet where you can see the error after that it happens. Although my clocks are synced it seems that the time in the packets is about 3 seconds off ? I checked my computer, homeassistant server and the accesspoint all have time sync’ed and yet 3 second delay in esphome?
It’s not reconnecting but seems there are some malformed packets not sure why that is.
Installed logviewer and I noticed that this also happens on my other clients. Only saw it once on one other client last 15mins and like every minute on the emporia.
EDIT: I do get the same error on all clients if i restart homeassistant.
Lowered the update_interval to 5s and …
now each time it saves preferences to flash i get the disconnect
[15:17:21][D][esp32.preferences:114]: Saving preferences to flash...
INFO 10.13.x.x: Error while reading incoming messages: Error while reading data: [Errno 104] Connection reset by peer
INFO Disconnected from ESPHome API for 10.13.x.x
WARNING Disconnected from API
INFO 10.13.x.x: Ping Failed: Error while reading data: [Errno 104] Connection reset by peer
INFO Successfully connected to 10.13.x.x
EDIT: never mind was maybe just coincidence … changed flash write to 3600s and it still has connection error only now the flash write message is gone.
Changed update intervals on the main to 3s and the other sensors to 5s, in the beginning it timed out like every 10-20s, not sure if that’s the case after a while it settled also the flow of information is becoming more continuous and now it has been going for the last 2 mins without disconnecting.
I think the issue is that the chip can’t handle the load. Seems like it’s still disconnecting but not as often.
For some reason my notifications didn’t show this thread being updated. Saw the pcap, will summarize my thoughts here as well. Definitely share privately.
Are you saying you are exporting AP syslog?
Disconnects shouldn’t really happen on any clients but likely aligns w/ the observations (below).
Not sure what chip you are referring to.
[Recap of DM observations]
Below error graph from the provided pcap from the AP is showing significant level of error (dup ack’s).
Given the level of duplicate ack’s noted by the AP, and other clients disconnecting, it really suggests wider network-level problems than an ESPHome or device configuration issue. Even if the Emporia is more sensitive than other clients.
Hopefully AP syslog events might help here to provide further clues.
I just done a capture again it being connected to a different AP same thing. Then I took a look at other clients in the network also connected to home assistant. No errors it is specific to this client. The errors are client specific.
Both are one the same accesspoint and both are connected to esphome.
The left one is of course emporia, Could it be that the powercables are interfering with the wifi? I don’t think it’s network releated but has something to do with the chip, it’s client side related.
Other clients are not disconnecting it used to be few months ago. Sometimes this is because of unifi they suck when it comes to firmware. Sometimes it works really well and other times it’s like they had a contest how can we make it so bad it barely works. And sometimes it’s because of esphome.
Ive sorted on wifi experiance, … 2 clients out of like 50 clients have a bit bad experiance.
well emporia is a power meter located in the power cabinet, the enclosure isn’t from metal however it’s close to the power cables. But that’s the case for all emporia installations.
If anyone who works on ESPhome wants the capture logs feel free to ask. But the problem definitely lies with ESPhome a TCP connection does not work just for one service. It’s either bad or a good one.
Hell didn’t even screw in the antenna. It might be the clients libraries I don’t know but something is going on. And basically both the server and the client is compiled in ESPhome, so no matter how you twist and turn. The problem lies with ESPhome.
EDIT: Does anyone know where I can submit a bug report? For ESPhome?
Question… I have a set of both the originial mains ct clamps and the flexible upgraded ones since my main rail is flat metal and not round wires. Could I use the 2 larger ct clamps on one of the side inputs so i could monitor my generator when i need to switch over?
