New router - HA works, but ESPHome wifi devices do not

ESPHome
1

Hi.

I just got a new Zyxel router to replace my old Zyxel router.
Nothing fancy - just the typical ones you get from your internet provider.

When the new router is connected (and all router settings are changed to match the old one), all laptops, tablets, smartphones and desktops (wired and wifi) works fine.
My Home Assistant Raspberry Pi also connects fine (wired) and I can access my Home Assistant from both within the LAN and from the outside world using e.g. the HA app on a smartphone and also using a normal browser.

However, all my ESPHome wifi devices will suddenly not talk to Home Assistant any more. This means no sensor updates, no wifi logs, no OTA updates.
What’s curious is that I can see from my router that the devices (e.g. some ESPHome smartplugs) are connecting to my wifi and get an IP. And I can “visit” their IPs and see their sensor data. They just won’t talk to Home Assistant where they appear offline.
I have problems with both my ESPHome devices with fixed IP’s and those that do not hve fixed IP’s. But to keep it simple, I only ask for help regarding those that do not have fixed IP’s as I can see for sure that they actually do connect to my wifi.

When trying to delete one of the ESPHome devices Home Assistant rediscovers it. But if I then attempt to reconfigure it (after the message: Do you want to add the ESPHome node Athom SmartPlug 2 to Home Assistant?), it fails with the error message: Can’t connect to ESP. Please make sure your YAML file contains an ‘api:’ line.
And for the record, I do have an API line in my yaml file (with a password).

substitutions:
  name: athomsmartplug2
  friendly_name: Athom SmartPlug 2
packages:
  Athom_Technology.Smart_Plug_V2: github://athom-tech/athom-configs/athom-smart-plug-v2.yaml
esphome:
  name: ${name}
  name_add_mac_suffix: false
  friendly_name: ${friendly_name}
api:
  encryption:
    key: 90123456789012345678901234567890123456789012
wifi:
  ssid: !secret wifi_ssid
  password: !secret wifi_password

And all this yaml code works fine with the device as long as it is connected to the old router.
I can switch back to my old router and all my devices connects nicely to Home Assistant again. I.e. I can make the problem disappear and reappear by switching between routers.

Anybody got an idea before I have to send the new router back and keep the old one, which is already several years old?

Best regards,

Thomas.

2

Important point perhaps, the ESPHome devices do not connect to HA, HA connects to the ESPHome device.
Not sure about the rest, your Home Assistant perhaps has them configured with a different IP, but after deletion you should be able to add them.
Did you try to add them to HA using the GUI (add new ESPhome device, enter hostname or IP, enter api key when asked for) instead of using discovery?

Armin

3

Thanks for the input Armin.
I was not aware of the detail about Home Assistant connecting to the ESPHome device and not the other way around.
I found out that I can also ping my devices with fixed IPs, so they have indeed connected to the wifi.
And the smart plugs without fixed IP’s I can visit with a browser and see that they are alive and well.
So my ESPHome devices are either ignoring Home Assistant’s attempts to talk to them (maybe because of API password issues?) or Home Assistant is not trying to contact them at all for some strange reason.

I have now tried deleting them and adding them using the GUI (Add device) instead of using the auto-discovery function.
I get the same error message: Can’t connect to ESP. Please make sure your YAML file contains an ‘api:’ line.

So I never even get to the part where I have to enter an API key.

4

Could you post the IP addresses of your HA setup as well as a few of the ESPHome devices? If for some reason you set a bunch of your devices to use a static IPs and some of your other devices to get IPs via DHCP from your router, it’s possible that there is now a mismatch between either the subnet or the net mask. That could cause a communication issue between HA and the ESPHome devices. I know you said all the settings on your new router were the same, so I might be grasping at straws, but I also don’t want to miss something obvious on the network side.

5

Hi Kyle.

I am very open to the possibility of some network clash as I am no network engineer and have only basic router experience.

My Home Assistant Raspberry Pi has the fixed IP: 192.168.1.30
Router subnet mask: 255.255.0.0
Router IP address range: 192.168.0.1 ~ 192.168.255.254
My ESPHome devices with static IPs have the following IP’s (although only a few of them are on at the moment):
192.168.2.x (x = 1-10)
192.168.3.x (x = 1-10)
etc.

My non-static IP ESPHome devices currently have the IPs:
192.168.22.197
192.168.25.7
192.168.25.139

Do you see any potential problems here?

Br, Thomas.

6

Additional information:
I can get logs from my ESPHome devices when connecting one of them directly to my laptop using a serial USB cable.

The devices seems to work perfectly, but no Home Assistant API connection is ever made, which I guess is at the root cause of my problem.

Relevant log output from ESPHome device:

[C][api:025]: Setting up Home Assistant API server...
[D][main:310]: on_boot - Before intial delay
...
[C][api:139]: API Server:
[C][api:140]: Address: 192.168.2.8:6053
[C][api:142]: Using noise encryption: YES
...
[D][main:415]: Skipping sleep cycle - no API connection

The last log entry repeats forever as no API connection is ever made.

I can reflash the ESPHome devices using the serial cable, but this changes nothing.

7

The last log entry where it is skipping sleep cycle due to no API connection is my own code where I check the condition of api.connected and then refuse a sleep cycle if no API connection has been made.
This code is working fine and in this case it provides a handy log feedback regarding the API connection.

Of course the absence of a successful API connection in the log gives the same information :slight_smile:

8

My limited knowledge says 192.168.x.x should be a class C network not class B
your subnet mask should be 255.255.255.0
and only the last segment for the ip addresses should be different.
But I actaully thought this was more of a genaral rule then stuff not working.

maybe one component has this hardcoded?
what happends if you move everything to a class C network?

9

The private IP space is classless for the 192 network. So 192.168.0.0 is part of class C but the subnet is /16 for the designated private space, so they should be fine.

@Thomas11 I know the ESPhome Dashboard add-on uses mDNS by default to try to connect to the devices. It uses the friendly name appended with .local to reach out (e.g. friendly name of esp-device would be mDNS of esp-device.local).

Can you reach the web pages using the mDNS name of any of the devices?

1 Like
10

A /16-network in the 192.168.x.x range… ok … unusual, but should work.
Usually you would create the private /16-network in 172.y.x.x
Did you check the netmask in your Home-Assistant ? Is it 255.255.0.0 like on the router? Same for the ESPHome devices with static address?

11

Hi Ryan.

Forgive my ignorance regarding mDNS.

In my example I have an ESPHome device with the

device name = “esp32mini8”

and which has the

friendly name: “ESP32 Mini 8”.

It has the IP: 192.168.2.8

I can ping it with the following result:

ping 192.168.2.8
Reply from 192.168.2.8: bytes=32 time=3ms TTL=255
Reply from 192.168.2.8: bytes=32 time=5ms TTL=255
Reply from 192.168.2.8: bytes=32 time=12ms TTL=255
Reply from 192.168.2.8: bytes=32 time=5ms TTL=255

I am unsure as to how I can ping it using mDNS.
You suggested friendly name with .local appended, but won’t that make the command: ping ESP32 Mini 8.local
This obviously gives an error.

When I instead try to write: ping esp32mini8.local
I get the following error:
Ping request could not find host esp32mini8.local. Please check the name and try again.

Was that what you wanted me to try?

I can access the homepage of my smart plugs using their IP addresses (e.g. 192.168.22.197).
I cannot however reach the webpage of my “ordinary” esp32 modules by just typing their IP addresses in a browser (like 192.168.2.8 or 192.168.2.8:6053 or 192.168.2.8:50), but I do know that they are on the wifi network.
I had even added the following to the yaml configuration, but it did not help.

web_server:
  port: 80
12

Hi Armin.

I have not (recently) changed the netmask in Home Assistant.
I only recently changed it in the router and the ESPHome devices’ yaml files.
I was not aware of this having to also be changed in Home Assistant.
That being said, everything works perfectly with my old router. I am just not sure why :slight_smile:

If I go into Home Assistant and look at the network settings under Configure network interfaces it lists the IP address/Netmask as: 192.168.1.30/24

I am a network novice, but I am guessing that /24 may not correspond to 255.255.0.0 ?

Dinner is calling, but I will research this afterwards.

Br, Thomas.

13

/24 is 255.255.255.0
/16 is 255.255.0.0
/8 is 255.0.0.0

With an IP of 192.168.1.30 and a /24 netmask, a ESPHome-Device with IP 192.168.2.8 is considered to be in a different network

14

Yes. So that tells me mDNS isn’t working properly.

If your ordinary devices didn’t already have the web_server enabled, then that’s not going to help. Adding the config to the YAML doesn’t change anything on the device. For that particular setting, you’d have to add the config and then do an OTA update to have it added to the firmware that’s running on the device.

I would start by fixing this:

since you definitely want to make sure the IP config in HA matches what you are doing at the router level (in theory though, this should still work).

If that doesn’t fix things, then I would focus on a single device at first.

Edit the YAML again and add the use_address option under the WiFi section. This will tell the ESPhome Dashboard to attempt to connect to the device by IP instead of mDNS.

So your wifi section would look like:

wifi:
  ssid: !secret wifi_ssid
  password: !secret wifi_password
  use_address: 192.168.2.8

Save and see if the ESPhome dashboard can connect to it then.

15

Hi Ryan, Armin, Kyle and bkbartk.

TLDR: I got it working again - thanks a lot for all your help :slight_smile:

Long version:
It was my Home Assistant network setting Configure network interfaces where the IP address/Netmask was: 192.168.1.30/24
I changed it to 192.168.1.30/16 after which everything connected like a charm.
I had never touched the /24 part, which apparently is the same as a subnet mask of 255.255.255.0 meaning that only the last byte/octet can be used by individual devices - or at least that Home Assistant will not talk to anybody beyond that range.

Recently (a month ago) I decided to put my ESPHome devices on what I thought was some well planned IP ranges as shown here:

192.168.2.x all my esp32 minis
192.168.3.x all my esp32 devboards
192.168.4.x all my esp32-cams
192.168.5.x all my esp32 smartplugs
192.168.6.x all my esp8266 D1 minis
etc.

I knew I had to update the relevant yamls for the ESPHome devices as well as the router settings and this worked very well with my old router.
All ESPHome devices had originally been flashed and given a DHCP IP of 192.168.1.x, which I then changed to statics IPs.
They all connected nicely to Home Assistant with their new static IPs and I could make nice dashboards with all the sensor data.
Being a network novice, I had no idea I also had to change the /24 setting in Home Assistant as well and apparently it worked with the old router without changing that setting.
That shouldn’t have worked according to my new understanding, but it did.
Maybe the fact that they were accepted into Home Assistant when they still had the old IPs that were allowed according to the /24 setting was the reason that they were not thrown out of HA even though their IPs later changed beyond the /24 scope?

Anyway, my problem arrived when I took a new router into use even though it had the same settings as the old one.
And the problem went away as soon as the /24 setting was changed to /16.

I was a bit baffled about not getting a response from ping esp32mini8.local when I did get a response when pinging its IP directly.
But obviously this was because I was pinging from my Windows PC prompt and not from the Raspberry Pi Home Assistant terminal.
When I finally realized this, I tried ping esp32mini8.local from the HA terminal and got a response :slight_smile:

I can still get to my smart plugs web pages just by typing their IPs in a browser.
I am not, however, able to do that though for my ordinary ESP32 devices in spite of adding the web_server: port: 80 lines to their yaml (and of course flashing the devices).
But this is not so important for me right now.
I guess it requires more knowledge about the ESPHome web server component, CSS or other stuff in order to setup a web page that shows the sensor data.
I just expected that a web page would be created automatically on ESPHome devices.

Anyway - Thanks a lot for all the help.
You saved the day :slight_smile:

Br,

Thomas.

1 Like
16

Glad you got it sorted.

17

How do you make that change? I found the network settings, but cannot see a way to edit them …

18

Hi.

If you are looking for where to change the network settings in Home Assistant, then look in the address: http://[YOUR IP]:8123/config/network

You can get to this manually by chosing the menus: Settings then System then Network

Expand the IPv4 menu to see IP address/Netmask (only visible if the Static radio button is chosen).

Hope this helps.

Thomas.

1 Like
19

Yes, that was the bit I did not spot. I had got to the right page but did not figure out the combination of ‘expand’ and ‘select static’.

I successfully changed /24 to /16. Unfortunately, it did not solve my problem, so perhaps mine has a different cause. I will post it as a separate issue shortly.

20

Both devices need the same netmask.

1 Like