Hi folks,
Just want to check this isn't something I've done wrong here.
Currently the wifi is set as follows
wifi:
id: ota_connectivity
ssid: !secret wifi_ssid
password: !secret wifi_password
enable_on_boot: False
reboot_timeout: 0s
min_auth_mode: WPA3
ESPNow is also present, and does initialise on boot.
espnow:
id: espnow_out
peers:
- ${peer_mac}
channel: 1
enable_on_boot: True
(note, this is a file that runs on two identical devices, hence the substitution).
This leads into packet_transport which then manages sending all the stuff out (standard setup).
We have a script that triggers when the receiver over packet transport sends a value back on a specific sensor (linked to a template switch on the reciever).
- id: ota_prepare
mode: single
then:
- if:
condition:
- lambda: |-
return cycles_rtc == 1 && is_cold_boot;
then:
- script.stop: ota_prepare
- deep_sleep.prevent: deep_sleep_control
- delay: 1000ms
- component.suspend: packet_link
- wifi.enable
- delay: 500ms
- wait_until:
condition:
- lambda: |-
return id(ota_connectivity).is_connected();
- delay: 1000ms
- if:
condition:
- lambda: |-
return id(ota_requested).has_state() && id(ota_requested).state > 0.5;
then:
- ota.http_request.flash:
md5_url: http://192.168.1.66/${device_name}.ota.bin.md5
url: http://192.168.1.66/${device_name}.ota.bin
Prior to 2026.6.0, this worked absolutely fine.
But now:
[13:00:12][V][wifi_esp32:783]: Connected ssid='IoT' bssid=[MAC] channel=1, authmode=WPA3 PSK
[13:00:14]Guru Meditation Error: Core 0 panic'ed (InstrFetchProhibited). Exception was unhandled.
[13:00:14]
[13:00:14]Core 0 register dump:
[13:00:14]PC : 0x00000000 PS : 0x00060a30 A0 : 0x800f30c1 A1 : 0x3ffcd150
[13:00:14]A2 : 0x3ffd42cc A3 : 0x3ffd4604 A4 : 0x0000002a A5 : 0x3ffd590c
[13:00:14]A6 : 0x3ffb2734 A7 : 0x3ffd4fa8 A8 : 0x8015979c A9 : 0x3ffcd130
[13:00:14]A10 : 0x3ffd4fa8 A11 : 0x3ffd42cc A12 : 0x3ffd4fa8 A13 : 0x3ffd590c
[13:00:14]A14 : 0x3ffcd2fc A15 : 0x3ffd4604 SAR : 0x0000001b EXCCAUSE: 0x00000014
[13:00:14]EXCVADDR: 0x00000000 LBEG : 0x4000c2e0 LEND : 0x4000c2f6 LCOUNT : 0xffffffff
[13:00:14]
[13:00:14]
[13:00:14]Backtrace: 0xfffffffd:0x3ffcd150 0x400f30be:0x3ffcd170 0x40088a85:0x3ffcd1a0 0x4008ca73:0x3ffcd1c0 0x4008cb29:0x3ffcd220 0x40091221:0x3ffcd240 0x4008e0df:0x3ffcd260
We do also have a single .h file that just sets RTC variables to hold in memory across deep sleeps. This device wakes up to just throw out humidity/temperature and then goes back to sleep for a defined amount of time unless the OTA switch is flicked on the receiver. These variables are just the previous humidity/temperature so it can calculate whether it needs to send out (ie. if it's changed too much) and a few other little bits. Nothing that relates to these scripts/the wifi.
My thought here is a bug. Possibly linked the wifi driver no longer taking up RAM if enable_on_boot is set to false?
IDF version is set to recommended. Version is 2026.6.2 (at present, flashed via USB). Issue persisted across all the 2026.6.x versions so far. Device is a classic non-variant ESP32 (Firebeetle 2 ESP32-E).
Is there a simple solution I'm not seeing? I would put this on the issues on ESPHome's GitHub, but never done that before, so don't want to throw something on there that is something stupid I'm missing.
Thanks 