I have an ESP32 monitoring my web connectivity by pinging a remote site.
Its supplied from a powered USB hub.
Last night we were expecting a thunderstorm so I removed the mains supply (so the equipment was isolated.)
This morning when I reconnected the equipment I noticed the board had not restarted; (on board power LED on but no other activity) but after a reset via the button on the board all is working again.
As the connected PC wasnt turned on I have no knowledge of the serial output, but after I turned it on there were no messages until I reset it.
The web page was not being displayed, and the browser said the page was unavailable so likely the MDNS also not working.
The sketch is too long to post, but the code following the setup() I feel isnt likely to be relevant anyway, so here is the setup() code.
void setup() {
Serial.begin(BAUDRATE);
pinMode(trigger, INPUT_PULLUP);
delay(5000); //allow to start & clear serial monitor
Serial.println();
Serial.println(__FILE__); //print name of sketch for easy identification
pingSamples = 60.0 / timeInterval;
initIO(); //initialise hardware IO devices all leds off
flashAll(5, 200); //test leds
Serial.println("Serial comms established");
attachInterrupt(digitalPinToInterrupt(trigger), ISR, RISING);
if (!LittleFS.begin(FORMAT_LITTLEFS_IF_FAILED)) { //start LittleFS system
Serial.println("LittleFS Mount Failed");
return;
}
getCreds(); //in wifi tab; recover credentials that were saved in the int response
connectWiFi(); //also gets time from time server ntp
buildStrings(); //in webData.ino - creates local copies of time server values int YYYY, MM, DD, hh, mm, ss;
pRecoverStats(); //retrieve stats (preferences) after reset power fail
prepareStats(); //put them into strings for web page
// Initialize mDNS
if (!MDNS.begin(mDNSname)) { // Set the hostname
Serial.println("Error setting up MDNS responder!");
}
Serial.println("mDNS responder started");
manageServer(); //handle web page requests
addToFile(LittleFS, "/connData.csv", messRestart); //save a line to show unexpected reset has occurred
fileHead(); //repeat heading after restart event
//dont allow newMin or newH to run as they reset counts recovered from pRecoverStats();
currMin = mm;
currHour = hh;
currDay = DD;
currMonth = MM;
firstHour = 1; //following restart no valid hourly data to display
timeWas = millis(); //used for timing pings in loop
wpUpdate = millis(); //times the web page update
}
Normally when I remove and reinstate the power it resets as we would expect.
I don't actually know what happens when power is applied, does it ALWAYS start at a specific point in the code? What code, the user uploaded code, or the 'OS'. I have never seen a diagram of what happens but I have no difficulty believing the OP and therefore that it is NOT the first setup instruction.
"If a tree falls in the forest and no-one is there does it make a sound?"
The project is for continuous monitoring and recording (amongst other things) a power fail. I cant watch the LED all the time. Setup() includes a sequence to flash the LEDs .
I did a search and found this https://arduino.stackexchange.com/questions/25448/bootloader-sequence
so if its right the "setup()" and "loop()" live inside a control program.
If my understanding is correct the removal and reinstatemnet of the supply should have the same effect as a reset. So why has it not in this case?
int main(void) {
init();
setup();
for (;;) {
loop();
if (serialEventRun) serialEventRun();
}
return 0;
}
This is going to be very difficult to trouble shoot if you can't simulate the conditions which caused this failed startup by switching off power in a controlled way, to various devices, and later restoring it.
Such a test would include switching off the Internet router to ensure the ESP32 can recover from a dropout of the WLAN either at power up of the ESP32 or during its normal operation.
I suppose you could try an uninteruptable power supply and wait for the next thunderstorm to see if it works.
You could also have a task which checks for a continuous loss of connectivity and forces a clean restart periodically.
In the OP you described that the board had not restarted, after the power had been reconnected presumably, and that you had to go and reset the board. Thus at that point if a LED on the board was on you would know if the sketch had actually started to run after power was restored.
Mains supply to the house?
Did you wait until the router had restarted, before connecting the ESP.
I use WiFiManager with a portal and a 120sec timeout for that reason.
Leo..
The on-board lights showed power was on, but the serial monitor showed the board was not responding.
Hadnt thought of that. Its just on a breadboard with no connections other then what are shown in the schematic ; possibly static? As I said it normally restarts from a power down (however that occurs) with no issues.
Have you tried just switching off the internet router for a few minutes and then switching it back on again? It may be that the ESP gets stuck in a loop while trying but failing to restore the connection giving the impression that it [the ESP] did not start correctly.
