Can't seem to successfully restart after resetSettings()

[cmarrin]

Basic Infos

Hardware

WiFimanager Branch: Development

Esp8266

Hardware: WeMos D1 Menu

Core Version: tip of tree

Description

I'm building a clock with a WeMos D1 Mini. Everything works. The system comes up, network connects (credentials are already there), it fetches data from Weather Underground, snags the current time and displays it.

If I use the hack of calling resetSettings before calling autoConnect, that works too. I connect to the AP from my iPhone, it brings up the web page, I tap on my SSID, enter the password and hit save. The system starts back up and all is well.

But now I've added a menu which allows me to reset the network and start over. It dutifully brings up the config portal, I enter my credentials and hit save. But this time, it says "*WM: 5000 ms connectTimeout set, waiting for connect...." at which point it outputs dots for 5 seconds and then starts the config portal again. If I reset the ESP at this point it boots back up, connects to the network and everything works.

I've tried lots of things. It seems like WifiManager is getting itself into a funny state that it can't get out of. Any ideas?

Settings in IDE

Module: NodeMcu, Wemos D1

Additional libraries:

Sketch

https://github.com/cmarrin/Clocks/blob/master/OfficeClock/OfficeClock.ino

Libraries

My own

https://github.com/cmarrin/m8r

with these includes

include

include <m8r/Blinker.h>

include <m8r/BrightnessManager.h>

include <m8r/ButtonManager.h>

include <m8r/Max7219Display.h>

include <m8r/StateMachine.h>

include <m8r/WUnderground.h>

And these libraries installed from the LibraryManager

include

include

include

include

include

include

include

WiFiManager requires PR https://github.com/tzapu/WiFiManager/pull/584

[tablatronix]

I will look, are you setting a connecttimeout to 5s ?

[cmarrin]

Doesn't look like there's an option for that. After the ConfigPortal succeeds, it uses _saveTimeout and I don't see any way to modify that with the public API. I did try changing the value to 10 seconds and that didn't help.

But I have more information. My system scrolls a message telling the user to go to the config portal to setup. That message is still scrolling when it tries to connect. When I turn off that message it succeeds, so I think my scroller is starving the network logic.

But then I have another problem. The first thing I do on startup is to try to access a server using HttpClient and that fails. When I reset and the network starts normally from boot, it accesses it fine

[cmarrin]

And I'm finding that it's not just a timing issue, where I'm trying to access the network too soon. When the network sets up after a successful run of the ConfigPortal, it is not setup correctly. On a fail I try to use HttpClient to talk to the network again and it always fails.

[tablatronix]

connectTimeout is a user setting

[tablatronix]

your sketch doesn't exist

[cmarrin]

Sorry, sketch url updated.

It's not using the connectTimeout when restarting. It using the saveTimeout which I don't see any api to change.

I have a solution!!!

Not sure why, but after the config portal runs and succeeds it sets up the network. But it seems to set it up incorrectly because it can't connect to anything. I'm readying WiFi.localIP() which is correct. But doing this:

HTTPClient http;
http.begin(wuURL);

with a valid URL times out.

So I added a callback with setAPCallback and when it fires I set a flag. On return from autoConnect() if that flag is set I do ESP.reset() to restart the system and it comes back up normally and succeeds.

This is a good workaround for me. Is there any information I can gather to help understand why the network didn't come up correctly after ConfigPortal?

[tablatronix]

Ohh yeah I just added savetimeout change it could be a bug

[tablatronix]

Give me the entire log, you can increase the debug level in .h to 3 also.

So to summarize after autoconnect fail, configportal save, network does not actually work? Or your httpd doesnt work?

[cmarrin]

I increased the debug level to DEVELOPER. Here are the logs.

Coming up normally:

Office Clock v1.0

WM: [1] AutoConnect WM: [2] Connecting as wifi client... WM: [3] wifi station disconnect WM: [1] Connecting to saved AP: marrin WM: [3] enableSTA PERSISTENT ON WM: [1] connectTimeout not set, ESP waitForConnectResult... WM: [2] Connection result: WL_CONNECTED WM: [3] lastconxresult: WL_CONNECTED *WM: [1] IP Address: 10.0.1.109 Wifi connected, IP=10.0.1.109 Getting weather and time feed... URL='http://api.wunderground.com/api/xxxxxxxxxxxxxxxx/conditions/forecast/q/CA/Los_Altos.json?a=9069' got response: 200 Got payload, parsing... Time set to:Sun Apr 08, 2018 08:39:12 AM, next setting in 1308 seconds Temps - current:51, low:48, high:65

After asking it to reset network settings:

WM: [3] unloading WM: [1] SETTINGS ERASED WM: [3] enableSTA PERSISTENT ON WM: [1] AutoConnect WM: [2] Connecting as wifi client... WM: [3] wifi station disconnect WM: [1] No saved credentials, skipping wifi WM: [2] Connection result: WL_NO_SSID_AVAIL WM: [3] lastconxresult: WL_NO_SSID_AVAIL WM: [3] wifi station disconnect WM: [2] Disabling STA WM: [2] Enabling AP WM: [1] StartAP with SSID: MT Galileo Clock WM: [2] AP has anonymous access! WM: [1] AP IP address: 192.168.4.1 Entered config mode:ip=192.168.4.1, ssid='MT Galileo Clock' WM: [3] setupConfigPortal WM: [1] Starting Web Portal WM: [3] dns server started with ip: 192.168.4.1 WM: [2] HTTP server started WM: [2] WiFi Scan done WM: [2] Config Portal Running, blocking, waiting for clients... WM: [2] NUM CLIENTS: 0

After entering credentials into config portal:

WM: [2] NUM CLIENTS: 0 WM: [3] -> captive.apple.com WM: [2] <- Request redirected to captive portal WM: [2] <- HTTP Root WM: [3] -> 192.168.4.1 WM: [3] lastconxresult: WL_IDLE_STATUS WM: [2] WiFi Scan done WM: [3] -> captive.apple.com WM: [2] <- Request redirected to captive portal WM: [2] <- HTTP Root WM: [3] -> 192.168.4.1 WM: [3] lastconxresult: WL_IDLE_STATUS WM: [2] Scan is cached WM: [3] -> captive.apple.com WM: [2] <- Request redirected to captive portal WM: [2] <- HTTP Root WM: [3] -> 192.168.4.1 WM: [3] lastconxresult: WL_IDLE_STATUS WM: [2] Scan is cached WM: [2] <- HTTP Wifi WM: [2] Scan is cached WM: [1] 7 networks found WM: [2] DUP AP: marrin WM: [2] DUP AP: xfinitywifi WM: [2] AP: -24 marrin WM: [2] AP: -81 PabellonExt2.4 WM: [2] AP: -84 ATT296 WM: [2] AP: -88 xfinitywifi WM: [2] AP: -95 BNAN wifi WM: [3] lastconxresult: WL_IDLE_STATUS WM: [3] Sent config page WM: [3] -> captive.apple.com WM: [2] <- Request redirected to captive portal WM: [2] <- HTTP Root WM: [3] -> 192.168.4.1 WM: [3] lastconxresult: WL_IDLE_STATUS WM: [2] Scan is cached WM: [3] -> captive.apple.com WM: [2] <- Request redirected to captive portal WM: [2] <- HTTP Root WM: [3] -> 192.168.4.1 WM: [3] lastconxresult: WL_IDLE_STATUS WM: [2] WiFi Scan done WM: [2] <- HTTP WiFi save
WM: [3] Method: POST WM: [3] Sent wifi save page WM: [2] process connect WM: [2] Connecting as wifi client... WM: [3] wifi station disconnect WM: [1] Connecting to new AP: marrin WM: [3] enableSTA PERSISTENT ON WM: [1] 5000 ms connectTimeout set, waiting for connect.... WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] . WM: [2] Connection result: WL_CONNECTED WM: [3] lastconxresult: WL_CONNECTED WM: [1] Connect to new AP [SUCCESS] WM: [1] Got IP Address: WM: [1] 10.0.1.109 WM: [2] disconnect configportal WM: [1] config portal exiting Wifi connected, IP=10.0.1.109 Getting weather and time feed... URL='http://api.wunderground.com/api/xxxxxxxxxxxxxxxx/conditions/forecast/q/CA/Los_Altos.json?a=67207' [HTTP] GET... failed, error: connection refused(-1) Time set to:Sun Apr 08, 2018 08:45:01 AM, next setting in 120 seconds Temps - current:52, low:48, high:65

[cmarrin]

And I retry the URL every 2 minutes and it remains unable to connect. If I reset manually or add my logic to do ESP.reset() after config portal runs, all is well

[tablatronix]

K thanks, did you add a getmode to your code to see if sta mode is active after wm returns to your code?

[cmarrin]

I just checked. It's in STA+AP mode, which is what it's in when it's working, so it doesn't seem like that's it

[tablatronix]

K thanks, I will try to test and make a sketch. Are you using latest staging of all libs?

[cmarrin]

I'm using the latest released versions of everything, installed via Library Manager. The exception is Adafruit_GFX, where I'm using my own fork with one small change. My plan is to move the change I made there (the ability to get the xAdvance of a given character) into my own sketch so I won't need that anymore. I assume you won't be using any of that display logic in whatever test you're doing.

I will try to make a smaller test case on my end as well

Thanks

[tablatronix]

latest released? as in esp8266 2.4.1? You know it has some major bugs right? have you tried, master/staging to see if this is a esp8266 bug that has been fixed?

[cmarrin]

Oh, sorry. Wasn't thinking about the ESP toolchain. I did try 2.4.1 and it did not go well. I'm using 2.3.0 which I've been using for quite a while and it seems pretty stable.

[tablatronix]

2.3.0 has even bigger bugs. It doesnt even erase flash properly and currupts your flash.

[cmarrin]

Hmmm. I haven't been doing a lot with Flash storage lately. What version do you use?

Just for some background. I've used LOTS of versions of the ESP8266 SDK and many ways of building and running ESP code. Arduino has annoyed me in the past because it has a lot of RAM overhead. Building raw, using https://github.com/pfalcon/esp-open-sdk, I was finding I could save 5K or more. But that required me to use a custom Makefile and didn't give me all the nice Arduino libraries. So I bit the bullet and came back to Arduino. The Arduino toolchain has gotten more efficient in later versions and I've gotten aggressive about using PROGMEM, so my memory usage is fine now.

[tablatronix]

Flash as in wifi settings not eeprom. I added erase flash workarounds in dev but you have to enable it

[tablatronix]

I just fixed that savetimeout turns out 5 seconds was too short, I removed it entirely.

[cmarrin]

Thanks, I'll update and retry. What version of the ESP SDK do you use?

[tablatronix]

I use staging for dev, but I like 2.4.1

[tablatronix]

2.3.0 is ok, but you have to work around some issues #430 But this is going away fast, because its sdk 1.5.3