On 2019-06-26 17:11, Mark Webb-Johnson wrote:
You can rule-out the vehicle module as the cause of the problem by simply turning off the vehicle type, so no vehicle module is loaded at startup.
Duh, I should have thought of that, excellent suggestion. The crashes continued. What's interesting to me is my dev module (an early V3.1) is on the bench, powered by a 12V brick and running the same code but never seems to crash. The canbus is not connected to anything but also it does not have a modem card. Then see if it still crashes.
Similar for server connections (are you using v2 or v3?).
To be able to use the iphone app I have v2 enabled. On 2019-06-26 22:38, Michael Balzer wrote:
This type of crash is the reason I now build the edge version on dexters-web.de without CONFIG_WIFI_LWIP_ALLOCATION_FROM_SPIRAM_FIRST.
It seems to reduce these crashes, but I haven't done new statistics yet.
Craig, you could also try if that helps for you.
Another good suggestion; instead of just editing sdkconfig directly I used: gmake menuconfig and then Component config -> ESP32-specific -> SPI RAM config and then turned off: Try to allocate memories of WiFi and LWIP in SPIRAM firstly. That took the "y" off of CONFIG_WIFI_LWIP_ALLOCATION_FROM_SPIRAM_FIRST. So. I haven't driven the car (it's the Z06's turn this week) but no crashes for almost 12 hours, a new record since I've had the module in the Cadillac. And since I was worried that noise from my battery tender was crashing the module it's been running off of the (very new, large delco AGM) battery since Sunday and my DVM still shows 12.7V. So I'm thinking turning off the cellular and gps when parked for N minutes, waking up, getting a fix, uploading to the server and turning back off might a good first pass on power savings mode. (But I really need to get a meter inline and see what the actual draw is when running on wifi vs. cellular...) Craig