Lilygo_T3_V1.61, locking up

I’ve been running OMG on Lilygo for a week of two, getting familiar with it. I was concerned that it was somewhat unstable and suspected my WiFi and Internet (because they are not 100%), but after reading I added Steph’s changes.
Today I caught the error:

[1947528][E][WiFiClient.cpp:422] write(): fail on fd 51, errno: 11, "No more processes"
[1947558][E][WiFiClient.cpp:422] write(): fail on fd 51, errno: 113, "Software caused connection abort"
*wm:[2] [EVENT] WIFI_REASON:  200
W: MQTT connection...
[1947575][E][WiFiClient.cpp:242] connect(): connect on fd 50, errno: 118, "Host is unreachable"
W: failure_number_mqtt: 1
W: failed, rc=-2
W: MQTT connection...
N: Connected to broker
[1960053][E][WiFiClient.cpp:422] write(): fail on fd 51, errno: 11, "No more processes"

Which may be one of several possible causes of hangups, but at least that’s a start.

So here’s my question, how can I track the number of processes? Or more generically (if it’s memory related) find what is using memory?

Soo, I’ve pulled the latest from github with out the changes. It has run twice as liong as I was getting (45 minutes).

I’ll leave it alone till tomorrow.

I’m not seeing errors

would suggest to just set “#define UTCtimestamp true” in User_config.h
without other modifications and see if your problem is back

you aren’t alone with similar problem : LilyGO devices stop responding after a few hours

The gateway publish its freememory and other parameters to hom/OpenMQTTGateway/SYStoMQTT
Could you share what you have here before the freeze?

it’s been a lot more stable with the ‘vanilla’ version.

I’ll try adding “#define UTCtimestamp true” shortly.

I added the UTCtimestamp, and it seems stable, Free memory still around 110,000 Bytes.

The graph above showed the rapid decline I was getting back on the 25th (localtime). The one between 3 and 6 PM would have included a WiFi/Internet dropout, but it seems like the lockup is under 40,000 Bytes. What would need that much?

here’s a couple of the SYStoMQTT messages from MQTT Exporer, just before and after I did the re-build, but I’m not sure how I can go back to the earlier crashes.


________________ Rebuild with UTC time _________________

Would earlier stuff be in the HA database?

Could you share the configuration that you are using and has issues?

I’ve put Steph’s ‘add timestamp’ code in to ZgatewayRTL_433.ino. For the last 50 minutes free memory is sitting around 106-108 thousand Bytes.
So, all seems OK now.
One thing that may have changed was when I tried to go back to ‘original’ code I ended up with the latest ‘development’ code.


So, I made the changes at 3:43, all still looking stable.

After some investigation I think I started out using the pre-compiled binaries (esp32dev-rtl_433) that I downloaded on 8th April, and when I saw Steph’s timestamp changes I moved to VSCode and cloned the repository.
I see now that is a few hundred byes bigger, so there must have been a change there. Sorry I don’t know enough about github to look at what they were and what might be relevant.

All good now, so thanks for the support.


1 Like