Release v2.0.0-dev7 - possible Bug in Controller

[toffel969]

Yes please!

[Shardan]

Yes please!

A 4096-memsized one? Or do you need another flashsize?

Regards
Shardan

[toffel969]

Yes please!

A 4096-memsized one? Or do you need another flashsize?

Regards
Shardan

Yes a 4096-mesized. Target node is a Wemos D1 mini pro with 2 D18B20 Temperature probes, 4 buttons, analog soil moisture probe and an SSR connected directly to GPIO as output actuator.

regards

Sebastian

[Shardan]

Here you go.
This is the version i used for bulk testing, it is running on nodeMCU and WeMos for around 2 days without issues.
It is compiled with the "normal" option, so no testing or development parts are compiled in.

firmware_normal4096.zip

(346.57 KiB) Downloaded 523 times

.
Anyone who wants to use it:
This is NOT an official release from the developers but one i compiled myself.
Don't expect support. Further the 2.0.0 is a work in progress so instability and crashing may occur.

Use at own risk!

Regards
Shardan

[toffel969]

Here you go.
This is the version i used for bulk testing, it is running on nodeMCU and WeMos for around 2 days without issues.
It is compiled with the "normal" option, so no testing or development parts are compiled in.

firmware_normal4096.zip

.
Anyone who wants to use it:
This is NOT an official release from the developers but one i compiled myself.
Don't expect support. Further the 2.0.0 is a work in progress so instability and crashing may occur.

Use at own risk!

Regards
Shardan

Thanks, its flashed. I had dev-dev8 installed on the node, did OTA (node is already "in the field"). It reset all my devices and settings, will try to replace the devices (what were the gpios that I used !?! lack of documentation
is a bi**ch , particularly when there is nobody to blame but yourself ) and connect it to domoticz later when my daughter is sleeping.

Regards

[Shardan]

You're welcome

[toffel969]

You're welcome

Correction to above post... I flashed the wrong node it was still running RC147, so no wonder I lost the devices and settings. However now I flashed the correct node, it didn't reset the settings. Now on the correct node I can also find my sensors . Its reporting the data correctly to domoticz and seems to run stable. will let it run at least until tomorrow morning.

[Shardan]

Thanks for testing.
Check the free RAM on the webside, it should not decrease over time.

Regards
Shardan

[toffel969]

Thanks for testing.
Check the free RAM on the webside, it should not decrease over time.

Regards
Shardan

After 2h30 operation
Free Mem: 22064 (11720)
so far stable and reporting values to domoticz. So it seems to be working.

If I feel adventurous I will try to flash dev8 back tomorrow morning. Gegenversuch

Regards

[Shardan]

Would be interesting indeed.

Regards
Shardan

[toffel969]

Would be interesting indeed.

Regards
Shardan

Update:
Uptime: 0 days 15 hours 57 minutes
Free Mem: 21408 (11720)

still going strong. I consider this working and will now try to flash dev-dev8 to see if "this unknown sniplet of code " (at least to me) fixes the problem with the controller/unresponsiveness or if it was somewhat already included in the binaries of version dev.8

[toffel969]

I flashed the dev-dev8 version. It just continues to work and send values to Domoticz. Now running for 1h stable , web interface is responsive, Data is reported to Domoticz.

Free Mem: 21840 (28520)

With the previous bug, it would have stopped working quite some time ago. Now I am not sure if the problem arises when you setup the controller, because thus far, this is where it always started going wrong. Now i setup the controller in the "shardan"-version and just OTA the dev8.

I will continue to let it run on Dev8 for now

[Shardan]

Okay.

I analyzed the failing connection with help from psy0rz and we found the memory
rapidly decreasing when my FHEM controller is connected, a memory leak.
Psy0rz was able to reproduce the bug and fixed it, That's what is compiled into
the binary.

So it would be nice if you check the free RAM now and then.

Regards
Shardan

[toffel969]

GIT version: v2.0.0-dev8
Local Time: 2017-05-06 21:55
Uptime: 0 days 9 hours 18 minutes
Load: 10% (LC=26251)
Free Mem: 21472 (28520)

still going strong. maybe psy0orz has an explanation. next would be dev 7 though im thinking thats a bit pointless

[maluko]

GIT version: v2.0.0-dev8
Local Time: 2017-05-07 0:39
Uptime: 2 days 13 hours 51 minutes
Load: 20% (LC=22798)
Free Mem: 23656 (16312)

this work with 12 tasks ( 8 mpc switch, 1 wifi info, 1 uptime, 1 lux, 1 temp) all sending to domotiz.

edit: more rules for each mpc switch with timerset

[toffel969]

maybe i jumped the gun. last night at 0:00 i checked again and the webserver was not responding. it was still reporting sensor data to domoticz and responding to pings. i also managed to reboot it via http , but webserver was definitly not working anymore. however after reboot it has now 9h uptime again. this is not the same behaviour i saw with dev7 (after reboot i had to hurry to ota to set itback to rc148) unfortunatly i cant give serial output though the node is already installed in a ip54 box outdoors.

[Shardan]

Hello toffel,

how long is the delay for reporting?
The memory leak occurred here every time the ESP reported to my FHEM server.
If i disabled the "Send data" in the tasks it stopped leaking.

So for testing we used delays of 1s and 5s bringing down the free ram very fast
until the ESP first stopped website, then stopped reporting and at last crashed.
Keep in mind I'm using FHEM, with other servers it may behave different.

Have a nice sunday

Regards
Shardan

PS: If that is possible with your server, maybe use "System Info" in an additional task and
report "Free RAM" to your server. I've logged the data with FHEM for being lazy

[toffel969]

After the last crash I just restarted the node.

Unit: 12 GIT version: v2.0.0-dev8
Local Time: 2017-05-08 12:23 Uptime: 1 days 11 hours 57 minutes
Load: 8% (LC=26666) Free Mem: 21656 (28520)

Now its starting to get weird. I am running with a 60 second delay.


So maybe it wasn't the controller bug after all. Not sure what to think....

[Shardan]

hmm... interesting (Spock would say "fascinating"..)

Did you re-download the dev8 or is it the same file as before?

Would be the easiest explanation.

Regards
Shardan.

[toffel969]

hmm... interesting (Spock would say "fascinating"..)

Did you re-download the dev8 or is it the same file as before?

Would be the easiest explanation.

Regards
Shardan.

I didnt reflash between the crashes. So its still the very same version that malfunctioned
Unit: 12 GIT version: v2.0.0-dev8
Local Time: 2017-05-08 21:00 Uptime: 1 days 20 hours 30 minutes
Load: 14% (LC=24912) Free Mem: 20984 (28520)

Free mem now dropping a little bit

[toffel969]

Unit: 12 GIT version: v2.0.0-dev8
Local Time: 2017-05-09 19:30 Uptime: 2 days 18 hours 50 minutes
Load: 10% (LC=26397) Free Mem: 21840 (28520)

[Shardan]

Strange.
Well i have to pass on that point.
With FHEM the issue was reproducable easily.
I've not hte slightest idea why it disappeared on your settings.

Regards
Shardan

[toffel969]

Strange.
Well i have to pass on that point.
With FHEM the issue was reproducable easily.
I've not hte slightest idea why it disappeared on your settings.

Regards
Shardan

Ya I also give up on understanding it... I will keep the dev8 on there and monitor.

[pppp33]

Hi there,
while definitely solved the reboot issue, with uptime of days, now (with 2.0.0-dev8) I can't read my pulse counter (that worked in dev7)
Am I the only one ?
Thx & regards
Paolo

[Shardan]

Hi Paolo,

i don't use a pulse counter atm so i can't say.
It might be better to open a new thread for that as noone will find your posting here.

Regards
Shardan