Crashes node app

[olekenneth]

Hi

I get this error several times a day.

node: ../deps/uv/src/unix/threadpool.c:236: uv__queue_done: Assertion `((&(req->loop)->active_reqs == (&(req->loop)->active_reqs)->prev) == 0)' failed.

I use the Tellstick Duo with firmware 12 on a Debian machine.

[Hexagon]

Hmm, that's a tricky one!

64 or 32 bits? Which version of node?

[olekenneth]

root@server:~# uname -m
x86_64
root@server:~# node -v
v0.10.25

[michaelsund]

I have the same problem (x64 with node v0.10.22), and sometimes my listener just stops recieving updates from sensors, allthough i can still see them getting updated with telldus --list

[olekenneth]

@michaelsund does it crash your app or does the app stop getting updates?

[michaelsund]

Both, ive gotten the same error as you maybe once a day and nodemon doesnt restart the app. And maybe every other day the sensor event listener just stops recieving data. allthough the rest of the app continues to work.

[olekenneth]

Do you have the latest firmware?

[michaelsund]

Nope, gonna do a upgrade and see if it makes any difference

Edit: With the latest firmware (12 i think it was for duo) i still get problem with events suddenly not being recieved anymore until a restart of the application.

Havent seen the threadpool error yet though.

[Hexagon]

Well, I might have some progress on this, I've noticed there is a quite heavy memory leak when running with lots of configured sensors and devices. That could have something to do with the threadpool error above ( which i still hasn't been able to reproduce :/ )

[michaelsund]

Got a new error today, ive stripped out everything in my application except telldus and a func to keep the program running. Segmentation fault (core dumped)

[kmpm]

Do you have a sample code?

[hugosp]

node: ../deps/uv/src/unix/threadpool.c:236: uv__queue_done: Assertion `((&(req->loop)->active_reqs == (&(req->loop)->active_reqs)->prev) == 0)' failed.

same error here , always in the middle of the night , anyone found a solution yet ?

[michaelsund]

I have been running my application with this module for a while, seems like when i updated my tellstick duo's firmware some of the problems was resolved. Cant say for sure though, i still need to restart the app sometimes because the events just stops. Im currently forwarding the tellstick via a hypervisor to my virtual machine running the app. Im gonna set it up on a dedicated host to atleast rule that layer out, but thats another story :) nothing todo with above error.

[hugosp]

Strange , i'm on latest firmware on my duo ,version 12 if thats the one you mean and no beta , seams like the errors started when i added a telldus.addDeviceEventListener to my project , can it be somekind of memory-leak in that function ? becouse it happens after about 6 hours of it running .

havent been digging into your code and checking , but just my thoughts

[Hexagon]

Interesting that it only happens after a device event listener is added, that will certainly help in isolating the exact issue. Will give it another go in a couple of weeks unless anyone else beats me to it :)

[kmpm]

I'v made a small project that I will leave running that uses addDeviceEventListener. https://github.com/kmpm/node-tellmqtt

I'll leave it running on a Raspberry Pi for some time and see if I get something similar.

[Hexagon]

A possible fix is merged and ready for testing!

[Hexagon]

Any news on this in the later versions?

[michaelsund]

I have been running the lib on a raspberry pi for 2 weeks without a glitch. Guess the usb-forwarding to the hypervisor was one of my problems.

[Hexagon]

Thanks @michaelsund , sounds promising!

@olekenneth ?

[olekenneth]

All good :-)

[Hexagon]

Nice! Closing this, don't hesitate to reopen if the problem returns :)

[Z3TA]

"version": "0.0.9" nodejs: ../deps/uv/src/unix/threadpool.c:236: uv__queue_done: Assertion `((&(req->loop)->active_reqs == (&(req->loop)->active_reqs)->prev) == 0)' failed.

uname -m x86_64 nodejs -v v0.10.25

Telldus firmware: 12

[Z3TA]

Here's another error, same system:

* Error in `/usr/bin/nodejs': free(): invalid pointer: 0x00000000007654e8 *

[Z3TA]

Here's some debug info before the error in threadpool.c

2015-11-11 (09:48:08_401 debug: sensorEvent: deviceId=199, protocol=temperaturehumidity, model=fineoffset, type=2 value=43, timestamp=1447231688 2015-11-11 (09:48:15_472 debug: RAW: class:sensor;protocol:fineoffset;id:135;model:temperaturehumidity;humidity:66;temp:8.6; 2015-11-11 (09:48:15_472 debug: sensorEvent: deviceId=135, protocol=temperaturehumidity, model=fineoffset, type=1 value=8.6, timestamp=1447231695 2015-11-11 (09:48:15_472 debug: sensorEvent: deviceId=135, protocol=temperaturehumidity, model=fineoffset, type=2 value=66, timestamp=1447231695 2015-11-11 (09:48:22_624 debug: RAW: class:sensor;protocol:fineoffset;id:215;model:temperaturehumidity;humidity:55;temp:11.6; 2015-11-11 (09:48:22_624 debug: sensorEvent: deviceId=215, protocol=temperaturehumidity, model=fineoffset, type=1 value=11.6, timestamp=1447231702 2015-11-11 (09:48:22_625 debug: sensorEvent: deviceId=215, protocol=temperaturehumidity, model=fineoffset, type=2 value=55, timestamp=1447231702 2015-11-11 (09:48:30_495 debug: RAW: class:sensor;protocol:fineoffset;id:167;model:temperaturehumidity;humidity:41;temp:20.0; 2015-11-11 (09:48:30_496 debug: sensorEvent: deviceId=167, protocol=temperaturehumidity, model=fineoffset, type=1 value=20.0, timestamp=1447231710 2015-11-11 (09:48:30_496 debug: sensorEvent: deviceId=167, protocol=temperaturehumidity, model=fineoffset, type=2 value=41, timestamp=1447231710 2015-11-11 (09:48:39_983 debug: RAW: class:sensor;protocol:fineoffset;id:151;model:temperaturehumidity;humidity:82;temp:5.2; nodejs: ../deps/uv/src/unix/threadpool.c:236: uv__queue_done: Assertion `((&(req->loop)->active_reqs == (&(req->loop)->active_reqs)->prev) == 0)' failed.

[Hexagon]

This seem to be because uv_queue isn't thread safe,i guess the chances of this happening is increasing tremendously if listening to both rawDataEvent and sensorEvent at the same time. As these two happens at the exact same time.

Try that as a workaround, I'll look into a solution when there is time :)

[Hexagon]

(removed jibberish)

[Hexagon]

A real solution pushed, replace uv_queue_work with uv_async

[Z3TA]

I've tested the new version and it no longer crashes, but it randomly
stops receiving data witch is much worse!

I get around 4-5 sensor events per second but do not notice anything weird
before it just stops receiving any data.

On Tue, 15 Dec 2015 20:17:06 +0100, Robin Nilsson
notifications@github.com wrote:

I just pushed a possible solution for this, where i removed the deletes
from reqest objects. Can anyone who had this error try out the latest
version on >npm?

— Reply to this email directly or view it on GitHub.

[Hexagon]

How often (in average) does this happen, and in which listener (device/sensor/raw)?

[Z3TA]

I've made my app restart if there is no sensor activity (all listeners) for two minutes and that happens about two times per day. I also still get "./deps/uv/src/unix/core.c:161: uv_close: Assertion `0' failed." error message, but only 1-3 times per week.