R352 Raspian stops working after 2hrs

[Jens]

Hi Gene,

today I upgraded from R345 to R352 on Raspian

- stop HG
- remove the package
- clean up sudo apt-get autoremove
- update sudo apt-get update
- sudo gdebi ...

Within R352 I started from scartch with creating my config

I use X10, Hue and UpnP within Interfaces and deselect the APPs which I do not make use ogf like ZWave.

After Reboot I have HG running

~ $ sudo /etc/init.d/homegenie status
root      2303  2302  0 11:37 ?        00:00:00 sudo /usr/bin/mono HomeGenie.exe
root      2308  2303 51 11:37 ?        00:00:28 /usr/bin/mono HomeGenie.exe
[ ok ] homegenie is running.

After approx 2hrs HG stops working and is not any longer part of the processes running

The last reliable Version running for long is R345 in this setup

Please help

Thanks
Jens

[Gene]

Hi Jens,

I'm running r352 and it's still up after about a day.
My setup is almost complete: X10, Z-Wave, Philips Hue, UPnP.
What you can do to help in finding the problem is:

1) install screen (sudo apt-get install screen)
2) stop HG as a service (sudo /etc/init.d/homegenie stop)
3) open a screen session (screen)
4) run hg in the screen session (cd /usr/local/bin/homegenie; sudo mono HomeGenie.exe)

With CTRL+A, D you can detach the screen session and logout.
When it happens that HG stops running again, you can attach the HG screen session with screen -x and see in the console the error that made HG stop.
Also enable logging, there might be useful informations in the log file to better understand what happened.

Cheers,
g.

[Gene]

Before all of that also do a:

Code: [Select]

sudo apt-get update
sudo apt-get dist-upgrade
sudo rpi-update
sudo reboot

since new kernel drivers and mono 3.2.8 are now available.

Cheers,
g.

[Jens]

Thanks for the tips. I turned on logging after it stopped last and will wait for the next stop.
Then I give it a stepwise approach (your second post first)

This was already the problem with R350/351 both stopped the same way, why this time I started from scratch with a new config instead of restoring the config from R345 into R352.

Regards
Jens

[bkenobi]

FWIW, I updated from r351 (broken) to r352 about 2 days ago.  It is still responding for me.  I am running X10 only with most programs disabled.  I have only 4 modules set up for testing and it is not set up to send any commands since I have this setup filtered off from my main PLC network.

I just moved the CM15A off the filter and will leave it there for a while to see if it becomes unstable for some reason though that was not the case in previous builds.

[Jens]

So I did perform another clean install yesterday evening
including the suggested commands in the install procedure

sudo apt-get update
sudo apt-get dist-upgrade
sudo rpi-update
sudo reboot

HG stopped shortly before midnight, these are the last entries in homegenie.log

2014-03-29T23:43:28.4959560+01:00   HomeAutomation.X10   A5   X10 Module   Meter.Watts   0
2014-03-29T23:43:58.6911650+01:00   HomeAutomation.X10   A5   X10 Module   Meter.Watts   0
2014-03-29T23:44:29.0925520+01:00   HomeAutomation.X10   A5   X10 Module   Meter.Watts   0
2014-03-29T23:44:59.2447200+01:00   HomeAutomation.X10   A5   X10 Module   Meter.Watts   0
2014-03-29T23:45:29.4372370+01:00   HomeAutomation.X10   A5   X10 Module   Meter.Watts   0
2014-03-29T23:45:59.5872970+01:00   HomeAutomation.X10   A5   X10 Module   Meter.Watts   0
2014-03-29T23:46:30.2473010+01:00   HomeAutomation.X10   A5   X10 Module   Meter.Watts   0


There's nothing in any log in /var/log/ around this time, HG was up for 3hrs40mins.

The light A5 has configured a value for VM Watts and is showing up like above every 30sec,
the other lights e.g. A10 which are not configured for VM Watts and still have 0 for that entry
do not show up in HG log - a hint?

Thanks

Regards
Jens

[badfinger]

Hi Gene,
my HG stops faster.......it don't needs 2 hrs...
Something seems to be buggy.  I made all updates. It is a full new installation from the Raspi image.
Attached you'll find my complete log from the start to the end of a running session. I hope you'll find something.

stay tuned

Alex

[Jens]

It's running <in Screen> still alive, but horribly slow. the Android APP cannot access HG anymore, my guess is due to timeout. I'll wait for the complete stop.

TOP tells me MONO is using less than 5% CPU and 15% MEM.

R
Jens

[Gene]

If you're using Z-Wave, try unplugging the controller for 10 seconds and then replugging it.
Yesterday I caught into the same bug, but have no clue how to replicate it systematically.
You seem to hold the key of this =) There's something in your configuration that make the bug alive at every run.
More infos about your configurations and automation programs you're using could help.

Thanks,
g.

[badfinger]

Hi Gene,
unplugging the z-wave stick for 10 seconds bring no better results. HG crashes again. Is it possible to get a more detailed log file?

stay tuned

Alex

[Jens]

Hi Gene

I have no Zwave at all, the corresponding APPs are deactivated within HG UI

- Setup is Raspberry B Model with CM15Pro and 8GB UltraFlashCard with expanded filesystem
- Shairport running as the only additional program (I am not streaming during these test phases)
- no additional programs besides the ones you provide
- 5 additional wizard scripts in scenes (react on forwarded InfraRed Commands)
- Use of HG_Led and DHT11
- Interfaces X10 HC A and P, UPnP
- Hue lightning with 6 bulbs (updated trigger code from 5->6) and wheatherunderground activated
- Some devices make use of the scheduler, no additional entries within the scheduler itself

maybe I can post my config file later on - already to slow right now, currently HG ist still running in Screen getting slower and slower but still alive ;-)

Entries in Screen so far

pi@raspberrypi /usr/local/bin/homegenie $ sudo mono HomeGenie.exe                                                                                           
`string' does not contain a definition for `moon_phase'
`string' does not contain a definition for `moon_phase'
`string' does not contain a definition for `moon_phase'
`string' does not contain a definition for `moon_phase'
`string' does not contain a definition for `moon_phase'
`string' does not contain a definition for `moon_phase'
`string' does not contain a definition for `moon_phase'
`string' does not contain a definition for `location'


After a while I could enter the UI, within interfaces X10 and UpnP show now up as disabled, but the icons at the top of the webpage are still happy. Enabling/disabling has no effect anymore on the icons.


Thanks
Jens

[Gene]

Thanks for the detailed report.
I was thinking the bug was about a problem in the SerialPortLib (used by ZWaveLib as well), but your setup is not using it at all.
I made a little change to SerialPortLib (replaced the ThreadPool with Thread with high priority, since on rpi z-wave was not preforming well, expecially in the initialization phase at startup).
Also made some relevant improvement to the UI and real-time server-to-client events.
Published all of that as r356:

https://sourceforge.net/projects/homegenie/files/testing/

this is not likely solve the bug, but we never know (damn Murphy!) =)

Cheers,
g.

[badfinger]

Hi Gene,
hmmm i'm wondering. Since i deactivate my CM15pro it seems to be working.    My z-wave is running good. Is there something with the X10 libary?

stay tuned

Alex

[Gene]

I will check for now I'll let r356 run for a while with complete setup.
If you have a little time, upgrade to r356, I need more testers =D

g.

[Jens]

I am about to install 356 on a separate raspberry.

I checked my log, 336 was the last version running for days.
Then I went to 345 which is already slowing down a little bit,
but I cannot remember a complete stop. I did not try any version
between 336 and 345.
350 got me into all the trouble as I posted already.

Cheers
Jens