Announcement: AppDaemon 1.5.1

A little bugfix - kinda.

Some of the functionality around logging substitution seems to be very sensitive to Python versions and is causing problems for folks not on 3.5. I am temporarily removing that functionality until I can figure it out (and I am travelling a lot on business for the next couple of weeks and didn’t want to leave you hanging)…

1.5.1 (2017-01-30)

Features

None

Fixes

  • Functionality to substitute line umbers and module names in log statements temporarily removed

Breaking Changes

  • Functionality to substitute line umbers and module names in log statements temporarily removed

Looks like this doesn’t play nice when running the dev version…

2017-01-30 08:51:47.626748 INFO AppDaemon Version 1.5.1 starting

Traceback (most recent call last):
File "/usr/local/bin/appdaemon", line 11, in <module>
load_entry_point('appdaemon==1.5.1', 'console_scripts', 'appdaemon')()
File "/usr/local/lib/python3.5/site-packages/appdaemon/appdaemon.py", line 1250, in main
conf.version = int(ha_config["version"].replace(".", ""))

ValueError: invalid literal for int() with base 10: '0380dev0

Thanks for creating the issue - response is on GitHub.

Since upgrading to 1.5.1 I have noticed a considerable increase in CPU usage by appDaemon - from 1-2% to 10-11%. Is this expected?

The only thing that might have caused this is the swap to websockets - I’ll do some experiments when I get home.

Did you find anything on this? I’m also getting a lot of

2017-02-20 13:16:27.361831 WARNING Scheduler clock skew detected - delta = 8.361763954162598 - resetting
2017-02-20 13:16:27.467506 WARNING Scheduler clock skew detected - delta = 1.5325579643249512 - resetting
2017-02-20 13:16:27.572861 WARNING Scheduler clock skew detected - delta = 1.4272029399871826 - resetting
2017-02-20 13:16:27.677291 WARNING Scheduler clock skew detected - delta = 1.3227729797363281 - resetting
2017-02-20 13:16:27.781781 WARNING Scheduler clock skew detected - delta = 1.21828293800354 - resetting

messages in my appdaemon log. running top, my load averages are around in the high to mid 20’s and my appdaemon cpu utilization is consistently around 10-15%.

I’m on HA 0.38.3 and AD 1.5.2

You can try reverting to SSE as a test - --commtype SSE. Let me know if that changes anything.

is that one or two dashes

It is 2. (10 characters)

ExecStart=/usr/local/bin/appdaemon --commtype SSE -c /home/homeassistant/appdaemon/conf/appdaemon.cfg

if that is correct, SSE is hitting homeassistant harder and driving it’s utilization up. AD is all over the place with peaks at around 102% of the cpu and mins down around 4%. With the sockets it was just fluctuated a little around 10-15%. I’m getting around the same number of clock skews.

I also tried reverting to SSE, and it made no difference to me.

My appdaemon is still running around 10-12% of the cpu, on a Pi B+. However, I am not seeing any clock skew.

I did some profiling on a test VM over the weekend, but couldn’t see anything obvious that stood out.

I just looked at the numbers again after giving it a while to settle down. No difference between SSE and non-SSE. Same behavior. The high numbers were just startup numbers where everything was initializing.