gunicorn high CPU usage and alarmdecoder-webapp unresponsive
Posted: Sun May 01, 2016 10:50 pmHello,
I've got a problem that starting occurring whereby the alarmdecoder-webapp no longer functions. I try going to the web URL and I get a 504 error, gateway timeout.
I'm running this on a Raspberry Pi and have been successfully for months. There are a couple things that I've done recently. One is that I attempted to set up cameras in it and had trouble. So, I followed the manual instructions to enable cameras (though the option was already there, it wasn't working for me--when it attempted to show the camera, it would show a broken image icon and just reload rapidly). That process, after a long time of trying to build things, eventually failed. The webapp continued to work fine from that point, however.
The other thing I did was that I set up OpenRemote to query the REST API every 2 seconds for two different calls. Both were just the GET /alarmdecoder API--once to get the armed status, the other to get faulted zones. This seemed to work fine for a day or two.
Then, for whatever reason, I couldn't access the webapp--nor could OpenRemote.
When I SSH'd into the Pi, I ran Top and noticed that 'python' was using 75 to 90-something % CPU. 'ps auxww | grep python" showed that it was gunicorn running the webapp.
I tried to enable debug logging, but I don't see anything interesting in the instances log, other than it having a debug message "1 workers" every second.
Any thoughts on how I can get more telling logs of why this is sucking up so much CPU? Ideally we could figure out what is wrong and fix it directly, and learn from the experience. But, if wiping this and reinstalling the webapp can be done pretty easily, I'd be open to that, as well.
Oh, and I started from the AlarmDecoder-provided image previously.
Thanks for any help!
Carlin
I've got a problem that starting occurring whereby the alarmdecoder-webapp no longer functions. I try going to the web URL and I get a 504 error, gateway timeout.
I'm running this on a Raspberry Pi and have been successfully for months. There are a couple things that I've done recently. One is that I attempted to set up cameras in it and had trouble. So, I followed the manual instructions to enable cameras (though the option was already there, it wasn't working for me--when it attempted to show the camera, it would show a broken image icon and just reload rapidly). That process, after a long time of trying to build things, eventually failed. The webapp continued to work fine from that point, however.
The other thing I did was that I set up OpenRemote to query the REST API every 2 seconds for two different calls. Both were just the GET /alarmdecoder API--once to get the armed status, the other to get faulted zones. This seemed to work fine for a day or two.
Then, for whatever reason, I couldn't access the webapp--nor could OpenRemote.
When I SSH'd into the Pi, I ran Top and noticed that 'python' was using 75 to 90-something % CPU. 'ps auxww | grep python" showed that it was gunicorn running the webapp.
I tried to enable debug logging, but I don't see anything interesting in the instances log, other than it having a debug message "1 workers" every second.
Any thoughts on how I can get more telling logs of why this is sucking up so much CPU? Ideally we could figure out what is wrong and fix it directly, and learn from the experience. But, if wiping this and reinstalling the webapp can be done pretty easily, I'd be open to that, as well.
Oh, and I started from the AlarmDecoder-provided image previously.
Thanks for any help!
Carlin