Been having an issue on a fresh Digital Ocean/KBPM build where drinks are measured correctly for each pour, but more often then not they are not being displayed in the tablet interfaces activity feed or on the sessions page of the website when saved. I checked the /kegbot-data/media/pics folder and can see that all of the images captured during pours were uploaded to the droplet correctly for the drinks that aren’t showing up. When I view the System --> Drinks page in the web interface there are numerous missing drinks.
Anyone had any issues like this before? I am running on the smallest droplet on Digital Ocean. I notice when I log in that the memory usage is hovering around 90%. Is this normal, or could it have an effect on the server receiving and logging pour information from the tablet correctly?
Here is a screenshot showing the gaps in drink logging.
Another user on here mentioned his 512 MB RAM Digital Ocean droplet was only utilizing about 18% of memory so I’m starting to think my install is a bit d*cked. I ran htop and see multiple instances of various processes (screenshot attached). Any idea what is causing the duplication/unnecessary consumption of resources? I didn’t seem to have any issues upgrading my fresh Digital Ocean install to 1.2.1 and here are the steps I took, based off of my late night notes at the time:
**> Complete Digital Ocean install steps:
HowTo: Install Kegbot Server on Digital Ocean
Complete steps to upgrade from 1.1.1 to 1.2.1:
[email protected]:~# su - kegbot
[email protected]:~$ . kegbot-server.venv/bin/activate
(kegbot-server.venv)[email protected]:~$ pip install --upgrade kegbot
(kegbot-server.venv)[email protected]:~$ kegbot upgrade
At this point after the upgrade and reboot, the web page will display 502 bad gateway errors. To get past this, modify the kegbot.conf file located in /etc/supervisor/conf.d/ to include the following statements:
command=/home/kegbot/kegbot-server.venv/bin/gunicorn pykeg.web.wsgi:application -w 3
Perhaps I messed up the install and upgrade to 1.2.1, but I never got any error messages and everything seemed to work alright until I noticed the random unlogged pours and resource exhaustion. As always, thanks.
Here are two more screenshots of htop running where userland threads are hidden, sorting is top down by memory usage and entire command strings are shown.