As we get closer to system deployment (March 2014), I am running more system level tests. The first of these (the week of November 17-24) revealed the following problems that need to be addressed:
- The DHT22 sensor is returning more errors that it should and those errors are getting logged as good data. Either we aren't using the checksum correctly or the sensor is actually sending bad data in which case we will put a software filter on the output. See the spikes in the following picture from RasPiConnect:
 |
| DHT22 Humidity Bad Data Spikes |
Note the windspeed chart. We are adding a pitot tube windspeed indicator to our environmental subsystem in the next few weeks.
- We are experiencing occasional failures in sending email from ProjectCuracao outside of the systems. In one case (the initial Reboot and startup email) it is not in a separate thread so it takes the whole Project Curacao main software down when it bombs. We have good log information so we should be able to track this down. We will put in Try/Recover blocks, but we'd still like to see what is causing the problem.
- After three days on the Pi, the Project Curacao system is failing because we can't open any more files. We installed "lsof" on the Pi and we are tracking this down. This is a lower priority bug at this point because the BatteryWatchdog will shut down the Pi on a daily basis in any case. Some where, someone is not shutting a file that has been opened and they accumulate.
We swapped out a LiPo Rider Pro board that seemed to have a problem with charging the Watchdog battery properly. It was discharging and charging in a much different manner than the charge controller for the Pi. We'll know in a couple of days whether it works properly.
No comments:
Post a Comment