Finkle noticed the ping count was unexpectedly high and dropped again to expected numbers after a few days – he discovered that several clients were sending extremely large ping counts. Let's figure out what this was about! I had some thoughts this could be automation so let's make sure we're not sending pings from our automation (bug 1270191). We should also consider whether Google Play tests our builds or not. I believe Finkle is looking into this further.

This issue was caused by a bug in my ETL script. The get_pings method was using a date range, not a single day - which is how the ETL scripts are designed to work. Once I fixed that, the high ping counts are gone. Things look normal. The reason this manifested at all was the backfill I did for the first 5 days, but did single dates after that. The 5 days were combined into each single day, which was the bug. Fixed in the latest script.