Resolved -
Removing the added instrumentation has brought the situation back to normal.
May 19, 13:31 PDT
Monitoring -
We believe added instrumentation to some of our lambda queries were to blame for rare errors. We reverted the change and things seem to have improved rapidly. We are keeping an eye on the situation in case something else is at play.
May 19, 13:05 PDT
Investigating -
We have noticed higher than usual error rates when running Honeycomb queries. We are currently investigating the potential sources behind this issue.
May 19, 12:59 PDT