A few hours ago, we deployed a change to our error capturing process to improve infrastructure performance. During the deployment, one of our services failed to come back online due to a locked file, but due to a gap in our instrumentation we failed to see the problem right away.
You may see a small gap in your error data today while the service was locked.
I regret the mistake and we are taking steps to improve our instrumentation and deployment process so that it does not happen again.