From about 04:04 UTC the API was unavailable for about 21 minutes after a routine database connection restart stopped the API process, and the process did not restart on its own. The website stayed up but sign-in and data requests failed. Service was restored by a manual restart, and the API now survives dropped database connections and restarts automatically on failure.
Between 14:22–15:07 UTC some API requests experienced elevated response times of 2–5 seconds. Root cause was a slow database query introduced in a deploy. A hotfix was shipped at 15:07 UTC and all metrics returned to normal.
Volunteer reminder emails were delayed by up to 45 minutes due to a transient issue with our email provider. No emails were lost. All queued messages were delivered by 11:30 UTC.