We use Amazon S3 to store assets such as website files and Psiphon server discovery lists. We sometimes enable logging of downloads of these files. Analyzing these logs helps us to answer questions like "how many users are starting but not completing the download of the server discovery list?", "how is the downloaded data split between website assets and server discovery?", and "is an attacker making a denial-of-service attempt against our websites?" </p> <p> S3 bucket access logs contain IP addresses, user agents, and timestamps. These logs are stored in S3 itself, so Amazon has access to these logs.