Configure log retention and sampling for my app. Most log providers meter by ingestion and storage volume, and default settings will happily bill you for terabytes of debug logs nobody reads. Do the following: 1. Set retention windows by log type: short (7-14 days) for high-volume info/debug logs, longer (30-90 days) for warn/error and audit-relevant logs. Anything required by compliance goes into a separate archive tier with the mandated retention. 2. Sample high-volume logs: if I'm logging every request, sample at 10% for successes and 100% for errors. Same for third-party API calls or hot code paths. Don't sample audit events. 3. Cut noise at the source: level-check before formatting expensive log messages, silence chatty libraries that log 'connected' every 30 seconds, and stop logging the same thing at multiple levels. 4. Never log secrets, tokens, or full request bodies. Add filters at the logger level so PII scrubbing doesn't rely on remembering. 5. Audit-relevant events (auth, permission changes, admin actions, payments) get logged separately from operational logs, with longer retention and stricter access. These are the logs you'll need in a security incident and don't want mixed into the firehose. 6. Cost check: look at current log volume and estimated monthly cost. Flag the top three noisy sources; often 80% of the volume is one bad log line. Don't cut error retention to save money. The deliverable is retention configured per log type, sampling in place on high-volume paths, noisy sources trimmed, audit logs isolated, and a documented monthly cost estimate.