Choosing a Simple JSON Logging Service to Detect Stalled SaaS Import Jobs

The article argues that a logging service is only worth adopting if it preserves structured run events, supports safe retries, and offers enough retention and queryability to reveal missing import results. It stresses that alerting logic belongs in the application, since a log entry saying a job finished does not prove it succeeded. The recommended design records a structured event per run with a stable run ID and a terminal status, plus an alert triggered when a normally productive schedule has no recent successful result. It separates domain facts, such as whether a run accepted results, from transport evidence like an HTTP 200 from a collector, which can mask a failed import. The author also notes that hosted endpoints suit some teams but may conflict with data residency, retention, or buffering needs, where a self-managed collector or queue fits better.