Engineering for Reliability
Understand your services with Cloud Logging
Twenty-three substantive episodes on SLOs, telemetry, alerting, investigation, and operations
Twenty-three 2021–2022 episodes connect user-facing reliability targets to metrics, logs, traces, and alerts. Examples show instrumentation, Prometheus collection, quota monitoring, GKE investigation, and observability cost controls.
Google Cloud Tech 13 of 23
In this collection Browse 23 summaries 13 of 23
Once measurements show an elevated error rate or a slow service, logs help explain which events occurred. This introduction follows an event through Cloud Logging, from collection and storage to filtering the records needed for an investigation 00:00–01:01(opens in a new tab).
Key Points Covered
- An entry describes an event and its origin: It contains a timestamp and a payload—the message, as text or structured data such as JSON—and may identify the resource, log name, and severity 00:00–02:05(opens in a new tab).
- Routing decides where a log is kept: A sink is a routing configuration whose filters select entries for a destination. Destinations include indexed Cloud Logging buckets, Cloud Storage for archives, BigQuery for analysis, and Pub/Sub for further processing. An exclusion prevents matching entries from reaching that sink's destination 01:01–03:07(opens in a new tab).
- Shared storage need not mean shared access: A log view limits which records a user can read within a bucket. The example centralizes several projects' logs while letting one user see only one project's records and another see all of them 03:07–04:11(opens in a new tab).
- Filter before inspecting or streaming: Logs Explorer narrows the example to one Kubernetes container's error logs, then exposes each entry's payload and metadata. A query can also select the logs to stream as new entries arrive during debugging 04:11–06:16(opens in a new tab).
Full video: https://www.youtube.com/watch?v=IlUCyV8mcS0(opens in a new tab)