Follow-up · depth 2
What metrics prove Artifact Promotion (Build Once) is healthy in prod?
What metrics prove Artifact Promotion (Build Once) is healthy in prod?
Answers use simple, clear English.
Quick interview answer
Prove Artifact Promotion (Build Once) is healthy with latency (p50/p95/p99), error/saturation rates, and queue/event-loop lag — not just CPU. Context: At Mid-level depth: start with the problem, then mechanism, then a short example. Build a versioned image/package once in CI; promote the same digest through staging→prod.
Detailed answer
Production health signals for Artifact Promotion (Build Once): • Latency: p50/p95/p99 of the critical path that uses Artifact Promotion (Build Once). • Errors: failure rate, timeouts, and retry storms. • Saturation: queue depth, event-loop delay, thread/pool utilization. • Business SLIs: request success and user-visible freshness where relevant. Trade-off lens: Pros: What you tested is what you ship. Cons: Needs good registry + digest pinning. Explain thresholds + alerts, then how you triage. Parent context: At Mid-level depth: start with the problem, then mechanism, then a short example. Build a versioned image/package once in CI; promote the same digest through staging→prod.
Full explanation
Good answers name measurable signals and what “bad” looks like for Artifact Promotion (Build Once). Build a versioned image/package once in CI; promote the same digest through staging→prod. Never rebuild for prod with different flags.
Follow-up questions
Only answered follow-ups are shown — click to open with full answers
Parent context — Artifact Promotion (Build Once)
At Mid-level depth: start with the problem, then mechanism, then a short example. Build a versioned image/package once in CI; promote the same digest through staging→prod.
View full parent question →