Combining Logs and Traces using OpenTelemetry
summary
When a transaction touches at least three services and you run fifty services finding its logs is needle‑in‑a‑haystack hard. OpenTelemetry adds trace id and span id to each log via a log appender or converts logs into span add event calls, preserving original timestamps and structured attributes. The result is indexed, correlated traces with trace id, span id, parent id, resource metadata and request attributes so you can query and aggregate causally linked events across services.
AI-written summary · about
chapters (17)
- 0:00Intro
- 0:27Which means you write logs.
- 0:50What is OpenTelemetry?
- 1:26Adding context turns logs into traces.
- 2:17Logs need context
- 3:26How do you identify which logs are part of which transactions?
- 3:49Finding all of the logs in a transaction is a kludge.
- 4:00Finding all of the events in a trace is a single lookup.
- 4:18Leveraging logs
- 5:58CONGRATULATIONS!
- 7:09What do you get?
- 7:27Example HTTP server logs
- 9:40LET'S TALK ABOUT ANALYSIS
- 10:04is correlated with
- 12:09Tracing is not a niche tool.
- 12:19It's structured data...
- 12:57Halting Problem
related talks
24mObservability and instrumentation via OpenTelemetry
37mThrough the looking glass: Effective observability for cloud native applications
13mStart Tracing on Day Zero
41mIntroducing logs in OpenTelemetry: Features and opportunities
59mOpenTelemetry will save your day and night
34mReduce Complexity with Distributing Tracing
26mOpenTracing Isn't Just Tracing: Measure Twice, Instrument Once
36mOpenTracing Project Deep Dive