Linking Errors To Traces

Link errors to traces and logs

Use a shared ID to connect records. A matching service and time do not prove that a log or span belongs to an error.

Find the exception

For an issue ID or URL, call posthog:query-error-tracking-issue-events with include: ["exception", "correlation"], filterTestAccounts: false, and a small limit. Choose one occurrence unless the user asks for several. Say when the result is a sample.

For an occurrence ID or session ID, read finding-exceptions.md (opens in a new tab) for the exact lookup. Use the user’s time range when given. If no event appears, state the window you checked. Do not claim that an event never existed based on a bounded search.

Follow the IDs

1. Replay: If the event has $session_id, call posthog:session-recording-get with that ID. found: false means no recording was found; the error or session may still exist. Use posthog:generate-app-url with /replay/{id} when a recording exists.
2. Exact trace: For each unique, nonzero, 32-character hex $trace_id, call posthog:apm-trace-get with that ID and excludeAttributes: true. Use its _posthogUrl for the trace link. If the event has $span_id, compare it with returned span IDs as hex, ignoring case. An empty result means no retained span was found for that trace ID. Logs may still exist.
3. Session trace candidates: If an event lacks $trace_id and the user asked about session traces, use posthog:apm-attributes-list to find a span attribute that stores the session ID. If a clear key exists, call posthog:query-apm-spans with an exact span_attribute filter, a short window around the error, flatSpans: true, and a small limit. This finds matching child spans as well as root spans. Call posthog:apm-trace-get for a candidate that needs a link. Label these as session matches. Time alone does not prove which request raised the error.
4. Same-trace logs: When an event has a trace ID, read finding-trace-logs.md (opens in a new tab) for the bounded service lookup and exact log filter. Include every severity, including info. A trace can have logs even when no span was retained.

Use posthog:generate-app-url with /error-tracking/{id} when the issue ID is known. Do not build PostHog links by hand.

Report the result

For each checked occurrence, show the error time, IDs used for the join, and links to the issue, replay, and retained trace when available. Summarize matching logs with their time and severity. Exclude secrets and unneeded personal data from log excerpts.

Name each gap separately: no trace ID on the event, trace ID but no retained span, no session attribute on spans, no recording, or no logs with that trace ID in the checked window. A session-matched span is a candidate until an exact trace or span ID connects it to the error. State any row limit or time range that kept the search from being complete. If a lookup fails, say that its result is unknown instead of reporting that no matching data exists.