Executions and Logs
Every invocation writes an execution row plus logs and metrics. Use dashboard and CLI to investigate failures quickly.
Step-by-Step Guide
1.Execution status model
Execution records store lifecycle status and runtime metrics.
- Statuses:
ok,fn_error,infra_error,limit_exceeded. - Metrics include wall time, CPU time, memory peak, egress bytes, and subrequest count.
- Composition metadata includes
parentExecutionIdandcallDepth.
2.Filtering and list APIs
Execution list APIs support filters by status, trigger kind, and optional date range/cursor windows.
- Dashboard filtering currently exposes status + trigger chips.
- Backend APIs also accept
from,to, andcursorfor pagination windows.
3.Logs and ingestion
Runtime/tail ingestion writes structured logs and runtime metrics to control-plane storage.
- Dashboard execution detail consumes live updates through server APIs.
- CLI logs route:
/api/cli/executions/logs(latest or by execution id). - Structured log fields are preserved with each log line.
4.Practical debug workflow
Use execution detail + logs together when investigating failures.
- Check trigger kind and parent execution linkage first.
- Inspect
errorMessageand log sequence around failure timestamps. - Use filtered execution views to isolate regressions after deploy.