Skip to main content

Troubleshooting

Three layers:

  1. Error codes — every E* / boundary / quota code the runtime emits.
  2. Runbook index — every runbook_url from the shipped Prometheus alert rules.
  3. Flowcharts — scoped to the two high-value first-week questions:

My install failed

If none of those branches match, open an issue with the full installer output.

Logging

When reproducing an issue, capture the daemon's per-agent log files (~/.declaragent/logs/<agent>.log, tail live with declaragent logs -f) and attach them to the report. (There is no log-level env var today — the runtime logs at its default verbosity.)

Still stuck?