Reading logs
Each Aurral writes logs/aurral.log in its data folder, one JSON object per line, with the newest line at the end. A dev worktree writes backend/data/logs/aurral.log. Worker lines land in the same file, tagged with proc and the job that wrote them. Lines from a request share a req id.
| Field | Meaning |
|---|---|
id |
Order in the file. Use it as the paging cursor |
level |
debug, info, warn, or error |
cat |
The area, such as library, jobs, http, or browser |
fp |
The same value for repeats of one problem, ignoring ids and numbers |
data, err |
Details, and an error with its stack frames. Secrets are masked |
Follow warnings and errors on a local install:
tail -F backend/data/logs/aurral.log | jq -c 'select(.level == "warn" or .level == "error")'On a remote install, use the API key. Start with the repeated problems from the last day, then read one in full:
curl -H "X-Api-Key: $KEY" "https://aurral.example.com/api/logs/problems"curl -H "X-Api-Key: $KEY" "https://aurral.example.com/api/logs?fp=<fp>&limit=20"/api/logs also filters by req and job, so you can follow one failure from start to end. Page back with before=<id>, and poll for new lines with after=<id>. See the endpoint reference.