ourfault
Reads every error your services emit, decides whether it is your fault, and files the issue.
How it works
-
Connect with OpenTelemetry
Point your OpenTelemetry SDK or Collector at the ingest endpoint with a workspace token. Records that carry an exception are read; the rest are counted and let go.
-
We triage, with evidence from your deployments
Each new failure is set against your recent deployments on GitHub and judged once: a fault in your code, someone else's, or too close to call. Failures seen before are counted, not judged again.
-
A proposed issue lands in your repository
When the fault is yours, ourfault reads the files the stack trace names and opens an issue with the root cause, marked as a proposal. Close calls are held, not filed.
What is kept
Kept
- A hash for each distinct failure
- Counts of records and occurrences
- Service names, repositories and times
- The verdict, its probability, and the issue it filed
Not kept
- Log and message bodies
- Exception messages and stack traces
- Attribute values
- The source files read to find the cause
A new failure's message and trace pass through the pipeline while it is judged, a buffer declared with every stored field in the data retention contract. A filed issue holds the trace and the root cause, in your repository, under your control.