Skip to content
Reloqui Docs

Rollbacks & incidents

When something goes wrong after a release, Reloqui keeps the record honest.

Roll back a published release

Section titled “Roll back a published release”

On a published release, click Roll back and provide a reason:

POST /api/v1/releases/{id}/rollback
{ "reason": "Elevated checkout error rate in production; reverting to 2.2.0." }

The release moves to rolled_back, emits release.rolled_back (fanning out to your channels), and records the reason in the audit trail. Only published releases can be rolled back — the guard returns 422 otherwise.

Rollback evidence & confidence

Section titled “Rollback evidence & confidence”

A deployment.rolled_back event in a release’s window is real evidence: it’s counted (counts.rollbacks), raises an exception, and applies a −0.20 confidence deduction — so a release that had to be rolled back reflects that in its score.

Manual deployment import

Section titled “Manual deployment import”

If your deploy tooling has no connector yet, record a deployment (or rollback) by hand — it becomes the same canonical evidence a webhook would produce:

POST /api/v1/evidence/deployments
{ "projectId": "proj_…", "serviceId": "svc_…", "environment": "production",
  "status": "succeeded", "version": "2.3.0", "commitSha": "a1b2c3d4", "actor": "argocd" }

Use "status": "rolled_back" to record a rollback deployment.

Incident correlation

Section titled “Incident correlation”

Connect an incident source (PagerDuty, Opsgenie) or post to the generic incident webhook. Incidents that fall in a release’s lifecycle window for the same service are correlated to it and shown on the Release 360.

Incidents are deliberately correlation-only — they describe operations, not release content, so they are never scored and never counted as evidence. They give you the “what happened after we shipped” picture without distorting the confidence of “what shipped”.

GET /api/v1/releases/{id}/incidents

Environment drift

Section titled “Environment drift”

Reloqui also detects environment drift — when the latest artifact differs across environments for a service — on the Operations page, so you can spot a stale or mismatched rollout.

Section titled “Related”