Errors & rate limits

What the API returns when things go wrong, and what to expect under throttling or backend pressure.

Error envelope

All 4xx and 5xx responses share the same shape: a JSON body with a single error field.

json

{ "error": "Rate limit exceeded. Maximum 10 pings per minute." }

Status codes

Field	Type	Description
`200`	`OK`	Request succeeded — a heartbeat ping, or an alert filtered by rules. Inspect the response body for the structured result.
`201`	`Created`	The trigger endpoint created an alert, reopened one, or added an occurrence.
`400`	`Bad request`	Validation failed — required field missing, invalid enum, or payload size exceeded.
`401`	`Unauthorized`	Missing or invalid X-Webhook-Signature.
`403`	`Forbidden`	Team subscription is inactive or expired.
`404`	`Not found`	Unknown ping URL or webhook URL, or the resource is paused/disabled.
`429`	`Too many requests`	Rate limit or monthly usage quota exceeded. The body explains which.
`5xx`	`Server error`	Unexpected failure on our end. Retry with backoff; if persistent, contact support.

Rate limits

Pings — 10 requests per minute per ping URL. Bursting above that returns 429 for the rest of the minute.
Alerts — limited primarily by the webhook's configured rate cap and the team monthly quota. Identical alerts are deduplicated server-side, so high-volume failure modes don't immediately hit the limit.

Monthly quotas

Each plan includes a monthly cap on heartbeats and alerts. When the cap is exhausted, further requests return 429 with a reason like "Monthly heartbeat quota exceeded". Usage resets at the start of each billing period; usage and cap are visible in Settings → Usage.

Fail-safe quota checks

Quota checks return allowed when our analytics backend is briefly unreachable — by design. A transient outage on our side should never drop your pings or alerts on the floor.

Retry guidance

401 / 403 / 404 — do not retry. The request is fundamentally rejected; fix the cause.
429 — retry with exponential backoff. The SDKs do not retry automatically, so add retry logic if your workload warrants it.
5xx — retry with exponential backoff, capped at a few attempts. Beyond that, surface the failure to your logs.

Errors from the SDKs

Both SDKs surface failures as OpShiftError with statusCode / status_code. Network failures (DNS, connection refused, timeout) surface with code 0 so you can distinguish "couldn't reach the API" from "the API said no".

typescript

try {
  await opshift.alert("deploy-failures", { title: "..." });
} catch (error) {
  if (error instanceof OpShiftError) {
    if (error.statusCode === 0) {
      // network-level failure: queue locally, retry later
    } else if (error.statusCode === 429) {
      // back off
    } else {
      // unrecoverable: surface to your log pipeline
    }
  }
  throw error;
}

Trigger alert