Skip to content
Demo: example dataThe Acme Demo workspace and its providers are fictional. Nothing you do here is saved or sent.Get your free API key
GridRouterhome
Sign up free

Alerts and drains

Rules on error rate, latency, spend, billing mismatches and security events, sent to Slack, PagerDuty, email or a webhook. Every call can also stream to your own log tools.

Open alerts

Alerts that are firing now. Acknowledge one to say you're on it (PagerDuty and Opsgenie are told too); it resolves on its own once the metric recovers.

  • CriticalError rate 31% on Example Eta

    31 of 100 calls failed in 5 minutes (threshold 25%). Most were 401: check the vendor key.

    Started Sep 30, 09:00 UTC · seen 4 times

    Timeline (3)
    1. Sep 30, 09:00 UTC Error rate 31% (threshold 25%).
    2. Sep 30, 09:00 UTC Sent to #gtm-oncall.
    3. Sep 30, 09:00 UTC Paged PagerDuty: data platform.

Alert rules

Each rule watches one metric in a window and opens an alert when it crosses the threshold. Repeats are grouped into the open alert; after it resolves, the rule stays quiet for its cooldown.

  • Error rate above 25%CriticalFiringDefault

    Alerts when 25% or more of calls for example-eta fail over 5 minutes (once there are at least 20 calls).

    Failed share of calls per provider, over 5 minutes.

    Checked on every call · Notifies #gtm-oncall, PagerDuty: data platform, On-call email · dashboard toast · last 31% at Sep 30, 15:59 UTC

  • Spend over $5 in 5 minutesWarningOKDefault

    Alerts when calls cost $5.00 or more within 5 minutes.

    Checked on every call · Notifies #gtm-oncall, PagerDuty: data platform, On-call email · dashboard toast · last $1.24 at Sep 30, 15:59 UTC

  • Billing mismatch on any providerWarningOK

    Alerts when 1 or more calls are billed against the vendor's published rule within an hour.

    A provider charged for a call its published rule says is free.

    Checked every 5 minutes · Notifies #gtm-oncall, PagerDuty: data platform, On-call email · dashboard toast · last 0 at Sep 30, 15:59 UTC

  • Key used from a new IPInfoOKDefault

    Alerts when a key is used from an IP address it hasn't used before.

    Raised when it happens · Notifies On-call email · dashboard toast

History

Resolved alerts, newest first, with who was notified and when. Kept for the last 2,000 alerts.

  • WarningSpend $5.40 in 5 minutes on Research agent

    Key Research agent spent $5.40 in 5 minutes (limit $5.00). The key's daily cap still applies.

    Started Sep 29, 14:00 UTC · resolved Sep 29, 15:00 UTC

    Timeline (0)

    1. InfoNew IP for Acme outbound (production)

      First call from 203.0.113.77 (DE). Revoke the key if you don't recognize it.

      Started Sep 27, 16:00 UTC · resolved Sep 27, 16:12 UTC

      Timeline (0)

      Log drains

      Stream every call, run, job and alert to your own tools as it happens: batched, retried with backoff, and dead-lettered after eight failed attempts. Credentials are encrypted and write-only.

      • Datadog (production)DatadogHealthy

        datadoghq.com · gridrouter · every event

        Events delivered
        184,220
        Failed batches
        2
        Dead-lettered
        0
        Last success
        Sep 30, 15:59 UTC

        Last error (Sep 24, 16:00 UTC): 503 from the receiver

      • Slack alerts relayHTTPS webhookDegraded

        https://hooks.acme-demo.example/gridrouter · alert

        Events delivered
        312
        Failed batches
        9
        Dead-lettered
        1
        Last success
        Sep 30, 14:00 UTC

        Last error (Sep 30, 13:00 UTC): Timed out after 10 s