Scalingo Metric Alert Notification Frequency and Duration Constraints

Answer

Scalingo’s metric‑alert engine fires as soon as a sample exceeds the threshold; it does not require the breach to persist for any number of minutes before sending a notification. Likewise, once an alert is in the firing state, the platform does not automatically suppress or delay subsequent notifications – every new sample that continues to be above the threshold will generate a fresh alert message.

Likely Explanation

• Alerts are evaluated per‑sample (typically every 30 s) and trigger immediately when the comparison becomes true.
• There is no UI field or API parameter for a “minimum duration” or a “cool‑down” period.
• Because the system treats each sample independently, an ongoing spike that stays above the threshold will produce a notification on every evaluation cycle.

Confirmed Facts (from current Scalingo documentation)

• Threshold value, comparison operator, and notification channels are configurable.
• No documented option for sustained‑breach duration or repeat‑interval.
• Alerts are sent immediately when the condition evaluates to true.

Steps for Your Specific Case

1. Navigate to the Alerts section of your Scalingo dashboard.
2. Open the alert you want to test and note the current threshold, comparison operator, and notification channels.
3. Generate a controlled spike in the relevant metric (e.g., run a CPU‑heavy job for a few minutes).
4. Observe the notification stream: you should receive a message on the first sample that exceeds the threshold, and then every subsequent sample that also exceeds it.
5. If you want to reduce noise, consider raising the threshold or adding a higher‑level filter (e.g., only notify when the value is above 90 % for more than 5 consecutive samples) in your own monitoring logic or by using a third‑party integration that supports de‑duplication.

What to Do If You Need a Cool‑Down

• Use a custom webhook that implements its own rate‑limiting logic.
• Set up a downstream service (e.g., a small Lambda or Cloud Function) that receives the webhook, tracks the last notification time per container, and only forwards a new alert if a configurable interval has passed.
• Alternatively, rely on Slack’s or email’s built‑in de‑duplication features if available.

Missing Diagnostic Detail

Have you inspected the alert configuration UI for any hidden fields that might allow a duration or repeat interval? If you see a field named something like duration or repeat_interval, that would change the recommendation.