Monitors and Alerts

Use monitors to reduce downtime and controls costs by notifying your team when a specific condition has been met.


Monitors can be used to track one or more timeseries. The general idea is to create 1) a set of filters / groupings to identify specific data .2) conditions that must be met .and 3) mechanisms by which to deliver an alert when conditions are met.

You can create a monitor from the Monitors Tab in the UI.

Select Add Monitor in the upper right

General / Specifying data

The interface below can be used to specify the timeseries that apply to a monitor. They are similar to the ones in the Dashboards page and can be used to select a metric, source grouping, source filters, destination groupings, and destination filters. The specific timeseries selected can be previewed on the right hand side of the page. Note that multiple timeseries may be selected by the filter criteria so a single monitor can apply a large portion of the overall dataset without additional configuration.


The conditions identify a set of thresholds after which to generate an alert. Two thresholds may be specified for "Alert" and "Warning".

The condition clause takes the form: When the <aggregation> of the metric is <comparison operator> the threshold over <timerange>.

  • Aggregations include sum, mean, min and max so they summarize the timeseries in the timerange to a scalar value.

  • Comparisons can be above or below.

  • Timeranges specify the block of time to use for the aggregation and comparison. Windows as short as 30 seconds or as long as 1 day can be configured.


Notifications for the monitor can be sent to Slack or Pagerduty. Note these must be configured in "Settings" tab in Flowmill.

Also, the "Events" tab in Flowmill provides a running list of alerts and warnings that have been generated. Note that each monitor can generate multiple alerts if it monitors multiple timeseries.