Set up alerts for Openflow¶
Openflow alerts are Snowflake alerts that are based on Openflow telemetry. Snowflake provides alert templates for Openflow connectors and runtimes. An alert created from a template regularly queries the event table and notifies you when its condition is met. The template defines the condition and the action, and you supply only the settings: connectors, runtimes, or deployments to monitor, and the notification type and target.
Note
Templates come in separate gen 1 and gen 2 versions. Each version monitors only connectors and runtimes of its own generation, so to monitor both, create an alert for each version.
Before you begin¶
- Openflow must write telemetry to an event table. See Prerequisites for Openflow alerts.
- To send notifications, create a notification integration. See Setting up notifications for alerts.
- Use a role that can create and monitor alerts. See Creating a role for accessing alerts in Snowsight.
Create an alert from a template¶
In Snowsight, choose a template from the OPENFLOW template group. See Creating a new alert. To create the alert with SQL, see Creating an alert from a template with SQL.
Connector templates¶
The following templates monitor connectors. Each alert notification identifies the connector and links to its details page in Snowsight.
| Template | Description | Settings |
|---|---|---|
| Connector backpressure (object count) | Fires when a connection of a connector applies back pressure based on the number of queued FlowFiles for at least the configured fraction of the lookback window. Can be configured for the account, a deployment, a runtime, or a connector. |
|
| Connector backpressure (bytes) | Fires when a connection of a connector applies back pressure based on the size of queued FlowFiles for at least the configured fraction of the lookback window. Can be configured for the account, a deployment, a runtime, or a connector. |
|
| High queued count alert | Fires when the number of FlowFiles queued in a connection of a connector exceeds a fraction of the back pressure threshold for at least the configured fraction of the lookback window. Can be configured for the account, a deployment, a runtime, or a connector. |
|
| High queued bytes alert | Fires when the size of the FlowFiles queued in a connection of a connector exceeds a fraction of the back pressure threshold for at least the configured fraction of the lookback window. Can be configured for the account, a deployment, a runtime, or a connector. |
|
| High error rate alert | Fires when the components of a connector log at least the threshold number of errors during the lookback window. Can be configured for the account, a deployment, a runtime, or a connector. |
|
| No data alert | Fires when a connector doesn’t receive data or doesn’t send data during the lookback window. Can be configured for the account, a deployment, a runtime, or a connector. | Lookback window (hours): Default: 4. |
| Openflow connector health alert | Fires when a connector becomes unhealthy. The definition of health differs across connector types. Can be configured for the account, a deployment, a runtime, or a connector. | Lookback window (hours): Default: 1. |
Runtime templates¶
The following templates monitor runtimes. Each alert notification identifies the runtime.
| Template | Description | Settings |
|---|---|---|
| Out of memory alert | Fires when a runtime logs at least the threshold number of OutOfMemoryError errors during the lookback window. Can be configured for the account, a deployment, or a runtime. |
|
| High runtime storage usage alert | Fires when the disk usage of a runtime reaches the threshold. Can be configured for the account, a deployment, or a runtime. | Storage usage threshold: Fraction of the storage capacity. Default: 0.85. |
Connector-specific templates¶
Some connectors have additional alerts that are based on metrics specific to them. For those templates, see the monitoring page of the connector: