Alerting CLI

In addition to being able to control alerting rules via UI, you can manage alerting rules in the databand using the command line.

Make sure, prior to running, that you have DATABAND_URL and DATABAND_ACCESS_TOKEN set to the correct values. You can read more about how to do that on the Connecting DBND to Databand page.

⚠️

Supported Alert Types

This version of CLI only supports individual pipeline alerts

List

This command lists all existing alert definitions.

Paramеters

• —pipeline or -p Name of the pipeline on which to show alerts.

List Example

dbnd alerts list --pipeline my_pipeline

Create

Create or update the alert definition

Parameters:

• --update or uid or -u UID of an existing alert to update. Used in case of update.
• --pipeline or -p Pipeline name on which alert will be created.
• --pipeline-id Pipeline id on which alert will be created.
• —severity or -s Alert severity, one of the following values are supported - CRITICAL, HIGH, MEDIUM, LOW.

Run duration of pipeline in seconds.

`run-duration`

Parameters

• —op or -o Operator one of ==, !=, <=, >=, <, >, ML, range.
  ML operator requires two additional params:
• --look-back or -lb Look Back - number of runs used to train anomaly alerts.
• --sensitivity” or -sn` Sensitivity of anomaly detection alerts.

range operator requires two additional params:

• --baseline or -b Base number from which range alert will be measured.
• --range or -r Percent of baseline to include in range alerts.
• --value or -v Numerical value of the run-duration. Required in numerical operators ==, !=, <=, >=, <, >. Do not provide for ML or range alert type.

Run Duration Alert Examples:

dbnd alerts create --pipeline my_pipeline --severity MEDIUM run-duration --op '==' -v 100
dbnd alerts create --pipeline my_pipeline --severity HIGH run-duration --op ML --look-back 10 --sensitivity 4
dbnd alerts create --pipeline my_pipeline --severity HIGH run-duration --op range --baseline 10 --range 4

State of the run to alert on.

 `run-state`

Parameters:

• --value or -v String value of the run state, one of running, success, failed, shutdown, cancelled. Note: operator is hardcoded to “==” and shouldn’t be provided.
  

Example for Run State Alert

dbnd alerts create --pipeline my_pipeline --severity MEDIUM run-state -v failed

Duration in seconds from the last successful run from a pipeline.

ran-last-x-seconds
Parameters:

• --value or -v Numerical value of last ran seconds. Note that the operator is hardcoded to > and shouldn’t be provided.

Example

dbnd alerts create --pipeline my_pipeline --severity CRITICAL ran-last-x-seconds -v 120

State of the task run to alert on.

`task-state`

Parameters

*—value or -v - String value of task state, one of scheduled, queued, running, success, failed, cancelled, shutdown, upstream_failed, skipped, up_for_retry, removed. Note that operator is hardcoded to “==” and shouldn’t be provided.

Example

dbnd alerts create --pipeline my_pipeline --severity MEDIUM --task my_pipeline_task task-state -v cancelled

Alert on metric value

`custom-metric`

Parameters

• --metric-name or -m Custom metric name to alert.
• --str-value Boolean param indicating whether metric has a float of string values.
• --op or -o Operator of the metric. One of ==, !=, <=, >=, <, >, ML, range. if --str-value set to false, otherwise operator can take only take == or != values.
• --value or -v. Numerical value of the metric corresponding ==, !=, <=, >=, <, > operators, Or String value corresponding ==, != operators based on --str-value param. Required only on the listed above operators except ML or range.

ML operator requires two additional params:

• --look-back or -lb Look Back number of runs used to train anomaly alerts.
• --sensitivity or -sn Sensitivity of anomaly detection alerts.

Range operator requires two additional params:

• --baseline or -bl Base number from which range alert will be measured.
• --range or -r Percent of baseline to include in range alerts.
  

Example

dbnd alerts create --pipeline my_pipeline --severity CRITICAL --task my_pipeline_task custom-metric --metric-name my_metric --op '<=' -v 100 --str-value false

Delete

Delete individual alert definition, all alert definitions for a specific pipeline, or all existing alert definitions.

Parameters

• --uid or -u Alert UID to delete.
• --wipe Delete all alerts.
• --pipeline or -p Pipeline name on which alerts will be deleted. Can be used with --name or -n indicating alert custom name.

Delete Examples:

dbnd alerts delete --uid my_alert_uid
dbnd alerts delete --pipeline my_pipeline
dbnd alerts delete --wipe