During planned maintenance, known incidents, or recurring low-priority alerts, notification noise can overwhelm your team and obscure actionable alerts. Silence policies filter and converge alerts that match specific rules, reducing the likelihood of alert storms, so your team only receives notifications that require attention.
Silence policies take precedence over notification policies. Alert events that match a silence policy are silenced and are not matched by any notification policy.
How silence policies work
A silence policy consists of two parts:
| Component | Purpose | Example |
|---|---|---|
| Matching rules | Define which alerts to silence, using tags and conditions | Silence all P4 alerts from cluster prod-east |
| Effective time | Define when the policy is active | Every Saturday and Sunday, or a one-time maintenance window |
Combine these two components to cover common scenarios:
Planned maintenance -- Silence alerts from specific clusters or services during a scheduled maintenance window using a custom time period.
Recurring low-priority noise -- Silence P4 or Default severity alerts on weekends using a recurring schedule.
Known incident suppression -- Silence alerts related to a known issue by matching the
alertnameor_aliyun_arms_involvedObject_nametag, with a persistent or custom time period.
Create a silence policy
Log on to the Managed Service for OpenTelemetry console.
In the left-side navigation pane, choose Alert Management > Alarm policy.
Click the Silence Policies tab, then click Create Silence Policy.
Enter a name for the silence policy.
In the Silent Event Matching Rule section, configure the rules that determine which alerts to silence.
Select a data source.
To filter alerts from a specific source, select that source.
To filter all alerts regardless of source, select No Preset Source.
Define one or more expressions using existing tags or custom tags.
Available tags for matching expressions:
Tags defined in alert rule expressions. For details on creating tags in Managed Service for Prometheus, see Create an alert rule for a Prometheus instance.
Default ARMS tags, listed in the following tables.
Common fields
Tag Description alertnameAlert name. clusternameKubernetes cluster name. severitySeverity level. Valid values: P1, P2, P3, P4, Default. namespaceKubernetes namespace. pod_nameKubernetes pod name. Preset system fields
Tag Description _aliyun_arms_integration_nameIntegration name specified as the data source. Default: ARMS-DEFAULT. _aliyun_arms_involvedObject_idID of the object that triggers the alert. _aliyun_arms_involvedObject_nameName of the object that triggers the alert. _aliyun_arms_region_idARMS region ID. _aliyun_arms_alert_rule_idAlert rule ID. _aliyun_arms_alert_typeAlert rule type. Valid values: 101 (Prometheus alert), 5 (Application Monitoring alert), 4 (Browser Monitoring alert). Combine conditions and rules to define matching logic.
Action Logic Use case Click + Add Condition AND -- the alert must meet all conditions within the same rule Silence alerts that are both P4 severity AND from a specific cluster Click + Add Rule OR -- the alert must meet any one of the rules Silence alerts that match rule A OR rule B
Set the effective time.
Option Behavior Example Persistent The policy stays active indefinitely. Suppress known noisy alerts with no end date. Recurring The policy activates on a repeating schedule. Select Every Day or Weekly, then click Create. Suppress non-critical alerts every weekend. Custom Time Period The policy activates during a specific time window. Click Create and define the period. Suppress alerts during a two-hour maintenance window on March 15. Click Save.
Manage silence policies
On the Silence Policies tab, perform any of the following operations:
| Operation | Steps |
|---|---|
| Edit | Click the policy name, or click Edit in the Actions column. Modify the settings and click Save. |
| Enable or disable | Toggle the switch in the Status column. |
| Delete | Click Delete in the Actions column, then click OK to confirm. |
| Copy | Click Copy in the Actions column to duplicate the policy. |