695ac91290
What is this feature? This PR introduces a new alert rule configuration option, keep_firing_for (Prometheus documentation). keep_firing_for prevents alerts from resolving immediately after the alert condition returns to normal. Instead, they transition into a "Recovering" state and are not considered resolved by the Alertmanager. Once the recovery period ends (or after the next evaluation if it is bigger than keep_firing_for), the alert transitions to "Normal" if it doesn't start alerting again: Before +----------+ +----------+ | Alerting |---->| Normal | +----------+ +----------+ ----- After +----------+ +------------+ +----------+ | Alerting |----->| Recovering |---->| Normal | +----------+ +------------+ +----------+ Why do we need this feature? This feature prevents flapping alerts by adding a recovery period. This helps avoid false resolutions caused by brief alert
110 lines
2.7 KiB
Go
110 lines
2.7 KiB
Go
package models
|
|
|
|
import (
|
|
"errors"
|
|
"fmt"
|
|
"testing"
|
|
|
|
"github.com/stretchr/testify/require"
|
|
)
|
|
|
|
func TestInstanceStateType_IsValid(t *testing.T) {
|
|
testCases := []struct {
|
|
instanceType InstanceStateType
|
|
expectedValidity bool
|
|
}{
|
|
{
|
|
instanceType: InstanceStateFiring,
|
|
expectedValidity: true,
|
|
},
|
|
{
|
|
instanceType: InstanceStateNormal,
|
|
expectedValidity: true,
|
|
},
|
|
{
|
|
instanceType: InstanceStatePending,
|
|
expectedValidity: true,
|
|
},
|
|
{
|
|
instanceType: InstanceStateNoData,
|
|
expectedValidity: true,
|
|
},
|
|
{
|
|
instanceType: InstanceStateError,
|
|
expectedValidity: true,
|
|
},
|
|
{
|
|
instanceType: InstanceStateRecovering,
|
|
expectedValidity: true,
|
|
},
|
|
{
|
|
instanceType: InstanceStateType("notAValidInstanceStateType"),
|
|
expectedValidity: false,
|
|
},
|
|
}
|
|
|
|
for _, tc := range testCases {
|
|
t.Run(buildTestInstanceStateTypeIsValidName(tc.instanceType, tc.expectedValidity), func(t *testing.T) {
|
|
require.Equal(t, tc.expectedValidity, tc.instanceType.IsValid())
|
|
})
|
|
}
|
|
}
|
|
|
|
func buildTestInstanceStateTypeIsValidName(instanceType InstanceStateType, expectedValidity bool) string {
|
|
if expectedValidity {
|
|
return fmt.Sprintf("%q should be valid", instanceType)
|
|
}
|
|
return fmt.Sprintf("%q should not be valid", instanceType)
|
|
}
|
|
|
|
func TestValidateAlertInstance(t *testing.T) {
|
|
testCases := []struct {
|
|
name string
|
|
orgId int64
|
|
uid string
|
|
currentState InstanceStateType
|
|
err error
|
|
}{
|
|
{
|
|
name: "fails if orgID is empty",
|
|
orgId: 0,
|
|
uid: "validUid",
|
|
currentState: InstanceStateNormal,
|
|
err: errors.New("alert instance is invalid due to missing alert rule organisation"),
|
|
},
|
|
{
|
|
name: "fails if uid is empty",
|
|
orgId: 1,
|
|
uid: "",
|
|
currentState: InstanceStateNormal,
|
|
err: errors.New("alert instance is invalid due to missing alert rule uid"),
|
|
},
|
|
{
|
|
name: "fails if current state is not valid",
|
|
orgId: 1,
|
|
uid: "validUid",
|
|
currentState: InstanceStateType("notAValidType"),
|
|
err: errors.New("alert instance is invalid because the state 'notAValidType' is invalid"),
|
|
},
|
|
{
|
|
name: "ok if validated fields are correct",
|
|
orgId: 1,
|
|
uid: "validUid",
|
|
currentState: InstanceStateNormal,
|
|
err: nil,
|
|
},
|
|
}
|
|
|
|
for _, tc := range testCases {
|
|
t.Run(tc.name, func(t *testing.T) {
|
|
instance := AlertInstanceGen(func(instance *AlertInstance) {
|
|
instance.AlertInstanceKey.RuleOrgID = tc.orgId
|
|
instance.AlertInstanceKey.RuleUID = tc.uid
|
|
instance.CurrentState = tc.currentState
|
|
})
|
|
|
|
require.Equal(t, tc.err, ValidateAlertInstance(*instance))
|
|
})
|
|
}
|
|
}
|