public/grafana
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Alerting: Ensure errors cleared when Alerting after error (#105246)

Browse Source 
When a rule configured with `ExecErrState` state of `Alerting`, has an instance which is Alerting then has a data source error, then successfully evaluates and continues to be Alerting, the cached instance keeps the error cached until it is no longer firing.

This is unexpected and leads to misleading results.
This commit is contained in:
Moustafa Baiou authored and GitHub committed 2025-06-04 12:16:14 +02:00
1 parent 5bfcbc1f47
commit 0ce086bd2e
3 files changed
+202
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

pkg/services/ngalert/state/manager_private_test.go
+199
View File
@@ -3779,6 +3779,205 @@ func TestProcessEvalResults_StateTransitions(t *testing.T) {
},
},
},
{
desc: "t1[1:alerting] t2[QueryError] t3[1:alerting] and 'for'=1 at t2,t3",
ruleMutators: []ngmodels.AlertRuleMutator{ngmodels.RuleMuts.WithForNTimes(1)},
results: map[time.Time]eval.Results{
t1: {
newResult(eval.WithState(eval.Alerting), eval.WithLabels(labels1), eval.WithValues(map[string]eval.NumberValueCapture{"A": {Var: "A", Value: util.Pointer(1.0)}})),
},
t2: {
newResult(eval.WithError(datasourceError)),
},
t3: {
newResult(eval.WithState(eval.Alerting), eval.WithLabels(labels1), eval.WithValues(map[string]eval.NumberValueCapture{"A": {Var: "A", Value: util.Pointer(1.0)}})),
},
},
expectedTransitions: map[ngmodels.ExecutionErrorState]map[time.Time][]StateTransition{
ngmodels.ErrorErrState: {
t2: {
{
PreviousState: eval.Pending,
State: &State{
Labels: labels["system + rule + labels1"],
State: eval.Pending,
LatestResult: newEvaluationWithValues(t1, eval.Alerting, map[string]float64{"A": 1.0}),
StartsAt: t1,
EndsAt: t1.Add(ResendDelay * 4),
LastEvaluationTime: t1,
Values: map[string]float64{"A": 1.0},
},
},
{
PreviousState: eval.Normal,
State: &State{
CacheID: labels["system + rule"].Fingerprint(),
Labels: labels["system + rule + datasource-error"],
State: eval.Error,
Error: datasourceError,
LatestResult: newEvaluation(t2, eval.Error),
StartsAt: t2,
EndsAt: t2.Add(ResendDelay * 4),
LastEvaluationTime: t2,
LastSentAt: &t2,
Annotations: mergeLabels(baseRule.Annotations, data.Labels{
"Error": datasourceError.Error(),
}),
},
},
},
t3: {
{
PreviousState: eval.Pending,
State: &State{
Labels: labels["system + rule + labels1"],
State: eval.Alerting,
Error: nil,
LatestResult: newEvaluationWithValues(t3, eval.Alerting, map[string]float64{"A": 1.0}),
StartsAt: t3,
FiredAt: &t3,
EndsAt: t3.Add(ResendDelay * 4),
LastEvaluationTime: t3,
LastSentAt: &t3,
Annotations: baseRule.Annotations,
Values: map[string]float64{"A": 1.0},
},
},
{
PreviousState: eval.Error,
State: &State{
CacheID: labels["system + rule"].Fingerprint(),
Labels: labels["system + rule + datasource-error"],
State: eval.Error,
Error: datasourceError,
LatestResult: newEvaluation(t2, eval.Error),
StartsAt: t2,
EndsAt: t2.Add(ResendDelay * 4),
LastEvaluationTime: t2,
LastSentAt: &t2,
Annotations: mergeLabels(baseRule.Annotations, data.Labels{
"Error": datasourceError.Error(),
}),
},
},
},
},
ngmodels.AlertingErrState: {
t2: {
{
PreviousState: eval.Pending,
State: &State{
Labels: labels["system + rule + labels1"],
State: eval.Alerting,
StateReason: eval.Error.String(),
Error: datasourceError,
Annotations: datasourceErrorAnnotations,
LatestResult: newEvaluationWithValues(t2, eval.Error, map[string]float64{"A": -1}),
StartsAt: t2,
EndsAt: t2.Add(ResendDelay * 4),
FiredAt: &t2,
LastEvaluationTime: t2,
LastSentAt: &t2,
Values: map[string]float64{"A": -1},
},
},
},
t3: {
{
PreviousStateReason: eval.Error.String(),
PreviousState: eval.Alerting,
State: &State{
Labels: labels["system + rule + labels1"],
State: eval.Alerting,
Error: nil,
Annotations: baseRule.Annotations,
LatestResult: newEvaluationWithValues(t3, eval.Alerting, map[string]float64{"A": 1.0}),
StartsAt: t2,
EndsAt: t3.Add(ResendDelay * 4),
FiredAt: &t2,
LastEvaluationTime: t3,
LastSentAt: &t2,
Values: map[string]float64{"A": 1.0},
},
},
},
},
ngmodels.OkErrState: {
t2: {
{
PreviousState: eval.Pending,
State: &State{
Labels: labels["system + rule + labels1"],
State: eval.Normal,
StateReason: eval.Error.String(),
Annotations: datasourceErrorAnnotations,
LatestResult: newEvaluationWithValues(t2, eval.Error, map[string]float64{"A": float64(-1)}),
StartsAt: t2,
EndsAt: t2,
LastEvaluationTime: t2,
Values: map[string]float64{"A": float64(-1)},
},
},
},
t3: {
{
PreviousState: eval.Normal,
PreviousStateReason: eval.Error.String(),
State: &State{
Labels: labels["system + rule + labels1"],
State: eval.Pending,
Annotations: baseRule.Annotations,
LatestResult: newEvaluationWithValues(t3, eval.Alerting, map[string]float64{"A": 1.0}),
StartsAt: t3,
EndsAt: t3.Add(ResendDelay * 4),
LastEvaluationTime: t3,
Values: map[string]float64{"A": 1.0},
},
},
},
},
ngmodels.KeepLastErrState: {
t2: {
{
PreviousState: eval.Pending,
State: &State{
Labels: labels["system + rule + labels1"],
State: eval.Alerting,
StateReason: ngmodels.ConcatReasons(eval.Error.String(), ngmodels.StateReasonKeepLast),
Annotations: datasourceErrorAnnotations,
LatestResult: newEvaluationWithValues(t2, eval.Error, map[string]float64{"A": float64(-1)}),
StartsAt: t2,
EndsAt: t2.Add(ResendDelay * 4),
FiredAt: &t2,
LastEvaluationTime: t2,
LastSentAt: &t2,
Values: map[string]float64{"A": float64(-1)},
},
},
},
t3: {
{
PreviousState: eval.Alerting,
PreviousStateReason: ngmodels.ConcatReasons(eval.Error.String(), ngmodels.StateReasonKeepLast),
State: &State{
Labels: labels["system + rule + labels1"],
State: eval.Alerting,
StateReason: "",
Error: nil,
LatestResult: newEvaluationWithValues(t3, eval.Alerting, map[string]float64{"A": 1.0}),
StartsAt: t2,
FiredAt: &t2,
EndsAt: t3.Add(ResendDelay * 4),
LastEvaluationTime: t3,
LastSentAt: &t2,
Annotations: baseRule.Annotations,
Values: map[string]float64{"A": 1.0},
},
},
},
},
},
},
{
desc: "t1[1:normal] t2[QueryError] at t2",
results: map[time.Time]eval.Results{
pkg/services/ngalert/state/manager_test.go
+1
View File
@@ -1013,6 +1013,7 @@ func TestProcessEvalResults(t *testing.T) {
},
},
{
// TODO(@moustafab): figure out why this test doesn't fail as is
desc: "classic condition, execution Error as Error (alerting -> query error -> alerting)",
alertRule: baseRuleWith(m.WithErrorExecAs(models.ErrorErrState)),
expectedAnnotations: 3,
pkg/services/ngalert/state/state.go
+2
View File
@@ -401,6 +401,8 @@ func resultAlerting(state *State, rule *models.AlertRule, result eval.Result, lo
case eval.Alerting:
prevEndsAt := state.EndsAt
state.Maintain(rule.IntervalSeconds, result.EvaluatedAt)
// explicitly clear errors
state.Error = nil
logger.Debug("Keeping state",
"state",
state.State,