VictoriaMetrics/app/vmalert/alerting.go
Roman Khavronenko ab10178c85
Vmalert compliance 2 (#2340)
* vmalert: split alert's `Start` field into `ActiveAt` and `Start`

The `ActiveAt` field identifies when alert becomes active for rules
with `for > 0`. Previously, this value was stored in field `Start`.

The field `Start` now identifies the moment alert became `FIRING`.

The split is needed in order to distinguish these two moments
in the API responses for alerts.

Signed-off-by: hagen1778 <roman@victoriametrics.com>

* vmalert: support specific moment of time for rules evaluation

The Querier interface was extended to accept a new argument
used as a timestamp at which evaluation should be made.

It is needed to align rules execution time within the group.

Signed-off-by: hagen1778 <roman@victoriametrics.com>

* vmalert: mark disappeared series as stale

Series generated by alerting rules, which were sent to remote write
now will be marked as stale if they will disappear on the next
evaluation. This would make ALERTS and ALERTS_FOR_TIME series
more precise.

Signed-off-by: hagen1778 <roman@victoriametrics.com>

* wip

Signed-off-by: hagen1778 <roman@victoriametrics.com>

* vmalert: evaluate rules at fixed timestamp

Before, time at which rules were evaluated was calculated
right before rule execution. The change makes sure
that timestamp is calculated only once per evalution round
and all rules are using the same timestamp.

It also updates the logic of resending of already resolved
alert notification.

Signed-off-by: hagen1778 <roman@victoriametrics.com>

* vmalert: allow overridin `alertname` label value if it is present in response

Previously, `alertname` was always equal to the Alerting Rule name. Now,
its value can be overriden if series in response containt the different value
for this label.

The change is needed for improving compatibility with Prometheus.

Signed-off-by: hagen1778 <roman@victoriametrics.com>

* vmalert: align rules evaluation in time

Now, evaluation timestamp for rules evaluates as if
there was no delay in rules evaluation. It means, that
rules will be evaluated at fixed timestamps+group_interval.
This way provides more consistent evaluation results and
improves compatibility with Prometheus,

Signed-off-by: hagen1778 <roman@victoriametrics.com>

* vmalert: add metric for missed iterations

New metric `vmalert_iteration_missed_total` will show
whether rules evaluation round was missed.

Signed-off-by: hagen1778 <roman@victoriametrics.com>

* vmalert: reduce delay before the initial rule evaluation in group

Signed-off-by: hagen1778 <roman@victoriametrics.com>

* vmalert: rollback alertname override

According to the spec:
```
The alert name from the alerting rule (HighRequestLatency from the example above) MUST be added to the labels of the alert with the label name as alertname. It MUST override any existing alertname label.
```

https://github.com/prometheus/compliance/blob/main/alert_generator/specification.md#step-3
Signed-off-by: hagen1778 <roman@victoriametrics.com>

* vmalert: throw err immediately on dedup detection

```
The execution of an alerting rule MUST error out immediately and MUST NOT send any alerts
or add samples to samples receiver if there is more than one alert with the same labels
```

https://github.com/prometheus/compliance/blob/main/alert_generator/specification.md#step-4
Signed-off-by: hagen1778 <roman@victoriametrics.com>

* vmalert: cleanup

Signed-off-by: hagen1778 <roman@victoriametrics.com>

* vmalert: use strings builder to reduce allocs

Signed-off-by: hagen1778 <roman@victoriametrics.com>
2022-04-01 12:03:41 +03:00

604 lines
17 KiB
Go

package main
import (
"context"
"fmt"
"hash/fnv"
"sort"
"strconv"
"sync"
"time"
"github.com/VictoriaMetrics/VictoriaMetrics/app/vmalert/config"
"github.com/VictoriaMetrics/VictoriaMetrics/app/vmalert/datasource"
"github.com/VictoriaMetrics/VictoriaMetrics/app/vmalert/notifier"
"github.com/VictoriaMetrics/VictoriaMetrics/app/vmalert/utils"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/prompbmarshal"
)
// AlertingRule is basic alert entity
type AlertingRule struct {
Type datasource.Type
RuleID uint64
Name string
Expr string
For time.Duration
Labels map[string]string
Annotations map[string]string
GroupID uint64
GroupName string
EvalInterval time.Duration
q datasource.Querier
// guard status fields
mu sync.RWMutex
// stores list of active alerts
alerts map[uint64]*notifier.Alert
// stores last moment of time Exec was called
lastExecTime time.Time
// stores the duration of the last Exec call
lastExecDuration time.Duration
// stores last error that happened in Exec func
// resets on every successful Exec
// may be used as Health state
lastExecError error
// stores the number of samples returned during
// the last evaluation
lastExecSamples int
metrics *alertingRuleMetrics
}
type alertingRuleMetrics struct {
errors *utils.Gauge
pending *utils.Gauge
active *utils.Gauge
samples *utils.Gauge
}
func newAlertingRule(qb datasource.QuerierBuilder, group *Group, cfg config.Rule) *AlertingRule {
ar := &AlertingRule{
Type: group.Type,
RuleID: cfg.ID,
Name: cfg.Alert,
Expr: cfg.Expr,
For: cfg.For.Duration(),
Labels: cfg.Labels,
Annotations: cfg.Annotations,
GroupID: group.ID(),
GroupName: group.Name,
EvalInterval: group.Interval,
q: qb.BuildWithParams(datasource.QuerierParams{
DataSourceType: &group.Type,
EvaluationInterval: group.Interval,
QueryParams: group.Params,
}),
alerts: make(map[uint64]*notifier.Alert),
metrics: &alertingRuleMetrics{},
}
labels := fmt.Sprintf(`alertname=%q, group=%q, id="%d"`, ar.Name, group.Name, ar.ID())
ar.metrics.pending = utils.GetOrCreateGauge(fmt.Sprintf(`vmalert_alerts_pending{%s}`, labels),
func() float64 {
ar.mu.RLock()
defer ar.mu.RUnlock()
var num int
for _, a := range ar.alerts {
if a.State == notifier.StatePending {
num++
}
}
return float64(num)
})
ar.metrics.active = utils.GetOrCreateGauge(fmt.Sprintf(`vmalert_alerts_firing{%s}`, labels),
func() float64 {
ar.mu.RLock()
defer ar.mu.RUnlock()
var num int
for _, a := range ar.alerts {
if a.State == notifier.StateFiring {
num++
}
}
return float64(num)
})
ar.metrics.errors = utils.GetOrCreateGauge(fmt.Sprintf(`vmalert_alerting_rules_error{%s}`, labels),
func() float64 {
ar.mu.RLock()
defer ar.mu.RUnlock()
if ar.lastExecError == nil {
return 0
}
return 1
})
ar.metrics.samples = utils.GetOrCreateGauge(fmt.Sprintf(`vmalert_alerting_rules_last_evaluation_samples{%s}`, labels),
func() float64 {
ar.mu.RLock()
defer ar.mu.RUnlock()
return float64(ar.lastExecSamples)
})
return ar
}
// Close unregisters rule metrics
func (ar *AlertingRule) Close() {
ar.metrics.active.Unregister()
ar.metrics.pending.Unregister()
ar.metrics.errors.Unregister()
ar.metrics.samples.Unregister()
}
// String implements Stringer interface
func (ar *AlertingRule) String() string {
return ar.Name
}
// ID returns unique Rule ID
// within the parent Group.
func (ar *AlertingRule) ID() uint64 {
return ar.RuleID
}
// ExecRange executes alerting rule on the given time range similarly to Exec.
// It doesn't update internal states of the Rule and meant to be used just
// to get time series for backfilling.
// It returns ALERT and ALERT_FOR_STATE time series as result.
func (ar *AlertingRule) ExecRange(ctx context.Context, start, end time.Time) ([]prompbmarshal.TimeSeries, error) {
series, err := ar.q.QueryRange(ctx, ar.Expr, start, end)
if err != nil {
return nil, err
}
var result []prompbmarshal.TimeSeries
qFn := func(query string) ([]datasource.Metric, error) {
return nil, fmt.Errorf("`query` template isn't supported in replay mode")
}
for _, s := range series {
// set additional labels to identify group and rule Name
if ar.Name != "" {
s.SetLabel(alertNameLabel, ar.Name)
}
if !*disableAlertGroupLabel && ar.GroupName != "" {
s.SetLabel(alertGroupNameLabel, ar.GroupName)
}
// extra labels could contain templates, so we expand them first
labels, err := expandLabels(s, qFn, ar)
if err != nil {
return nil, fmt.Errorf("failed to expand labels: %s", err)
}
for k, v := range labels {
// apply extra labels to datasource
// so the hash key will be consistent on restore
s.SetLabel(k, v)
}
a, err := ar.newAlert(s, time.Time{}, qFn) // initial alert
if err != nil {
return nil, fmt.Errorf("failed to create alert: %s", err)
}
if ar.For == 0 { // if alert is instant
a.State = notifier.StateFiring
for i := range s.Values {
result = append(result, ar.alertToTimeSeries(a, s.Timestamps[i])...)
}
continue
}
// if alert with For > 0
prevT := time.Time{}
for i := range s.Values {
at := time.Unix(s.Timestamps[i], 0)
if at.Sub(prevT) > ar.EvalInterval {
// reset to Pending if there are gaps > EvalInterval between DPs
a.State = notifier.StatePending
a.ActiveAt = at
} else if at.Sub(a.ActiveAt) >= ar.For {
a.State = notifier.StateFiring
a.Start = at
}
prevT = at
result = append(result, ar.alertToTimeSeries(a, s.Timestamps[i])...)
}
}
return result, nil
}
// resolvedRetention is the duration for which a resolved alert instance
// is kept in memory state and consequently repeatedly sent to the AlertManager.
const resolvedRetention = 15 * time.Minute
// Exec executes AlertingRule expression via the given Querier.
// Based on the Querier results AlertingRule maintains notifier.Alerts
func (ar *AlertingRule) Exec(ctx context.Context, ts time.Time) ([]prompbmarshal.TimeSeries, error) {
start := time.Now()
qMetrics, err := ar.q.Query(ctx, ar.Expr, ts)
ar.mu.Lock()
defer ar.mu.Unlock()
ar.lastExecTime = start
ar.lastExecDuration = time.Since(start)
ar.lastExecError = err
ar.lastExecSamples = len(qMetrics)
if err != nil {
return nil, fmt.Errorf("failed to execute query %q: %w", ar.Expr, err)
}
for h, a := range ar.alerts {
// cleanup inactive alerts from previous Exec
if a.State == notifier.StateInactive && ts.Sub(a.ResolvedAt) > resolvedRetention {
delete(ar.alerts, h)
}
}
qFn := func(query string) ([]datasource.Metric, error) { return ar.q.Query(ctx, query, ts) }
updated := make(map[uint64]struct{})
// update list of active alerts
for _, m := range qMetrics {
// set additional labels to identify group and rule name
if ar.Name != "" {
m.SetLabel(alertNameLabel, ar.Name)
}
if !*disableAlertGroupLabel && ar.GroupName != "" {
m.SetLabel(alertGroupNameLabel, ar.GroupName)
}
// extra labels could contain templates, so we expand them first
labels, err := expandLabels(m, qFn, ar)
if err != nil {
return nil, fmt.Errorf("failed to expand labels: %s", err)
}
for k, v := range labels {
// apply extra labels to datasource
// so the hash key will be consistent on restore
m.SetLabel(k, v)
}
h := hash(m)
if _, ok := updated[h]; ok {
// duplicate may be caused by extra labels
// conflicting with the metric labels
ar.lastExecError = fmt.Errorf("labels %v: %w", m.Labels, errDuplicate)
return nil, ar.lastExecError
}
updated[h] = struct{}{}
if a, ok := ar.alerts[h]; ok {
if a.State == notifier.StateInactive {
// alert could be in inactive state for resolvedRetention
// so when we again receive metrics for it - we switch it
// back to notifier.StatePending
a.State = notifier.StatePending
a.ActiveAt = ts
}
if a.Value != m.Values[0] {
// update Value field with latest value
a.Value = m.Values[0]
// and re-exec template since Value can be used
// in annotations
a.Annotations, err = a.ExecTemplate(qFn, ar.Annotations)
if err != nil {
return nil, err
}
}
continue
}
a, err := ar.newAlert(m, ar.lastExecTime, qFn)
if err != nil {
ar.lastExecError = err
return nil, fmt.Errorf("failed to create alert: %w", err)
}
a.ID = h
a.State = notifier.StatePending
a.ActiveAt = ts
ar.alerts[h] = a
}
for h, a := range ar.alerts {
// if alert wasn't updated in this iteration
// means it is resolved already
if _, ok := updated[h]; !ok {
if a.State == notifier.StatePending {
// alert was in Pending state - it is not
// active anymore
delete(ar.alerts, h)
continue
}
if a.State == notifier.StateFiring {
a.State = notifier.StateInactive
a.ResolvedAt = ts
}
continue
}
if a.State == notifier.StatePending && time.Since(a.ActiveAt) >= ar.For {
a.State = notifier.StateFiring
a.Start = ts
alertsFired.Inc()
}
}
return ar.toTimeSeries(ts.Unix()), nil
}
func expandLabels(m datasource.Metric, q notifier.QueryFn, ar *AlertingRule) (map[string]string, error) {
metricLabels := make(map[string]string)
for _, l := range m.Labels {
metricLabels[l.Name] = l.Value
}
tpl := notifier.AlertTplData{
Labels: metricLabels,
Value: m.Values[0],
Expr: ar.Expr,
}
return notifier.ExecTemplate(q, ar.Labels, tpl)
}
func (ar *AlertingRule) toTimeSeries(timestamp int64) []prompbmarshal.TimeSeries {
var tss []prompbmarshal.TimeSeries
for _, a := range ar.alerts {
if a.State == notifier.StateInactive {
continue
}
ts := ar.alertToTimeSeries(a, timestamp)
tss = append(tss, ts...)
}
return tss
}
// UpdateWith copies all significant fields.
// alerts state isn't copied since
// it should be updated in next 2 Execs
func (ar *AlertingRule) UpdateWith(r Rule) error {
nr, ok := r.(*AlertingRule)
if !ok {
return fmt.Errorf("BUG: attempt to update alerting rule with wrong type %#v", r)
}
ar.Expr = nr.Expr
ar.For = nr.For
ar.Labels = nr.Labels
ar.Annotations = nr.Annotations
ar.EvalInterval = nr.EvalInterval
ar.q = nr.q
return nil
}
// TODO: consider hashing algorithm in VM
func hash(m datasource.Metric) uint64 {
hash := fnv.New64a()
labels := m.Labels
sort.Slice(labels, func(i, j int) bool {
return labels[i].Name < labels[j].Name
})
for _, l := range labels {
// drop __name__ to be consistent with Prometheus alerting
if l.Name == "__name__" {
continue
}
hash.Write([]byte(l.Name))
hash.Write([]byte(l.Value))
hash.Write([]byte("\xff"))
}
return hash.Sum64()
}
func (ar *AlertingRule) newAlert(m datasource.Metric, start time.Time, qFn notifier.QueryFn) (*notifier.Alert, error) {
a := &notifier.Alert{
GroupID: ar.GroupID,
Name: ar.Name,
Labels: map[string]string{},
Value: m.Values[0],
ActiveAt: start,
Expr: ar.Expr,
}
for _, l := range m.Labels {
// drop __name__ to be consistent with Prometheus alerting
if l.Name == "__name__" {
continue
}
a.Labels[l.Name] = l.Value
}
var err error
a.Annotations, err = a.ExecTemplate(qFn, ar.Annotations)
return a, err
}
// AlertAPI generates APIAlert object from alert by its id(hash)
func (ar *AlertingRule) AlertAPI(id uint64) *APIAlert {
ar.mu.RLock()
defer ar.mu.RUnlock()
a, ok := ar.alerts[id]
if !ok {
return nil
}
return ar.newAlertAPI(*a)
}
// ToAPI returns Rule representation in form
// of APIRule
func (ar *AlertingRule) ToAPI() APIRule {
r := APIRule{
Type: "alerting",
DatasourceType: ar.Type.String(),
Name: ar.Name,
Query: ar.Expr,
Duration: ar.For.Seconds(),
Labels: ar.Labels,
Annotations: ar.Annotations,
LastEvaluation: ar.lastExecTime,
EvaluationTime: ar.lastExecDuration.Seconds(),
Health: "ok",
State: "inactive",
Alerts: ar.AlertsToAPI(),
LastSamples: ar.lastExecSamples,
// encode as strings to avoid rounding in JSON
ID: fmt.Sprintf("%d", ar.ID()),
GroupID: fmt.Sprintf("%d", ar.GroupID),
}
if ar.lastExecError != nil {
r.LastError = ar.lastExecError.Error()
r.Health = "err"
}
// satisfy APIRule.State logic
if len(r.Alerts) > 0 {
r.State = notifier.StatePending.String()
stateFiring := notifier.StateFiring.String()
for _, a := range r.Alerts {
if a.State == stateFiring {
r.State = stateFiring
break
}
}
}
return r
}
// AlertsToAPI generates list of APIAlert objects from existing alerts
func (ar *AlertingRule) AlertsToAPI() []*APIAlert {
var alerts []*APIAlert
ar.mu.RLock()
for _, a := range ar.alerts {
if a.State == notifier.StateInactive {
continue
}
alerts = append(alerts, ar.newAlertAPI(*a))
}
ar.mu.RUnlock()
return alerts
}
func (ar *AlertingRule) newAlertAPI(a notifier.Alert) *APIAlert {
aa := &APIAlert{
// encode as strings to avoid rounding
ID: fmt.Sprintf("%d", a.ID),
GroupID: fmt.Sprintf("%d", a.GroupID),
RuleID: fmt.Sprintf("%d", ar.RuleID),
Name: a.Name,
Expression: ar.Expr,
Labels: a.Labels,
Annotations: a.Annotations,
State: a.State.String(),
ActiveAt: a.ActiveAt,
Restored: a.Restored,
Value: strconv.FormatFloat(a.Value, 'f', -1, 32),
}
if alertURLGeneratorFn != nil {
aa.SourceLink = alertURLGeneratorFn(a)
}
return aa
}
const (
// alertMetricName is the metric name for synthetic alert timeseries.
alertMetricName = "ALERTS"
// alertForStateMetricName is the metric name for 'for' state of alert.
alertForStateMetricName = "ALERTS_FOR_STATE"
// alertNameLabel is the label name indicating the name of an alert.
alertNameLabel = "alertname"
// alertStateLabel is the label name indicating the state of an alert.
alertStateLabel = "alertstate"
// alertGroupNameLabel defines the label name attached for generated time series.
// attaching this label may be disabled via `-disableAlertgroupLabel` flag.
alertGroupNameLabel = "alertgroup"
)
// alertToTimeSeries converts the given alert with the given timestamp to time series
func (ar *AlertingRule) alertToTimeSeries(a *notifier.Alert, timestamp int64) []prompbmarshal.TimeSeries {
var tss []prompbmarshal.TimeSeries
tss = append(tss, alertToTimeSeries(a, timestamp))
if ar.For > 0 {
tss = append(tss, alertForToTimeSeries(a, timestamp))
}
return tss
}
func alertToTimeSeries(a *notifier.Alert, timestamp int64) prompbmarshal.TimeSeries {
labels := make(map[string]string)
for k, v := range a.Labels {
labels[k] = v
}
labels["__name__"] = alertMetricName
labels[alertStateLabel] = a.State.String()
return newTimeSeries([]float64{1}, []int64{timestamp}, labels)
}
// alertForToTimeSeries returns a timeseries that represents
// state of active alerts, where value is time when alert become active
func alertForToTimeSeries(a *notifier.Alert, timestamp int64) prompbmarshal.TimeSeries {
labels := make(map[string]string)
for k, v := range a.Labels {
labels[k] = v
}
labels["__name__"] = alertForStateMetricName
return newTimeSeries([]float64{float64(a.ActiveAt.Unix())}, []int64{timestamp}, labels)
}
// Restore restores the state of active alerts basing on previously written time series.
// Restore restores only ActiveAt field. Field State will be always Pending and supposed
// to be updated on next Exec, as well as Value field.
// Only rules with For > 0 will be restored.
func (ar *AlertingRule) Restore(ctx context.Context, q datasource.Querier, lookback time.Duration, labels map[string]string) error {
if q == nil {
return fmt.Errorf("querier is nil")
}
ts := time.Now()
qFn := func(query string) ([]datasource.Metric, error) { return ar.q.Query(ctx, query, ts) }
// account for external labels in filter
var labelsFilter string
for k, v := range labels {
labelsFilter += fmt.Sprintf(",%s=%q", k, v)
}
// Get the last data point in range via MetricsQL `last_over_time`.
// We don't use plain PromQL since Prometheus doesn't support
// remote write protocol which is used for state persistence in vmalert.
expr := fmt.Sprintf("last_over_time(%s{alertname=%q%s}[%ds])",
alertForStateMetricName, ar.Name, labelsFilter, int(lookback.Seconds()))
qMetrics, err := q.Query(ctx, expr, ts)
if err != nil {
return err
}
for _, m := range qMetrics {
a, err := ar.newAlert(m, time.Unix(int64(m.Values[0]), 0), qFn)
if err != nil {
return fmt.Errorf("failed to create alert: %w", err)
}
a.ID = hash(m)
a.State = notifier.StatePending
a.Restored = true
ar.alerts[a.ID] = a
logger.Infof("alert %q (%d) restored to state at %v", a.Name, a.ID, a.Start)
}
return nil
}
// alertsToSend walks through the current alerts of AlertingRule
// and returns only those which should be sent to notifier.
// Isn't concurrent safe.
func (ar *AlertingRule) alertsToSend(ts time.Time, resolveDuration, resendDelay time.Duration) []notifier.Alert {
needsSending := func(a *notifier.Alert) bool {
if a.State == notifier.StatePending {
return false
}
if a.ResolvedAt.After(a.LastSent) {
return true
}
return a.LastSent.Add(resendDelay).Before(ts)
}
var alerts []notifier.Alert
for _, a := range ar.alerts {
if !needsSending(a) {
continue
}
a.End = ts.Add(resolveDuration)
if a.State == notifier.StateInactive {
a.End = a.ResolvedAt
}
a.LastSent = ts
alerts = append(alerts, *a)
}
return alerts
}