VictoriaMetrics/app/vmalert/group.go

534 lines
14 KiB
Go
Raw Normal View History

package main
import (
"context"
"fmt"
"hash/fnv"
"net/url"
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>
2022-03-29 15:09:07 +02:00
"strconv"
"strings"
"sync"
"time"
"github.com/VictoriaMetrics/metrics"
"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/remotewrite"
"github.com/VictoriaMetrics/VictoriaMetrics/app/vmalert/utils"
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>
2022-03-29 15:09:07 +02:00
"github.com/VictoriaMetrics/VictoriaMetrics/lib/decimal"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
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>
2022-03-29 15:09:07 +02:00
"github.com/VictoriaMetrics/VictoriaMetrics/lib/prompbmarshal"
)
// Group is an entity for grouping rules
type Group struct {
mu sync.RWMutex
Name string
File string
Rules []Rule
Type config.Type
Interval time.Duration
Limit int
Concurrency int
Checksum string
LastEvaluation time.Time
Labels map[string]string
Params url.Values
Headers map[string]string
doneCh chan struct{}
finishedCh chan struct{}
// channel accepts new Group obj
// which supposed to update current group
updateCh chan *Group
app/vmalert: extend metrics set exported by `vmalert` #573 (#654) * app/vmalert: extend metrics set exported by `vmalert` #573 New metrics were added to improve observability: + vmalert_alerts_pending{alertname, group} - number of pending alerts per group per alert; + vmalert_alerts_acitve{alertname, group} - number of active alerts per group per alert; + vmalert_alerts_error{alertname, group} - is 1 if alertname ended up with error during prev execution, is 0 if no errors happened; + vmalert_recording_rules_error{recording, group} - is 1 if recording rule ended up with error during prev execution, is 0 if no errors happened; * vmalert_iteration_total{group, file} - now contains group and file name labels. This should improve control over specific groups; * vmalert_iteration_duration_seconds{group, file} - now contains group and file name labels. This should improve control over specific groups; Some collisions for alerts and recording rules are possible, because neither group name nor alert/recording rule name are unique for compatibility reasons. Commit contains list of TODOs for Unregistering metrics since groups and rules are ephemeral and could be removed without application restart. In order to unlock Unregistering feature corresponding PR was filed - https://github.com/VictoriaMetrics/metrics/pull/13 * app/vmalert: extend metrics set exported by `vmalert` #573 The changes are following: * add an ID label to rules metrics, since `name` collisions within one group is a common case - see the k8s example alerts; * supports metrics unregistering on rule updates. Consider the case when one rule was added or removed from the group, or the whole group was added or removed. The change depends on https://github.com/VictoriaMetrics/metrics/pull/16 where race condition for Unregister method was fixed.
2020-08-09 08:41:29 +02:00
metrics *groupMetrics
}
type groupMetrics struct {
iterationTotal *utils.Counter
iterationDuration *utils.Summary
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>
2022-03-29 15:09:07 +02:00
iterationMissed *utils.Counter
iterationInterval *utils.Gauge
app/vmalert: extend metrics set exported by `vmalert` #573 (#654) * app/vmalert: extend metrics set exported by `vmalert` #573 New metrics were added to improve observability: + vmalert_alerts_pending{alertname, group} - number of pending alerts per group per alert; + vmalert_alerts_acitve{alertname, group} - number of active alerts per group per alert; + vmalert_alerts_error{alertname, group} - is 1 if alertname ended up with error during prev execution, is 0 if no errors happened; + vmalert_recording_rules_error{recording, group} - is 1 if recording rule ended up with error during prev execution, is 0 if no errors happened; * vmalert_iteration_total{group, file} - now contains group and file name labels. This should improve control over specific groups; * vmalert_iteration_duration_seconds{group, file} - now contains group and file name labels. This should improve control over specific groups; Some collisions for alerts and recording rules are possible, because neither group name nor alert/recording rule name are unique for compatibility reasons. Commit contains list of TODOs for Unregistering metrics since groups and rules are ephemeral and could be removed without application restart. In order to unlock Unregistering feature corresponding PR was filed - https://github.com/VictoriaMetrics/metrics/pull/13 * app/vmalert: extend metrics set exported by `vmalert` #573 The changes are following: * add an ID label to rules metrics, since `name` collisions within one group is a common case - see the k8s example alerts; * supports metrics unregistering on rule updates. Consider the case when one rule was added or removed from the group, or the whole group was added or removed. The change depends on https://github.com/VictoriaMetrics/metrics/pull/16 where race condition for Unregister method was fixed.
2020-08-09 08:41:29 +02:00
}
func newGroupMetrics(g *Group) *groupMetrics {
app/vmalert: extend metrics set exported by `vmalert` #573 (#654) * app/vmalert: extend metrics set exported by `vmalert` #573 New metrics were added to improve observability: + vmalert_alerts_pending{alertname, group} - number of pending alerts per group per alert; + vmalert_alerts_acitve{alertname, group} - number of active alerts per group per alert; + vmalert_alerts_error{alertname, group} - is 1 if alertname ended up with error during prev execution, is 0 if no errors happened; + vmalert_recording_rules_error{recording, group} - is 1 if recording rule ended up with error during prev execution, is 0 if no errors happened; * vmalert_iteration_total{group, file} - now contains group and file name labels. This should improve control over specific groups; * vmalert_iteration_duration_seconds{group, file} - now contains group and file name labels. This should improve control over specific groups; Some collisions for alerts and recording rules are possible, because neither group name nor alert/recording rule name are unique for compatibility reasons. Commit contains list of TODOs for Unregistering metrics since groups and rules are ephemeral and could be removed without application restart. In order to unlock Unregistering feature corresponding PR was filed - https://github.com/VictoriaMetrics/metrics/pull/13 * app/vmalert: extend metrics set exported by `vmalert` #573 The changes are following: * add an ID label to rules metrics, since `name` collisions within one group is a common case - see the k8s example alerts; * supports metrics unregistering on rule updates. Consider the case when one rule was added or removed from the group, or the whole group was added or removed. The change depends on https://github.com/VictoriaMetrics/metrics/pull/16 where race condition for Unregister method was fixed.
2020-08-09 08:41:29 +02:00
m := &groupMetrics{}
labels := fmt.Sprintf(`group=%q, file=%q`, g.Name, g.File)
m.iterationTotal = utils.GetOrCreateCounter(fmt.Sprintf(`vmalert_iteration_total{%s}`, labels))
m.iterationDuration = utils.GetOrCreateSummary(fmt.Sprintf(`vmalert_iteration_duration_seconds{%s}`, labels))
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>
2022-03-29 15:09:07 +02:00
m.iterationMissed = utils.GetOrCreateCounter(fmt.Sprintf(`vmalert_iteration_missed_total{%s}`, labels))
m.iterationInterval = utils.GetOrCreateGauge(fmt.Sprintf(`vmalert_iteration_interval_seconds{%s}`, labels), func() float64 {
g.mu.RLock()
i := g.Interval.Seconds()
g.mu.RUnlock()
return i
})
app/vmalert: extend metrics set exported by `vmalert` #573 (#654) * app/vmalert: extend metrics set exported by `vmalert` #573 New metrics were added to improve observability: + vmalert_alerts_pending{alertname, group} - number of pending alerts per group per alert; + vmalert_alerts_acitve{alertname, group} - number of active alerts per group per alert; + vmalert_alerts_error{alertname, group} - is 1 if alertname ended up with error during prev execution, is 0 if no errors happened; + vmalert_recording_rules_error{recording, group} - is 1 if recording rule ended up with error during prev execution, is 0 if no errors happened; * vmalert_iteration_total{group, file} - now contains group and file name labels. This should improve control over specific groups; * vmalert_iteration_duration_seconds{group, file} - now contains group and file name labels. This should improve control over specific groups; Some collisions for alerts and recording rules are possible, because neither group name nor alert/recording rule name are unique for compatibility reasons. Commit contains list of TODOs for Unregistering metrics since groups and rules are ephemeral and could be removed without application restart. In order to unlock Unregistering feature corresponding PR was filed - https://github.com/VictoriaMetrics/metrics/pull/13 * app/vmalert: extend metrics set exported by `vmalert` #573 The changes are following: * add an ID label to rules metrics, since `name` collisions within one group is a common case - see the k8s example alerts; * supports metrics unregistering on rule updates. Consider the case when one rule was added or removed from the group, or the whole group was added or removed. The change depends on https://github.com/VictoriaMetrics/metrics/pull/16 where race condition for Unregister method was fixed.
2020-08-09 08:41:29 +02:00
return m
}
// merges group rule labels into result map
// set2 has priority over set1.
func mergeLabels(groupName, ruleName string, set1, set2 map[string]string) map[string]string {
r := map[string]string{}
for k, v := range set1 {
r[k] = v
}
for k, v := range set2 {
if prevV, ok := r[k]; ok {
logger.Infof("label %q=%q for rule %q.%q overwritten with external label %q=%q",
k, prevV, groupName, ruleName, k, v)
}
r[k] = v
}
return r
}
func newGroup(cfg config.Group, qb datasource.QuerierBuilder, defaultInterval time.Duration, labels map[string]string) *Group {
g := &Group{
Type: cfg.Type,
Name: cfg.Name,
File: cfg.File,
Interval: cfg.Interval.Duration(),
Limit: cfg.Limit,
Concurrency: cfg.Concurrency,
Checksum: cfg.Checksum,
Params: cfg.Params,
Headers: make(map[string]string),
Labels: cfg.Labels,
doneCh: make(chan struct{}),
finishedCh: make(chan struct{}),
updateCh: make(chan *Group),
}
if g.Interval == 0 {
g.Interval = defaultInterval
}
if g.Concurrency < 1 {
g.Concurrency = 1
}
for _, h := range cfg.Headers {
g.Headers[h.Key] = h.Value
}
g.metrics = newGroupMetrics(g)
rules := make([]Rule, len(cfg.Rules))
for i, r := range cfg.Rules {
var extraLabels map[string]string
// apply external labels
if len(labels) > 0 {
extraLabels = labels
}
// apply group labels, it has priority on external labels
if len(cfg.Labels) > 0 {
extraLabels = mergeLabels(g.Name, r.Name(), extraLabels, g.Labels)
}
// apply rules labels, it has priority on other labels
if len(extraLabels) > 0 {
r.Labels = mergeLabels(g.Name, r.Name(), extraLabels, r.Labels)
}
rules[i] = g.newRule(qb, r)
}
g.Rules = rules
return g
}
func (g *Group) newRule(qb datasource.QuerierBuilder, rule config.Rule) Rule {
if rule.Alert != "" {
return newAlertingRule(qb, g, rule)
}
return newRecordingRule(qb, g, rule)
}
// ID return unique group ID that consists of
// rules file and group Name
func (g *Group) ID() uint64 {
g.mu.RLock()
defer g.mu.RUnlock()
hash := fnv.New64a()
hash.Write([]byte(g.File))
hash.Write([]byte("\xff"))
hash.Write([]byte(g.Name))
hash.Write([]byte(g.Type.Get()))
return hash.Sum64()
}
// Restore restores alerts state for group rules
func (g *Group) Restore(ctx context.Context, qb datasource.QuerierBuilder, lookback time.Duration, labels map[string]string) error {
labels = mergeLabels(g.Name, "", labels, g.Labels)
for _, rule := range g.Rules {
rr, ok := rule.(*AlertingRule)
if !ok {
continue
}
if rr.For < 1 {
continue
}
// ignore QueryParams on purpose, because they could contain
// query filters. This may affect the restore procedure.
q := qb.BuildWithParams(datasource.QuerierParams{
DataSourceType: g.Type.String(),
Headers: g.Headers,
})
if err := rr.Restore(ctx, q, lookback, labels); err != nil {
return fmt.Errorf("error while restoring rule %q: %w", rule, err)
}
}
return nil
}
// updateWith updates existing group with
// passed group object. This function ignores group
// evaluation interval change. It supposed to be updated
// in group.start function.
// Not thread-safe.
func (g *Group) updateWith(newGroup *Group) error {
rulesRegistry := make(map[uint64]Rule)
for _, nr := range newGroup.Rules {
rulesRegistry[nr.ID()] = nr
}
for i, or := range g.Rules {
nr, ok := rulesRegistry[or.ID()]
if !ok {
// old rule is not present in the new list
// so we mark it for removing
app/vmalert: extend metrics set exported by `vmalert` #573 (#654) * app/vmalert: extend metrics set exported by `vmalert` #573 New metrics were added to improve observability: + vmalert_alerts_pending{alertname, group} - number of pending alerts per group per alert; + vmalert_alerts_acitve{alertname, group} - number of active alerts per group per alert; + vmalert_alerts_error{alertname, group} - is 1 if alertname ended up with error during prev execution, is 0 if no errors happened; + vmalert_recording_rules_error{recording, group} - is 1 if recording rule ended up with error during prev execution, is 0 if no errors happened; * vmalert_iteration_total{group, file} - now contains group and file name labels. This should improve control over specific groups; * vmalert_iteration_duration_seconds{group, file} - now contains group and file name labels. This should improve control over specific groups; Some collisions for alerts and recording rules are possible, because neither group name nor alert/recording rule name are unique for compatibility reasons. Commit contains list of TODOs for Unregistering metrics since groups and rules are ephemeral and could be removed without application restart. In order to unlock Unregistering feature corresponding PR was filed - https://github.com/VictoriaMetrics/metrics/pull/13 * app/vmalert: extend metrics set exported by `vmalert` #573 The changes are following: * add an ID label to rules metrics, since `name` collisions within one group is a common case - see the k8s example alerts; * supports metrics unregistering on rule updates. Consider the case when one rule was added or removed from the group, or the whole group was added or removed. The change depends on https://github.com/VictoriaMetrics/metrics/pull/16 where race condition for Unregister method was fixed.
2020-08-09 08:41:29 +02:00
g.Rules[i].Close()
g.Rules[i] = nil
continue
}
if err := or.UpdateWith(nr); err != nil {
return err
}
delete(rulesRegistry, nr.ID())
}
var newRules []Rule
for _, r := range g.Rules {
if r == nil {
// skip nil rules
continue
}
newRules = append(newRules, r)
}
// add the rest of rules from registry
for _, nr := range rulesRegistry {
newRules = append(newRules, nr)
}
// note that g.Interval is not updated here
// so the value can be compared later in
// group.Start function
g.Type = newGroup.Type
g.Concurrency = newGroup.Concurrency
g.Params = newGroup.Params
g.Headers = newGroup.Headers
g.Labels = newGroup.Labels
g.Limit = newGroup.Limit
g.Checksum = newGroup.Checksum
g.Rules = newRules
return nil
}
func (g *Group) close() {
if g.doneCh == nil {
return
}
close(g.doneCh)
<-g.finishedCh
app/vmalert: extend metrics set exported by `vmalert` #573 (#654) * app/vmalert: extend metrics set exported by `vmalert` #573 New metrics were added to improve observability: + vmalert_alerts_pending{alertname, group} - number of pending alerts per group per alert; + vmalert_alerts_acitve{alertname, group} - number of active alerts per group per alert; + vmalert_alerts_error{alertname, group} - is 1 if alertname ended up with error during prev execution, is 0 if no errors happened; + vmalert_recording_rules_error{recording, group} - is 1 if recording rule ended up with error during prev execution, is 0 if no errors happened; * vmalert_iteration_total{group, file} - now contains group and file name labels. This should improve control over specific groups; * vmalert_iteration_duration_seconds{group, file} - now contains group and file name labels. This should improve control over specific groups; Some collisions for alerts and recording rules are possible, because neither group name nor alert/recording rule name are unique for compatibility reasons. Commit contains list of TODOs for Unregistering metrics since groups and rules are ephemeral and could be removed without application restart. In order to unlock Unregistering feature corresponding PR was filed - https://github.com/VictoriaMetrics/metrics/pull/13 * app/vmalert: extend metrics set exported by `vmalert` #573 The changes are following: * add an ID label to rules metrics, since `name` collisions within one group is a common case - see the k8s example alerts; * supports metrics unregistering on rule updates. Consider the case when one rule was added or removed from the group, or the whole group was added or removed. The change depends on https://github.com/VictoriaMetrics/metrics/pull/16 where race condition for Unregister method was fixed.
2020-08-09 08:41:29 +02:00
g.metrics.iterationDuration.Unregister()
g.metrics.iterationTotal.Unregister()
g.metrics.iterationMissed.Unregister()
g.metrics.iterationInterval.Unregister()
app/vmalert: extend metrics set exported by `vmalert` #573 (#654) * app/vmalert: extend metrics set exported by `vmalert` #573 New metrics were added to improve observability: + vmalert_alerts_pending{alertname, group} - number of pending alerts per group per alert; + vmalert_alerts_acitve{alertname, group} - number of active alerts per group per alert; + vmalert_alerts_error{alertname, group} - is 1 if alertname ended up with error during prev execution, is 0 if no errors happened; + vmalert_recording_rules_error{recording, group} - is 1 if recording rule ended up with error during prev execution, is 0 if no errors happened; * vmalert_iteration_total{group, file} - now contains group and file name labels. This should improve control over specific groups; * vmalert_iteration_duration_seconds{group, file} - now contains group and file name labels. This should improve control over specific groups; Some collisions for alerts and recording rules are possible, because neither group name nor alert/recording rule name are unique for compatibility reasons. Commit contains list of TODOs for Unregistering metrics since groups and rules are ephemeral and could be removed without application restart. In order to unlock Unregistering feature corresponding PR was filed - https://github.com/VictoriaMetrics/metrics/pull/13 * app/vmalert: extend metrics set exported by `vmalert` #573 The changes are following: * add an ID label to rules metrics, since `name` collisions within one group is a common case - see the k8s example alerts; * supports metrics unregistering on rule updates. Consider the case when one rule was added or removed from the group, or the whole group was added or removed. The change depends on https://github.com/VictoriaMetrics/metrics/pull/16 where race condition for Unregister method was fixed.
2020-08-09 08:41:29 +02:00
for _, rule := range g.Rules {
rule.Close()
}
}
var skipRandSleepOnGroupStart bool
func (g *Group) start(ctx context.Context, nts func() []notifier.Notifier, rw *remotewrite.Client) {
defer func() { close(g.finishedCh) }()
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>
2022-03-29 15:09:07 +02:00
e := &executor{
rw: rw,
notifiers: nts,
previouslySentSeriesToRW: make(map[uint64]map[string][]prompbmarshal.Label)}
// Spread group rules evaluation over time in order to reduce load on VictoriaMetrics.
if !skipRandSleepOnGroupStart {
randSleep := uint64(float64(g.Interval) * (float64(g.ID()) / (1 << 64)))
sleepOffset := uint64(time.Now().UnixNano()) % uint64(g.Interval)
if randSleep < sleepOffset {
randSleep += uint64(g.Interval)
}
randSleep -= sleepOffset
sleepTimer := time.NewTimer(time.Duration(randSleep))
select {
case <-ctx.Done():
sleepTimer.Stop()
return
case <-g.doneCh:
sleepTimer.Stop()
return
case <-sleepTimer.C:
}
}
evalTS := time.Now()
logger.Infof("group %q started; interval=%v; concurrency=%d", g.Name, g.Interval, g.Concurrency)
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>
2022-03-29 15:09:07 +02:00
eval := func(ts time.Time) {
g.metrics.iterationTotal.Inc()
start := time.Now()
if len(g.Rules) < 1 {
g.metrics.iterationDuration.UpdateDuration(start)
g.LastEvaluation = start
return
}
resolveDuration := getResolveDuration(g.Interval, *resendDelay, *maxResolveDuration)
errs := e.execConcurrently(ctx, g.Rules, ts, g.Concurrency, resolveDuration, g.Limit)
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>
2022-03-29 15:09:07 +02:00
for err := range errs {
if err != nil {
logger.Errorf("group %q: %s", g.Name, err)
}
}
g.metrics.iterationDuration.UpdateDuration(start)
g.LastEvaluation = start
}
eval(evalTS)
t := time.NewTicker(g.Interval)
defer t.Stop()
for {
select {
case <-ctx.Done():
logger.Infof("group %q: context cancelled", g.Name)
return
case <-g.doneCh:
logger.Infof("group %q: received stop signal", g.Name)
return
case ng := <-g.updateCh:
g.mu.Lock()
err := g.updateWith(ng)
if err != nil {
logger.Errorf("group %q: failed to update: %s", g.Name, err)
g.mu.Unlock()
continue
}
// ensure that staleness is tracked or existing rules only
e.purgeStaleSeries(g.Rules)
if g.Interval != ng.Interval {
g.Interval = ng.Interval
t.Stop()
t = time.NewTicker(g.Interval)
}
g.mu.Unlock()
logger.Infof("group %q re-started; interval=%v; concurrency=%d", g.Name, g.Interval, g.Concurrency)
case <-t.C:
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>
2022-03-29 15:09:07 +02:00
missed := (time.Since(evalTS) / g.Interval) - 1
if missed > 0 {
g.metrics.iterationMissed.Inc()
}
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>
2022-03-29 15:09:07 +02:00
evalTS = evalTS.Add((missed + 1) * g.Interval)
eval(evalTS)
}
}
}
// getResolveDuration returns the duration after which firing alert
// can be considered as resolved.
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>
2022-03-29 15:09:07 +02:00
func getResolveDuration(groupInterval, delta, maxDuration time.Duration) time.Duration {
if groupInterval > delta {
delta = groupInterval
}
resolveDuration := delta * 4
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>
2022-03-29 15:09:07 +02:00
if maxDuration > 0 && resolveDuration > maxDuration {
resolveDuration = maxDuration
}
return resolveDuration
}
type executor struct {
notifiers func() []notifier.Notifier
rw *remotewrite.Client
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>
2022-03-29 15:09:07 +02:00
previouslySentSeriesToRWMu sync.Mutex
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>
2022-03-29 15:09:07 +02:00
// previouslySentSeriesToRW stores series sent to RW on previous iteration
// map[ruleID]map[ruleLabels][]prompb.Label
// where `ruleID` is ID of the Rule within a Group
// and `ruleLabels` is []prompb.Label marshalled to a string
previouslySentSeriesToRW map[uint64]map[string][]prompbmarshal.Label
}
func (e *executor) execConcurrently(ctx context.Context, rules []Rule, ts time.Time, concurrency int, resolveDuration time.Duration, limit int) chan error {
res := make(chan error, len(rules))
if concurrency == 1 {
// fast path
for _, rule := range rules {
res <- e.exec(ctx, rule, ts, resolveDuration, limit)
}
close(res)
return res
}
sem := make(chan struct{}, concurrency)
go func() {
wg := sync.WaitGroup{}
for _, rule := range rules {
sem <- struct{}{}
wg.Add(1)
go func(r Rule) {
res <- e.exec(ctx, r, ts, resolveDuration, limit)
<-sem
wg.Done()
}(rule)
}
wg.Wait()
close(res)
}()
return res
}
app/vmalert: extend metrics set exported by `vmalert` #573 (#654) * app/vmalert: extend metrics set exported by `vmalert` #573 New metrics were added to improve observability: + vmalert_alerts_pending{alertname, group} - number of pending alerts per group per alert; + vmalert_alerts_acitve{alertname, group} - number of active alerts per group per alert; + vmalert_alerts_error{alertname, group} - is 1 if alertname ended up with error during prev execution, is 0 if no errors happened; + vmalert_recording_rules_error{recording, group} - is 1 if recording rule ended up with error during prev execution, is 0 if no errors happened; * vmalert_iteration_total{group, file} - now contains group and file name labels. This should improve control over specific groups; * vmalert_iteration_duration_seconds{group, file} - now contains group and file name labels. This should improve control over specific groups; Some collisions for alerts and recording rules are possible, because neither group name nor alert/recording rule name are unique for compatibility reasons. Commit contains list of TODOs for Unregistering metrics since groups and rules are ephemeral and could be removed without application restart. In order to unlock Unregistering feature corresponding PR was filed - https://github.com/VictoriaMetrics/metrics/pull/13 * app/vmalert: extend metrics set exported by `vmalert` #573 The changes are following: * add an ID label to rules metrics, since `name` collisions within one group is a common case - see the k8s example alerts; * supports metrics unregistering on rule updates. Consider the case when one rule was added or removed from the group, or the whole group was added or removed. The change depends on https://github.com/VictoriaMetrics/metrics/pull/16 where race condition for Unregister method was fixed.
2020-08-09 08:41:29 +02:00
var (
alertsFired = metrics.NewCounter(`vmalert_alerts_fired_total`)
execTotal = metrics.NewCounter(`vmalert_execution_total`)
execErrors = metrics.NewCounter(`vmalert_execution_errors_total`)
app/vmalert: extend metrics set exported by `vmalert` #573 (#654) * app/vmalert: extend metrics set exported by `vmalert` #573 New metrics were added to improve observability: + vmalert_alerts_pending{alertname, group} - number of pending alerts per group per alert; + vmalert_alerts_acitve{alertname, group} - number of active alerts per group per alert; + vmalert_alerts_error{alertname, group} - is 1 if alertname ended up with error during prev execution, is 0 if no errors happened; + vmalert_recording_rules_error{recording, group} - is 1 if recording rule ended up with error during prev execution, is 0 if no errors happened; * vmalert_iteration_total{group, file} - now contains group and file name labels. This should improve control over specific groups; * vmalert_iteration_duration_seconds{group, file} - now contains group and file name labels. This should improve control over specific groups; Some collisions for alerts and recording rules are possible, because neither group name nor alert/recording rule name are unique for compatibility reasons. Commit contains list of TODOs for Unregistering metrics since groups and rules are ephemeral and could be removed without application restart. In order to unlock Unregistering feature corresponding PR was filed - https://github.com/VictoriaMetrics/metrics/pull/13 * app/vmalert: extend metrics set exported by `vmalert` #573 The changes are following: * add an ID label to rules metrics, since `name` collisions within one group is a common case - see the k8s example alerts; * supports metrics unregistering on rule updates. Consider the case when one rule was added or removed from the group, or the whole group was added or removed. The change depends on https://github.com/VictoriaMetrics/metrics/pull/16 where race condition for Unregister method was fixed.
2020-08-09 08:41:29 +02:00
remoteWriteErrors = metrics.NewCounter(`vmalert_remotewrite_errors_total`)
remoteWriteTotal = metrics.NewCounter(`vmalert_remotewrite_total`)
app/vmalert: extend metrics set exported by `vmalert` #573 (#654) * app/vmalert: extend metrics set exported by `vmalert` #573 New metrics were added to improve observability: + vmalert_alerts_pending{alertname, group} - number of pending alerts per group per alert; + vmalert_alerts_acitve{alertname, group} - number of active alerts per group per alert; + vmalert_alerts_error{alertname, group} - is 1 if alertname ended up with error during prev execution, is 0 if no errors happened; + vmalert_recording_rules_error{recording, group} - is 1 if recording rule ended up with error during prev execution, is 0 if no errors happened; * vmalert_iteration_total{group, file} - now contains group and file name labels. This should improve control over specific groups; * vmalert_iteration_duration_seconds{group, file} - now contains group and file name labels. This should improve control over specific groups; Some collisions for alerts and recording rules are possible, because neither group name nor alert/recording rule name are unique for compatibility reasons. Commit contains list of TODOs for Unregistering metrics since groups and rules are ephemeral and could be removed without application restart. In order to unlock Unregistering feature corresponding PR was filed - https://github.com/VictoriaMetrics/metrics/pull/13 * app/vmalert: extend metrics set exported by `vmalert` #573 The changes are following: * add an ID label to rules metrics, since `name` collisions within one group is a common case - see the k8s example alerts; * supports metrics unregistering on rule updates. Consider the case when one rule was added or removed from the group, or the whole group was added or removed. The change depends on https://github.com/VictoriaMetrics/metrics/pull/16 where race condition for Unregister method was fixed.
2020-08-09 08:41:29 +02:00
)
func (e *executor) exec(ctx context.Context, rule Rule, ts time.Time, resolveDuration time.Duration, limit int) error {
execTotal.Inc()
tss, err := rule.Exec(ctx, ts, limit)
if err != nil {
execErrors.Inc()
return fmt.Errorf("rule %q: failed to execute: %w", rule, err)
}
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>
2022-03-29 15:09:07 +02:00
errGr := new(utils.ErrGroup)
if e.rw != nil {
pushToRW := func(tss []prompbmarshal.TimeSeries) {
for _, ts := range tss {
remoteWriteTotal.Inc()
if err := e.rw.Push(ts); err != nil {
remoteWriteErrors.Inc()
errGr.Add(fmt.Errorf("rule %q: remote write failure: %w", rule, err))
}
}
}
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>
2022-03-29 15:09:07 +02:00
pushToRW(tss)
staleSeries := e.getStaleSeries(rule, tss, ts)
pushToRW(staleSeries)
}
ar, ok := rule.(*AlertingRule)
if !ok {
return nil
}
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>
2022-03-29 15:09:07 +02:00
alerts := ar.alertsToSend(ts, resolveDuration, *resendDelay)
if len(alerts) < 1 {
return nil
}
wg := sync.WaitGroup{}
for _, nt := range e.notifiers() {
wg.Add(1)
go func(nt notifier.Notifier) {
if err := nt.Send(ctx, alerts); err != nil {
errGr.Add(fmt.Errorf("rule %q: failed to send alerts to addr %q: %w", rule, nt.Addr(), err))
}
wg.Done()
}(nt)
}
wg.Wait()
return errGr.Err()
}
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>
2022-03-29 15:09:07 +02:00
// getStaledSeries checks whether there are stale series from previously sent ones.
func (e *executor) getStaleSeries(rule Rule, tss []prompbmarshal.TimeSeries, timestamp time.Time) []prompbmarshal.TimeSeries {
ruleLabels := make(map[string][]prompbmarshal.Label, len(tss))
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>
2022-03-29 15:09:07 +02:00
for _, ts := range tss {
// convert labels to strings so we can compare with previously sent series
key := labelsToString(ts.Labels)
ruleLabels[key] = ts.Labels
}
rID := rule.ID()
var staleS []prompbmarshal.TimeSeries
// check whether there are series which disappeared and need to be marked as stale
e.previouslySentSeriesToRWMu.Lock()
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>
2022-03-29 15:09:07 +02:00
for key, labels := range e.previouslySentSeriesToRW[rID] {
if _, ok := ruleLabels[key]; ok {
continue
}
// previously sent series are missing in current series, so we mark them as stale
ss := newTimeSeriesPB([]float64{decimal.StaleNaN}, []int64{timestamp.Unix()}, labels)
staleS = append(staleS, ss)
}
// set previous series to current
e.previouslySentSeriesToRW[rID] = ruleLabels
e.previouslySentSeriesToRWMu.Unlock()
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>
2022-03-29 15:09:07 +02:00
return staleS
}
// purgeStaleSeries deletes references in tracked
// previouslySentSeriesToRW list to Rules which aren't present
// in the given activeRules list. The method is used when the list
// of loaded rules has changed and executor has to remove
// references to non-existing rules.
func (e *executor) purgeStaleSeries(activeRules []Rule) {
newPreviouslySentSeriesToRW := make(map[uint64]map[string][]prompbmarshal.Label)
e.previouslySentSeriesToRWMu.Lock()
for _, rule := range activeRules {
id := rule.ID()
prev, ok := e.previouslySentSeriesToRW[id]
if ok {
// keep previous series for staleness detection
newPreviouslySentSeriesToRW[id] = prev
}
}
e.previouslySentSeriesToRW = nil
e.previouslySentSeriesToRW = newPreviouslySentSeriesToRW
e.previouslySentSeriesToRWMu.Unlock()
}
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>
2022-03-29 15:09:07 +02:00
func labelsToString(labels []prompbmarshal.Label) string {
var b strings.Builder
b.WriteRune('{')
for i, label := range labels {
if len(label.Name) == 0 {
b.WriteString("__name__")
} else {
b.WriteString(label.Name)
}
b.WriteRune('=')
b.WriteString(strconv.Quote(label.Value))
if i < len(labels)-1 {
b.WriteRune(',')
}
}
b.WriteRune('}')
return b.String()
}