216 lines
5.1 KiB
Go
216 lines
5.1 KiB
Go
// Copyright 2013 Prometheus Team
|
|
// Licensed under the Apache License, Version 2.0 (the "License");
|
|
// you may not use this file except in compliance with the License.
|
|
// You may obtain a copy of the License at
|
|
//
|
|
// http://www.apache.org/licenses/LICENSE-2.0
|
|
//
|
|
// Unless required by applicable law or agreed to in writing, software
|
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
// See the License for the specific language governing permissions and
|
|
// limitations under the License.
|
|
|
|
package manager
|
|
|
|
import (
|
|
"errors"
|
|
"log"
|
|
"sync"
|
|
"time"
|
|
)
|
|
|
|
const (
|
|
minimumRepeatRate = 5 * time.Minute
|
|
minimumRefreshPeriod = 5 * time.Minute
|
|
notificationRetryPeriod = 1 * time.Minute
|
|
)
|
|
|
|
// AggregationRule creates and manages the scope for received events.
|
|
type AggregationRule struct {
|
|
Filters Filters
|
|
RepeatRate time.Duration
|
|
NotificationConfigName string
|
|
}
|
|
|
|
type AggregationInstances []*AggregationInstance
|
|
|
|
type AggregationInstance struct {
|
|
Rule *AggregationRule
|
|
Event *Event
|
|
|
|
// When was this AggregationInstance created?
|
|
Created time.Time
|
|
// When was the last refresh received into this AggregationInstance?
|
|
LastRefreshed time.Time
|
|
|
|
// When was the last successful notification sent out for this
|
|
// AggregationInstance?
|
|
lastNotificationSent time.Time
|
|
// Timer used to trigger a notification retry/resend.
|
|
notificationResendTimer *time.Timer
|
|
// Timer used to trigger the deletion of the AggregationInstance after it
|
|
// hasn't been refreshed for too long.
|
|
expiryTimer *time.Timer
|
|
}
|
|
|
|
func (r *AggregationRule) Handles(e *Event) bool {
|
|
return r.Filters.Handles(e)
|
|
}
|
|
|
|
func (r *AggregationInstance) Ingest(e *Event) {
|
|
r.Event = e
|
|
r.LastRefreshed = time.Now()
|
|
|
|
r.expiryTimer.Reset(minimumRefreshPeriod)
|
|
}
|
|
|
|
func (r *AggregationInstance) SendNotification(n Notifier) {
|
|
if time.Since(r.lastNotificationSent) < r.Rule.RepeatRate {
|
|
return
|
|
}
|
|
|
|
err := n.QueueNotification(r.Event, r.Rule.NotificationConfigName)
|
|
if err != nil {
|
|
// BUG: Limit the number of retries.
|
|
log.Printf("Error while sending notification: %s, retrying in %v", err, notificationRetryPeriod)
|
|
r.resendNotificationAfter(notificationRetryPeriod, n)
|
|
return
|
|
}
|
|
|
|
r.resendNotificationAfter(r.Rule.RepeatRate, n)
|
|
r.lastNotificationSent = time.Now()
|
|
}
|
|
|
|
func (r *AggregationInstance) resendNotificationAfter(d time.Duration, n Notifier) {
|
|
r.notificationResendTimer = time.AfterFunc(d, func() {
|
|
r.SendNotification(n)
|
|
})
|
|
}
|
|
|
|
func (r *AggregationInstance) Close() {
|
|
if r.notificationResendTimer != nil {
|
|
r.notificationResendTimer.Stop()
|
|
}
|
|
if r.expiryTimer != nil {
|
|
r.expiryTimer.Stop()
|
|
}
|
|
}
|
|
|
|
type AggregationRules []*AggregationRule
|
|
|
|
type Aggregator struct {
|
|
Rules AggregationRules
|
|
Aggregates map[EventFingerprint]*AggregationInstance
|
|
Notifier Notifier
|
|
|
|
// Mutex to protect the above.
|
|
mu sync.Mutex
|
|
}
|
|
|
|
func NewAggregator(n Notifier) *Aggregator {
|
|
return &Aggregator{
|
|
Aggregates: make(map[EventFingerprint]*AggregationInstance),
|
|
Notifier: n,
|
|
}
|
|
}
|
|
|
|
func (a *Aggregator) Close() {
|
|
a.mu.Lock()
|
|
defer a.mu.Unlock()
|
|
|
|
for _, agg := range a.Aggregates {
|
|
agg.Close()
|
|
}
|
|
}
|
|
|
|
func (a *Aggregator) Receive(events Events) error {
|
|
a.mu.Lock()
|
|
defer a.mu.Unlock()
|
|
|
|
if len(a.Rules) == 0 {
|
|
return errors.New("No aggregation rules")
|
|
}
|
|
for _, e := range events {
|
|
for _, r := range a.Rules {
|
|
if r.Handles(e) {
|
|
fp := e.Fingerprint()
|
|
aggregation, ok := a.Aggregates[fp]
|
|
if !ok {
|
|
expTimer := time.AfterFunc(minimumRefreshPeriod, func() {
|
|
a.mu.Lock()
|
|
defer a.mu.Unlock()
|
|
a.removeAggregate(fp)
|
|
})
|
|
|
|
aggregation = &AggregationInstance{
|
|
Rule: r,
|
|
Created: time.Now(),
|
|
expiryTimer: expTimer,
|
|
}
|
|
|
|
a.Aggregates[fp] = aggregation
|
|
}
|
|
|
|
aggregation.Ingest(e)
|
|
aggregation.SendNotification(a.Notifier)
|
|
break
|
|
}
|
|
}
|
|
}
|
|
return nil
|
|
}
|
|
|
|
func (a *Aggregator) SetRules(rules AggregationRules) {
|
|
a.mu.Lock()
|
|
defer a.mu.Unlock()
|
|
|
|
log.Println("Replacing", len(rules), "aggregator rules...")
|
|
|
|
for _, rule := range rules {
|
|
if rule.RepeatRate < minimumRepeatRate {
|
|
log.Println("Rule repeat rate too low, setting to minimum value")
|
|
rule.RepeatRate = minimumRepeatRate
|
|
}
|
|
}
|
|
|
|
a.Rules = rules
|
|
|
|
// Reparent AggregationInstances to the first new matching rule, drop orphans
|
|
// that are not matched by any rule anymore. Expiry and notification resend
|
|
// timers are left untouched for reparented alerts, meaning that the last
|
|
// rule's RepeatRate needs to pass once before the new one is used.
|
|
for fp, agg := range a.Aggregates {
|
|
orphaned := true
|
|
|
|
for _, r := range a.Rules {
|
|
if r.Handles(agg.Event) {
|
|
agg.Rule = r
|
|
orphaned = false
|
|
break
|
|
}
|
|
}
|
|
|
|
if orphaned {
|
|
a.removeAggregate(fp)
|
|
}
|
|
}
|
|
}
|
|
|
|
func (a *Aggregator) AlertAggregates() AggregationInstances {
|
|
a.mu.Lock()
|
|
defer a.mu.Unlock()
|
|
|
|
aggs := make(AggregationInstances, 0, len(a.Aggregates))
|
|
for _, agg := range a.Aggregates {
|
|
aggs = append(aggs, agg)
|
|
}
|
|
return aggs
|
|
}
|
|
|
|
func (a *Aggregator) removeAggregate(fp EventFingerprint) {
|
|
log.Println("Deleting expired aggregation instance", a)
|
|
a.Aggregates[fp].Close()
|
|
delete(a.Aggregates, fp)
|
|
}
|