diff --git a/core/internal/httpserver/coordinator.go b/core/internal/httpserver/coordinator.go index 2c5f37f2..42a6347d 100644 --- a/core/internal/httpserver/coordinator.go +++ b/core/internal/httpserver/coordinator.go @@ -130,6 +130,9 @@ func (hc *Coordinator) Configure() { // This is a healthcheck URL. Please don't change it hc.router.GET("/burrow/admin", hc.handleAdmin) + // Prometheus metrics + hc.router.GET("/metrics", hc.handlePrometheusMetrics) + // All valid paths go here hc.router.GET("/v3/kafka", hc.handleClusterList) hc.router.GET("/v3/kafka/:cluster", hc.handleClusterDetail) diff --git a/core/internal/httpserver/prometheus.go b/core/internal/httpserver/prometheus.go new file mode 100644 index 00000000..7b129457 --- /dev/null +++ b/core/internal/httpserver/prometheus.go @@ -0,0 +1,290 @@ +/* Copyright 2017 LinkedIn Corp. 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. + */ + +package httpserver + +import ( + "fmt" + "github.com/julienschmidt/httprouter" + "github.com/linkedin/Burrow/core/protocol" + "net/http" + "os" + "sort" + "strconv" + "strings" +) + +type httpPrometheusMetric struct { + Name string + Value float64 + Labels *map[string]string + + Help string + Type string // counter, gauge, histogram, summary, untyped + + Replacer *strings.Replacer +} + +func (pm *httpPrometheusMetric) escape(s string) string { + if pm.Replacer == nil { + pm.Replacer = strings.NewReplacer("\\", `\\`, "\n", `\n`, "\"", `\"`) + } + return pm.Replacer.Replace(s) +} + +func (pm *httpPrometheusMetric) String() string { + metric := pm.escape(pm.Name) + "{" + if pm.Labels != nil && len(*pm.Labels) > 0 { + labelNameList := make([]string, len(*pm.Labels)) + labelNumber := 0 + + for labelName := range *pm.Labels { + labelNameList[labelNumber] = labelName + labelNumber++ + } + + sort.Strings(labelNameList) + + for _, labelName := range labelNameList { + labelValue := (*pm.Labels)[labelName] + metric += fmt.Sprintf("%s=\"%s\",", pm.escape(labelName), pm.escape(labelValue)) + } + } + metric = strings.TrimRight(metric, ",") + metric += fmt.Sprintf("} %f\n", pm.Value) + return metric +} + +func (pm *httpPrometheusMetric) Comment() string { + var comment = "" + if len(pm.Help) > 0 { + comment += fmt.Sprintf("# HELP %s %s\n", pm.escape(pm.Name), pm.escape(pm.Help)) + } + if len(pm.Type) > 0 { + comment += fmt.Sprintf("# TYPE %s %s\n", pm.escape(pm.Name), pm.escape(pm.Type)) + } + return comment +} + +func (pm *httpPrometheusMetric) MarshalText() ([]byte, error) { + return []byte(pm.String()), nil +} + +func (hc *Coordinator) handlePrometheusMetrics(w http.ResponseWriter, r *http.Request, params httprouter.Params) { + // Fetch cluster list from the storage module + requestFetchClusters := &protocol.StorageRequest{ + RequestType: protocol.StorageFetchClusters, + Reply: make(chan interface{}), + } + hc.App.StorageChannel <- requestFetchClusters + responseFetchClusters := <-requestFetchClusters.Reply + + replacer := strings.NewReplacer("\\", `\\`, "\n", `\n`, "\"", `\"`) + host, _ := os.Hostname() + hostLabel := replacer.Replace(host) + + var metric httpPrometheusMetric + metricList := make([]httpPrometheusMetric, 0) + + for _, cluster := range responseFetchClusters.([]string) { + clusterLabel := replacer.Replace(cluster) + + // Consumer Group List + + requestFetchConsumers := &protocol.StorageRequest{ + RequestType: protocol.StorageFetchConsumers, + Cluster: cluster, + Reply: make(chan interface{}), + } + hc.App.StorageChannel <- requestFetchConsumers + responseFetchConsumers := <-requestFetchConsumers.Reply + + // Consumer Group Status + + for _, consumerGroup := range responseFetchConsumers.([]string) { + + // request the consumer group status structure + requestConsumerGroupStatus := &protocol.EvaluatorRequest{ + Cluster: cluster, + Group: consumerGroup, + ShowAll: true, + Reply: make(chan *protocol.ConsumerGroupStatus), + } + hc.App.EvaluatorChannel <- requestConsumerGroupStatus + responseConsumerGroupStatus := <-requestConsumerGroupStatus.Reply + + // skip the consumer group if the data have not been fully gathered + if responseConsumerGroupStatus.Complete < 1.0 { + continue + } + + consumerGroupLabels := map[string]string{ + "host": hostLabel, + "cluster": clusterLabel, + "consumer_group": replacer.Replace(consumerGroup), + "consumer_group_status": replacer.Replace(responseConsumerGroupStatus.Status.String()), + } + + metric = httpPrometheusMetric{ + Name: "burrow_consumer_group_total_lag", + Value: float64(responseConsumerGroupStatus.TotalLag), + Help: "The sum of all consumed partition current lag values for the group.", + Type: "gauge", + Labels: &consumerGroupLabels, + } + metricList = append(metricList, metric) + + metric = httpPrometheusMetric{ + Name: "burrow_consumer_group_total_partitions", + Value: float64(responseConsumerGroupStatus.TotalPartitions), + Help: "The total count of all partitions that this groups has committed offsets for across all consumed topics. It may not be the same as the total number of partitions consumed by the group, if Burrow has not seen commits for all partitions yet.", + Type: "gauge", + Labels: &consumerGroupLabels, + } + metricList = append(metricList, metric) + + metric = httpPrometheusMetric{ + Name: "burrow_consumer_group_status_code", + Value: float64(responseConsumerGroupStatus.Status), + Help: "The status code of the consumer group. It is calculated from the highest status for the individual partitions. (Codes are index of this list starting from zero: NOTFOUND, OK, WARN, ERR, STOP, STALL, REWIND).", + Type: "gauge", + Labels: &consumerGroupLabels, + } + metricList = append(metricList, metric) + + if responseConsumerGroupStatus.Maxlag != nil { + metric = httpPrometheusMetric{ + Name: "burrow_consumer_group_max_lag", + Value: float64(responseConsumerGroupStatus.Maxlag.CurrentLag), + Help: "The current lag value of a partition consumed by this group which has the highest lag.", + Type: "gauge", + Labels: &consumerGroupLabels, + } + metricList = append(metricList, metric) + } + + // Consumer Group Partition Status + + for _, partitionStatus := range responseConsumerGroupStatus.Partitions { + + // skip the consumer group partition if the data have not been fully gathered + if partitionStatus.Complete < 1.0 { + continue + } + + consumerGroupPartitionLabels := map[string]string{ + "host": hostLabel, + "cluster": clusterLabel, + "consumer_group": replacer.Replace(consumerGroup), + "consumer_group_status": replacer.Replace(responseConsumerGroupStatus.Status.String()), + "topic": replacer.Replace(partitionStatus.Topic), + "partition": strconv.Itoa(int(partitionStatus.Partition)), + "partition_status": replacer.Replace(partitionStatus.Status.String()), + "owner": replacer.Replace(partitionStatus.Owner), + "client_id": replacer.Replace(partitionStatus.ClientID), + } + + metric = httpPrometheusMetric{ + Name: "burrow_consumer_group_partition_current_lag", + Value: float64(partitionStatus.CurrentLag), + Help: "The current number of messages that the consumer is behind for this partition. This is calculated using the last committed offset and the current broker end offset.", + Type: "gauge", + Labels: &consumerGroupPartitionLabels, + } + metricList = append(metricList, metric) + + metric = httpPrometheusMetric{ + Name: "burrow_consumer_group_partition_status_code", + Value: float64(partitionStatus.Status), + Help: "The status code of the consumer group partition. (Codes are index of this list starting from zero: NOTFOUND, OK, WARN, ERR, STOP, STALL, REWIND).", + Type: "gauge", + Labels: &consumerGroupPartitionLabels, + } + metricList = append(metricList, metric) + + metric = httpPrometheusMetric{ + Name: "burrow_consumer_group_partition_latest_offset", + Value: float64(partitionStatus.End.Offset), + Help: "The latest offset value that Burrow is storing for this partition.", + Type: "counter", + Labels: &consumerGroupPartitionLabels, + } + metricList = append(metricList, metric) + + } + } + + // Topic List + + requestFetchTopics := &protocol.StorageRequest{ + RequestType: protocol.StorageFetchTopics, + Cluster: cluster, + Reply: make(chan interface{}), + } + hc.App.StorageChannel <- requestFetchTopics + responseFetchTopics := <-requestFetchTopics.Reply + + // Topic Status + + for _, topic := range responseFetchTopics.([]string) { + + requestTopicDetail := &protocol.StorageRequest{ + RequestType: protocol.StorageFetchTopic, + Cluster: cluster, + Topic: topic, + Reply: make(chan interface{}), + } + hc.App.StorageChannel <- requestTopicDetail + responseTopicDetail := <-requestTopicDetail.Reply + + for partitionNumber, partitionOffset := range responseTopicDetail.([]int64) { + + topicDetailLabels := map[string]string{ + "host": hostLabel, + "cluster": clusterLabel, + "topic": replacer.Replace(topic), + "partition": strconv.Itoa(partitionNumber), + } + + metric = httpPrometheusMetric{ + Name: "burrow_topic_partition_offset", + Value: float64(partitionOffset), + Help: "The offset of the partition in this topic.", + Type: "counter", + Labels: &topicDetailLabels, + } + metricList = append(metricList, metric) + + } + + } + + } + + // Output + + w.Header().Set("Content-Type", "text/plain") + w.Header().Set("Access-Control-Allow-Origin", "*") + w.WriteHeader(http.StatusOK) + + sort.Slice(metricList, func(a, b int) bool { + return metricList[a].Name < metricList[b].Name + }) + + previousMetricName := "" + for _, metric := range metricList { + if metric.Name != previousMetricName { + _, _ = w.Write([]byte("\n" + metric.Comment())) + } + previousMetricName = metric.Name + _, _ = w.Write([]byte(metric.String())) + } + +}