網上看了很多，發現對於Skywalking支持哪些指標名稱metrics，官方文檔跟博客幾乎都是指明了一個路徑，沒有人詳細的解釋，支持哪些指標，這些指標的作用又有什么作用，導致大家自定義指標的時候有很多困難。
所以這里給大家總結下，如有錯誤，及時指正：
Skywalking的oap指標存放在：/apache-skywalking-apm-bin-es78/config/oal/*.oap 目錄下
先來看第一個oap文件:
core.oal
1 / All scope metrics 2 all_percentile = from(All.latency).percentile(10); // Multiple values including p50, p75, p90, p95, p99 3 all_heatmap = from(All.latency).histogram(100, 20); // 4 5 // Service scope metrics 服務 6 service_resp_time = from(Service.latency).longAvg(); // 服務的平均響應時間 7 service_sla = from(Service.*).percent(status == true); // 服務的請求成功率 8 service_cpm = from(Service.*).cpm(); //服務的每分鍾調用次數 9 service_percentile = from(Service.latency).percentile(10); // Multiple values including p50, p75, p90, p95, p99 10 service_apdex = from(Service.latency).apdex(name, status); // 服務的應用性能指標，apdex的衡量的是衡量滿意的響應時間與不滿意的響應時間的比率，默認的請求滿意時間是500ms 11 12 // Service relation scope metrics for topology 服務與服務間調用的調用度量指標 13 service_relation_client_cpm = from(ServiceRelation.*).filter(detectPoint == DetectPoint.CLIENT).cpm();//在客戶端檢測到的每分鍾調用次數 14 service_relation_server_cpm = from(ServiceRelation.*).filter(detectPoint == DetectPoint.SERVER).cpm();//在服務端檢測到的每分鍾調用的次數 15 service_relation_client_call_sla = from(ServiceRelation.*).filter(detectPoint == DetectPoint.CLIENT).percent(status == true);//在客戶端檢測到成功率 16 service_relation_server_call_sla = from(ServiceRelation.*).filter(detectPoint == DetectPoint.SERVER).percent(status == true);//在服務端檢測到的成功率 17 service_relation_client_resp_time = from(ServiceRelation.latency).filter(detectPoint == DetectPoint.CLIENT).longAvg();//在客戶端檢測到的平均響應時間 18 service_relation_server_resp_time = from(ServiceRelation.latency).filter(detectPoint == DetectPoint.SERVER).longAvg();//在服務端檢測到的平均響應時間 19 service_relation_client_percentile = from(ServiceRelation.latency).filter(detectPoint == DetectPoint.CLIENT).percentile(10); // Multiple values including p50, p75, p90, p95, p99 20 service_relation_server_percentile = from(ServiceRelation.latency).filter(detectPoint == DetectPoint.SERVER).percentile(10); // Multiple values including p50, p75, p90, p95, p99 21 22 // Service Instance relation scope metrics for topology 服務實例與服務實例之間的調用度量指標 23 service_instance_relation_client_cpm = from(ServiceInstanceRelation.*).filter(detectPoint == DetectPoint.CLIENT).cpm();//在客戶端實例檢測到的每分鍾調用次數 24 service_instance_relation_server_cpm = from(ServiceInstanceRelation.*).filter(detectPoint == DetectPoint.SERVER).cpm();//在服務端實例檢測到的每分鍾調用次數 25 service_instance_relation_client_call_sla = from(ServiceInstanceRelation.*).filter(detectPoint == DetectPoint.CLIENT).percent(status == true);//在客戶端實例檢測到的成功率 26 service_instance_relation_server_call_sla = from(ServiceInstanceRelation.*).filter(detectPoint == DetectPoint.SERVER).percent(status == true);//在服務端實例檢測到的成功率 27 service_instance_relation_client_resp_time = from(ServiceInstanceRelation.latency).filter(detectPoint == DetectPoint.CLIENT).longAvg();//在客戶端實例檢測到的平均響應時間 28 service_instance_relation_server_resp_time = from(ServiceInstanceRelation.latency).filter(detectPoint == DetectPoint.SERVER).longAvg();//在服務端實例檢測到的平均響應時間 29 service_instance_relation_client_percentile = from(ServiceInstanceRelation.latency).filter(detectPoint == DetectPoint.CLIENT).percentile(10); // Multiple values including p50, p75, p90, p95, p99 30 service_instance_relation_server_percentile = from(ServiceInstanceRelation.latency).filter(detectPoint == DetectPoint.SERVER).percentile(10); // Multiple values including p50, p75, p90, p95, p99 31 32 // Service Instance Scope metrics 33 service_instance_sla = from(ServiceInstance.*).percent(status == true);//服務實例的成功率 34 service_instance_resp_time= from(ServiceInstance.latency).longAvg();//服務實例的平均響應時間 35 service_instance_cpm = from(ServiceInstance.*).cpm();//服務實例的每分鍾調用次數 36 37 // Endpoint scope metrics 38 endpoint_cpm = from(Endpoint.*).cpm();//端點的每分鍾調用次數 39 endpoint_avg = from(Endpoint.latency).longAvg();//端口平均響應時間 40 endpoint_sla = from(Endpoint.*).percent(status == true);//端點的成功率 41 endpoint_percentile = from(Endpoint.latency).percentile(10); // Multiple values including p50, p75, p90, p95, p99 42 43 // Endpoint relation scope metrics 44 endpoint_relation_cpm = from(EndpointRelation.*).filter(detectPoint == DetectPoint.SERVER).cpm();//在服務端端點檢測到的每分鍾調用次數 45 endpoint_relation_resp_time = from(EndpointRelation.rpcLatency).filter(detectPoint == DetectPoint.SERVER).longAvg();//在服務端檢測到的rpc調用的平均耗時 46 endpoint_relation_sla = from(EndpointRelation.*).filter(detectPoint == DetectPoint.SERVER).percent(status == true);//在服務端檢測到的請求成功率 47 endpoint_relation_percentile = from(EndpointRelation.rpcLatency).filter(detectPoint == DetectPoint.SERVER).percentile(10); // Multiple values including p50, p75, p90, p95, p99 48 49 database_access_resp_time = from(DatabaseAccess.latency).longAvg();//數據庫的處理平均響應時間 50 database_access_sla = from(DatabaseAccess.*).percent(status == true);//數據庫的請求成功率 51 database_access_cpm = from(DatabaseAccess.*).cpm();//數據庫的每分鍾調用次數 52 database_access_percentile = from(DatabaseAccess.latency).percentile(10);
java-agent.oal
// JVM instance metrics instance_jvm_cpu = from(ServiceInstanceJVMCPU.usePercent).doubleAvg();//jvm 平均cpu耗時百分比 instance_jvm_memory_heap = from(ServiceInstanceJVMMemory.used).filter(heapStatus == true).longAvg();//jvm 堆空間的平均使用空間 instance_jvm_memory_noheap = from(ServiceInstanceJVMMemory.used).filter(heapStatus == false).longAvg();//jvm 非堆空間的平均使用空間 instance_jvm_memory_heap_max = from(ServiceInstanceJVMMemory.max).filter(heapStatus == true).longAvg();//jvm 最大堆內存的平均值 instance_jvm_memory_noheap_max = from(ServiceInstanceJVMMemory.max).filter(heapStatus == false).longAvg();//jvm 最大非堆內存的平均值 instance_jvm_young_gc_time = from(ServiceInstanceJVMGC.time).filter(phrase == GCPhrase.NEW).sum();//年輕代gc的耗時 instance_jvm_old_gc_time = from(ServiceInstanceJVMGC.time).filter(phrase == GCPhrase.OLD).sum();//老年代gc的耗時 instance_jvm_young_gc_count = from(ServiceInstanceJVMGC.count).filter(phrase == GCPhrase.NEW).sum();//年輕代gc的次數 instance_jvm_old_gc_count = from(ServiceInstanceJVMGC.count).filter(phrase == GCPhrase.OLD).sum();//老年代gc的次數 instance_jvm_thread_live_count = from(ServiceInstanceJVMThread.liveCount).longAvg();//存活的線程數 instance_jvm_thread_daemon_count = from(ServiceInstanceJVMThread.daemonCount).longAvg();//守護線程數 instance_jvm_thread_peak_count = from(ServiceInstanceJVMThread.peakCount).longAvg();//峰值線程數
告警的設置
rules: # 告警規則 名稱唯一 必須以_rule 結尾 service_resp_time_rule: # 度量名稱，只支持int long double metrics-name: service_resp_time # 操作符 op: ">" # 閾值 ms threshold: 1000 # 評估度量的時間長度 period: 10 # 度量有多少次符合告警條件后，才會觸發告警 count: 2 # 靜默時間 默認情況下，它和周期一樣，在同一個周期內只會觸發一次。 silence-period: 10 message: 服務【{name}】的平均響應時間在最近10分鍾內有2分鍾超過1秒 service_sla_rule: metrics-name: service_sla op: "<" threshold: 8000 period: 10 count: 2 silence-period: 10 message: 服務【{name}】的成功率在最近10分鍾內有2分鍾低於80％ composite-rules: # 規則名稱：在告警信息中顯示的唯一名稱，必須以_rule結尾 comp_rule: # 指定如何組成規則，支持&&, ||, ()操作符 expression: service_resp_time_rule && service_sla_rule message: 服務【{name}】在最近10分鍾內有2分鍾平均響應時間超過1秒並且成功率低於80％