效能計量
您應該收集計量,以了解叢集及在叢集中執行之應用程式的效能。 針對 Service Fabric 叢集,建議您收集下列效能計數器。
節點
針對您叢集中的機器,請考慮收集下列效能計數器,以進一步了解每部機器上的負載,並做出適當的叢集調整決策。
計數器類別 | 計數器名稱 |
---|---|
Logical Disk | 邏輯磁碟可用空間 |
PhysicalDisk(per Disk) | Avg.磁碟讀取佇列長度 |
PhysicalDisk(per Disk) | Avg.磁碟寫入佇列長度 |
PhysicalDisk(per Disk) | 平均磁碟秒數/讀取 |
PhysicalDisk(per Disk) | Avg.Disk sec/Write |
PhysicalDisk(per Disk) | 磁碟讀取/秒 |
PhysicalDisk(per Disk) | 磁碟讀取位元組/秒 |
PhysicalDisk(per Disk) | Disk Writes/sec |
PhysicalDisk(per Disk) | 磁碟寫入位元組/秒 |
記憶體 | 可用的 MB |
PagingFile | % 使用量 |
Processor(Total) | % Processor Time |
Process (per service) | % Processor Time |
Process (per service) | 識別碼處理序 |
Process (per service) | 私用位元組 |
Process (per service) | 對話計數 |
Process (per service) | 虛擬位元組 |
Process (per service) | 工作集 |
Process (per service) | 工作集 - 私用 |
Network Interface(all-instances) | 記錄的位元組 |
Network Interface(all-instances) | 傳送的位元組 |
Network Interface(all-instances) | 位元組總數 |
Network Interface(all-instances) | 輸出佇列長度 |
Network Interface(all-instances) | 已丟棄輸出封包 |
Network Interface(all-instances) | 已丟棄接收封包 |
Network Interface(all-instances) | 輸出封包錯誤 |
Network Interface(all-instances) | 已收到封包錯誤 |
.NET 應用程式與服務
如果您要將 .NET 服務部署到叢集,請收集下列計數器。
計數器類別 | 計數器名稱 |
---|---|
.NET CLR 記憶體 (每一服務) | 處理序識別碼 |
.NET CLR 記憶體 (每一服務) | # Total committed Bytes |
.NET CLR 記憶體 (每一服務) | # Total reserved Bytes |
.NET CLR 記憶體 (每一服務) | # Bytes in all Heaps |
.NET CLR 記憶體 (每一服務) | Large Object Heap size |
.NET CLR 記憶體 (每一服務) | # GC Handles |
.NET CLR 記憶體 (每一服務) | # Gen 0 Collections |
.NET CLR 記憶體 (每一服務) | # Gen 1 Collections |
.NET CLR 記憶體 (每一服務) | # Gen 2 Collections |
.NET CLR 記憶體 (每一服務) | % Time in GC |
Service Fabric 的自訂效能計數器
Service Fabric 可產生大量的自訂效能計數器。 若已安裝 SDK,便可在 Windows 機器上的 [效能監視器] 應用程式 ([開始] > [效能監視器]) 中看到完整的清單。
在您部署到叢集的應用程式中,如果您使用 Reliable Actors,請從 Service Fabric Actor
和 Service Fabric Actor Method
類別新增計數器 (請參閱 Service Fabric Reliable Actors 診斷)。
如果您使用 Reliable Services 或服務遠端,我們同樣擁有應從其中收集計數器的 Service Fabric Service
和 Service Fabric Service Method
計數器類別,請參閱使用服務遠端進行監視和可靠服務效能計數器。
如果您使用「可靠的集合」,建議您從 Service Fabric Transactional Replicator
新增 Avg. Transaction ms/Commit
,以收集每一交易的平均認可延遲計量。
下一步
- 深入了解 Service Fabric 中的平台層級的事件產生
- 透過 Log Analytics 代理程式收集效能計量