HDFS is an open-source distributed filesystem developed to manage large volumes of data distributed on low-cost, high-end hardware. In HDFS architecture, there are two types of nodes called NameNode and DataNode, the first type is responsible for managing data distribution and the other one is responsible for storing data and exchanging it with clients. Moein provides HDFS cluster monitoring by observing both types of nodes. Performance indicators of each type are as follow:

Performance indicators and parameters of NameNode monitored by Moien are:

NameNode General KPIs:

1. Total Create Operations
2. Total Created Files And Directories
3. Total Appended Files
4. Number Of GetBlockLocations Operations
5. Number Of Rename Operations
6. Number Of Directory Listing Operations
7. Total Delete Operations
8. Total Deleted Files And Directories
9. Total Get FileInfo Operations
10. Number Of AddBlock Operations
11. Number Of GetAdditionalDatanode Operations
12. Number Of CreateSymlink Operations
13. Number Of GetLinkTarget Operations
14. Number Of Files And Directories Listed By Directory Listing Operations
15. Number Of Processing Block Reports
16. Number Of Journal Transactions
17. Journal Transactions Average Time
18. Number Of Journal Syncs
19. Journal Syncs Average Time
20. Number Of Batched Transactions In Sync
21. Safe Mode Time
22. FSImage Load Time
23. Number Of Edits Downloads From Secondary Node
24. Average Edits Download Time
25. Total Number Of FSImage Downloads From Secondary Node
26. Average FSImage Download Time
27. Total Number Of FSImage Uploads To Secondary Node
28. Average FSImage Upload Time To Secondary Node

Cluster KPIs:

1. Total Configured Capacity
2. DFS Used Capacity
3. DFS Remaining Capacity
4. Current Number Of Connections
5. Current Number Of Allocated Blocks In The System
6. Current Number Of Files And Directories
7. Number Of Blocks Pending To Be Replicated
8. Number Of Blocks Under Replicated
9. Number Of Blocks Scheduled For Replications
10. Current FS State
11. Number Of Live Datanodes
12. Number Of Dead Datanodes

KPIs of DataNodes managed by NameNode:

1. DFS Capacity Usage
2. DFS Remaining Capacity Percentage
3. Non DFS Used Capacity
4. Cluster ID
5. Block Pool ID
6. Version
7. Block Pool Used Space
8. Block Pool Used Space Percentage
9. Total Blocks
10. Total Files
11. Number Of Missing Blocks

Live Node Information:

1. Datanode Name
2. Number Of Blocks
3. DFS Used Capacity
4. Admin State
5. Non DFS Used Capacity
6. Last Update Time
7. Total Capacity
8. DFS Capacity Used Percentage
9. DFS Remaining Capacity Percentage

Dead Node Information:

1. Datanode Name
2. Is Decommissioned?
3. Dead Time

Monitoring NameNode is enough for HDFS cluster monitoring. However, if DataNode monitoring is desired separately, the following indicators are provided for each DataNode by Moein:

General Datanode Parameters:

1. Number Of Active Connections
2. Version
3. Cluster ID
4. Namenode Address
5. DFS Free Space
6. DFS Used Space
7. DFS Used Space Percentage

DataNode KPIs:

1. Number Of Bytes Written To Datanode
2. Number Of Bytes Read From Datanode
3. Data Written Rate To Datanode
4. Data Read Rate From Datanode
5. Number Of Blocks Written To Datanode
6. Number Of Blocks Read From Datanode
7. Number Of Replicated Blocks
8. Number Of Removed Blocks
9. Number Of Verified Blocks
10. Number Of Verification Failures
11. Local Client Read Operations
12. Remote Client Read Operations
13. Local Client Write Operations
14. Remote Client Write Operations
15. Number Of Get Local Path Info Operations
16. Number Of Fsyncs
17. Number Of Volume Failures
18. Number Of Block Read Operations
19. Block Read Average Time
20. Number Of Block Write Operations
21. Block Write Average Time
22. Number Of Block Checksum Operations
23. Block Checksum Average Time
24. Number Of Block Copy Operations
25. Block Copy Average Time
26. Number Of Block Replace Operations
27. Block Replace Average Time
28. Number Of Heartbeats
29. Heartbeats Average Time
30. Number Of Block Reports Operations
31. Block Reports Average Time
32. Number Of ACK RTT
33. ACK RTT Average Time
34. Number Of Flushes
35. Flushes Average Time
36. Number Of Fsync Nanos
37. Fsync Nanos Average Time
38. Number Of Sending Packet Operations
39. Sending Packets Average Time

JVM KPIs:

HDFS is developed in Java and requires JVM to run. Details of JVM‌ indicators are provided in the Environments section.

Communication Protocols:

• REST
• JMX