Metrics
This topic will describe the monitoring metrics in Nebula Dashboard.
Machine
Note
- All the machine metrics listed below are for the Linux operating system.
- The default unit for Disk and Network is byte. The unit changes with the data magnitude as the page displays. For example, when the flow is less than 1 KB/s, the unit is Bytes/s.
- For all versions of Dashboard Enterprise Edition, the memory occupied by Buff and Cache will not be counted in the memory usage.
CPU
| Parameter |
Description |
cpu_utilization |
The percentage of used CPU. |
cpu_idle |
The percentage of idled CPU. |
cpu_wait |
The percentage of CPU waiting for IO operations. |
cpu_user |
The percentage of CPU used by users. |
cpu_system |
The percentage of CPU used by the system. |
Memory
| Parameter |
Description |
memory_utilization |
The percentage of used memory. |
memory_used |
The memory space used (including caches). |
memory_actual_used |
The memory space used (not including caches). |
memory_free |
The memory space available. |
Load
| Parameter |
Description |
load_1m |
The average load of the system in the last 1 minute. |
load_5m |
The average load of the system in the last 5 minutes. |
load_15m |
The average load of the system in the last 15 minutes. |
Disk
| Parameter |
Description |
disk_used |
The disk space used. |
disk_free |
The disk space available. |
disk_readbytes |
The number of bytes that the system reads in the disk per second. |
disk_writebytes |
The number of bytes that the system writes in the disk per second. |
disk_readiops |
The number of read queries that the disk receives per second. |
disk_writeiops |
The number of write queries that the disk receives per second. |
inode_utilization |
The percentage of used inode. |
Network
| Parameter |
Description |
network_in_rate |
The number of bytes that the network card receives per second. |
network_out_rate |
The number of bytes that the network card sends out per second. |
network_in_errs |
The number of wrong bytes that the network card receives per second. |
network_out_errs |
The number of wrong bytes that the network card sends out per second. |
network_in_packets |
The number of data packages that the network card receives per second. |
network_out_packets |
The number of data packages that the network card sends out per second. |
Service
Period
The period is the time range of counting metrics. It currently supports 5 seconds, 60 seconds, 600 seconds, and 3600 seconds, which respectively represent the last 5 seconds, the last 1 minute, the last 10 minutes, and the last 1 hour.
Metric methods
| Parameter |
Description |
rate |
The average rate of operations per second in a period. |
sum |
The sum of operations in the period. |
avg |
The average latency in the cycle. |
P75 |
The 75th percentile latency. |
P95 |
The 95th percentile latency. |
P99 |
The 99th percentile latency. |
P999 |
The 99.9th percentile latency. |
Graph
| Parameter |
Description |
num_queries |
The number of queries. |
num_slow_queries |
The number of slow queries. |
query_latency_us |
The average latency of queries. |
slow_query_latency_us |
The average latency of slow queries. |
num_query_errors |
The number of queries in error. |
| Parameter |
Description |
heartbeat_latency_us |
The latency of heartbeats. |
num_heartbeats |
The number of heartbeats. |
Storage
| Parameter |
Description |
add_edges_latency_us |
The average latency of adding edges. |
add_vertices_latency_us |
The average latency of adding vertices. |
delete_edges_latency_us |
The average latency of deleting edges. |
delete_vertices_latency_us |
The average latency of deleting vertices. |
get_neighbors_latency_us |
The average latency of querying neighbors. |
Last update:
March 13, 2023