Working with the Aerospike Monitoring Stack dashboards

Using locations, types, and labels

Metrics data from Aerospike is described by the location API that generated the metric. Metrics are further described by type for the scope, data type, and duration of the data. Each metric also may have labels that allow for sorting and filtering.

For a complete list of metrics available, see the Metrics Reference.

Location

Each metric is described for a location that refers to the API that generated the metric and corresponds to the asadm show command parameter. The Aerospike Prometheus Exporter prepends the metric with a string to identify the location. For example, the metric conns_in_use, generated for the Users location, becomes aerospike_users_conns_in_use when exported from a node. Following are the locations mapped to the string prefixes prepended by the Exporter. Use a metric with its exported prefix when creating alerts or displaying the metric in a dashboard.

Location	Exported prefix
Bins	aerospike_namespace_
Namespace	aerospike_namespace_
Sets	aerospike_sets_
Sindex	aerospike_sets_
Statistics	aerospike_node_stats_
Users	aerospike_users_
XDR	aerospike_xdr_
XDR - DC	aerospike_xdr_

Type

The Metrics Reference also describes the types that apply to a specific metric. These are shown with colored highlighting in the upper right corner of the metric's reference entry. Types describe the Aerospike server edition, the data duration, and the data type for the metric.

Edition

When the enterprise edition label is shown, the metric is available only in Aerospike Enterprise Edition. Otherwise, the metric is available for all Aerospike editions.

Duration

• instantaneous - the data is measured at a specific instant. For example, the memory_used_bytes metric is an instantaneous measure of the total bytes of memory used.

• cumulative - the data is accumulated from the last restart. For example, the batch_sub_delete_error metric reports the accumulated subtransactions that failed with an error since the last restart.

• moving average - the data reported is a moving average of all values since the last restart. For example, the device_compression_ratio measures the average compressed size to uncompressed size ratio since the last restart.

Data type

• integer

• decimal

• boolean

Labels

Metrics may also be described with labels in queries on the command line or in a dashboard. You can sort and filter metric data based on these labels. Following are some of the more prominent labels used in Aerospike:

• ns - corresponds to the specific namespace of the metric.

• set - a group of records within a collection.

• job_name - a label you can set on the Aerospike Prometheus Exporter. aerospike is the default. The label is customizable.

• cluster_name - corresponds to the name set for in your cluster configuration. It is required for monitoring multiple clusters.

• service - IP:port of the nodes.

• instance - IP:port of the Exporter on the node.

Example PromQL query used in the dashboard

An example query to compute and view transactions per second (TPS) for Aerospike read operations is detailed here.

The query passes the following parameters to the rate() function:

• The metric name aerospike_namespace_client_read_success.
• The job label as aerospike.
• The cluster_name label as aerospike-release-aerospike-enterprise.

rate(aerospike_namespace_client_read_success{job="aerospike", cluster_name="aerospike-release-aerospike-enterprise"}[1m])+
rate(aerospike_namespace_client_read_not_found{job="aerospike", cluster_name="aerospike-release-aerospike-enterprise"}[1m])

Query results

The query returns the per-second average rate of increase in TPS for Aerospike reads.

The following results are displayed in the dashboard.

Dashboard example console

Example graph

Dashboard example graph

Other operations on the dashboard

In addition to queries, you can also check the status of scrape targets, alert rules, configuration, and more.

Dashboard status menu