To get started with the basics of querying time series data, see Prometheus - Querying - Basics.

🚧

To monitor Tamr, it is not required that the Prometheus Web UI port (default 31390) is exposed. However, if you'd like to use the Prometheus console for query development, you must expose this port.

Quick Reference

The following list includes examples of useful queries you can run.

CPU

100- (avg(rate(node_cpu{mode="idle"}[5m])) BY (instance) * 100)

Memory

avg_over_time(node_memory_Active[5m] ) / node_memory_MemTotal * 100

Disk

Percentage time of CPU spent on I/O waiting.

(avg(rate(node_cpu{mode="iowait"}[5m])) BY (instance) * 100)

Elasticsearch

Elasticsearch JVM memory usage.

elasticsearch_jvm_memory_used_bytes / elasticsearch_jvm_memory_max_bytes

Other Metrics

Other useful metrics are:

• node_disk_io_time_ms
• node_disk_read_time_ms
• node_disk_write_time_ms

Note: These three metrics are "counters" and you need to wrap them in the rate() function.