As an SRE I'd like to be able to monitor my IO performance closely. node_exporter allows for polling and calculating in-depth disk related metrics, such as queue depth and latency (IOPS don't mean much without latency after all). Throughput and average request size are also a nice-to-have.
As an SRE I'd like to be able to monitor my IO performance closely.
node_exporterallows for polling and calculating in-depth disk related metrics, such as queue depth and latency (IOPS don't mean much without latency after all). Throughput and average request size are also a nice-to-have.This panel can be used, or similar metrics added to our own panels: https://grafana.com/grafana/dashboards/11801 Here's a guide for building a good comprehensive panel, again, nothing bu node_exporter is used: https://devconnected.com/monitoring-disk-i-o-on-linux-with-the-node-exporter/