/
deprecated

Ganglia metrics

Ganglia metrics is deprecated, replaced by Compute metrics (deprecated April 2023).

Ganglia metrics Compute metrics

The Ganglia distributed-monitoring system bundled on classic clusters, surfaced as a Ganglia UI link and a list of historical snapshots on the cluster's Metrics tab for watching hardware performance and health.

  • It arrived on Azure Databricks in the July 2019 platform release (version 2.101), advertised as a way to monitor cluster performance and health straight from the cluster details page.
  • Its historical view was a list of snapshots taken every 15 minutes, so a job that finished in 14 could leave behind exactly zero metrics files.
  • It only ever measured Spark container consumption, so whatever else a node was busy doing simply never appeared on its graphs.

Limitations: Only available on Databricks Runtime 12.2 LTS and below on AWS and Azure and never on Google Cloud; snapshots are taken every 15 minutes by default, adjustable through the DATABRICKS_GANGLIA_SNAPSHOT_PERIOD_MINUTES cluster environment variable.

Open in REbricked →
Category
Compute / BI
Deprecated
April 2023
Also known as
Ganglia, Ganglia UI
Verified
2026-09-11

Sources

Related in Compute / BI