statistics-to-grafana

Analyse Apache Geode & GemFire Statistics With Grafana

Apache Geode can collect Statistics about the distributed system and persist it in archive files. The StatisticsToGrafana tool helps to load the archive files in time-series databases such as InfluxDB used to feed Grafana dashboards. Use the stack to build comprehensive Grafana dashboards to visualize, analyse and compare statistics data from different cluster members or even different Geode clusters.

Build

Get the source code from github

git clone https://github.com/tzolov/geode-dashboard.git

From within the geode-dashboard/statistics-to-grafana directory run

mvn clean install

Quick Start

Build Grafana dashboard to analyze the statistics files collected on two Geode instances (e.g. members). Grafana and InfluxDB have to be installed first. Samples below expect InfluxDB on http://localhost:8086 and Grafana on http://localhost:3000.

Load statistics into an InfluxDB database

If the target InfluxDB database (GeodeArchive for example) doesn't exist yet create one. You can do it through the influx command-line tool:

influx> create database GeodeArchive
influx> show databases

or with the help of the --cleanDatabaseOnLoad=true parameter. Later removes and creates again the time-series database.

Load the server1_StatisticsArchiveFile.gfs file containing historical statistics for server1 cluster member into influx database GeodeArchive. Set the measurement archiveMember tag to ‘server1’.

java -jar ./target/statistics-to-grafana-0.0.2-SNAPSHOT.jar \
   --influxUrl=http://localhost:8086 \
   --influxDatabaseName=GeodeArchive \
   --geodeMemberName=server1 \
   --archiveFile=server1_StatisticsArchiveFile.gfs

Load the server2_StatisticsArchiveFile.gfs file containing historical statistics for server2 cluster member in the same influx database GeodeArchive:

java -jar ./target/statistics-to-grafana-0.0.2-SNAPSHOT.jar \
   --influxUrl=http://localhost:8086 \
   --influxDatabaseName=GeodeArchive \
   --geodeMemberName=server2 \
   --archiveFile=server2_StatisticsArchiveFile.gfs

Note that the statistics from both files is loaded in the same database: GeodeArchive! The archiveMember meta-tags (server1 and server2) is used to distinct both series. Use Grafana Templating to define multi-select query variables for the archiveMember meta-atag (e.g. show tag values with key="archiveMember"). This allow mixing and matching measurments from different archive files.

Complete list of statistics-to-grafana parameters:

Property Name	Default Value	Description
influxUrl	http://localhost:8086	InfulxDB connection URL
influxUser	admin	InfuxDB connection username
influxPassword	admin	InfluxDB connection password
cleanDatabaseOnLoad	false	If set the target TSDB will be (re)created on every statistics load
influxRetentionPolicy	autogen	InfluxDB retention policy
archiveFile	None	File path to a single Geode statistics archive file. Note: every Geode instance (e.g. member) generates a statistics file. It is helpful to copy all the statistics files from all members into one directory so that you can easily load the files into InfluxDb.
influxDatabaseName	GeodeArchive	Database to load the statistics into. Same database can be used to load statistics from multiple archive files. Use the geodeMemerName to distinct the time series
geodeMemberName	None	Name that uniquely identifies the Geode instance (e.g. member) which produced the statistics being loaded
allowedStatTypes	None	Comma separated list of Statistic Type Names to import. If empty all statistic is read
skipZeroValuesTimeSeries	true	When true the time series that contain only zero values will not be imported

Build Grafana Dashboard

	Create datasource:GeodeArchive to the GeodeArchive Influx database. Set approriate InfluxDB URL and credentials.
	Create new Grafana dashboard named ArchiveDashboard
	Inside ArchiveDashboard create member tag Template (show tag values with key="archiveMember"). Set GeodeArchive as Data Source and enable the Multi-value option. Later allows to select the members to show statistics for.
	Create Number-Of-Threads-Per-Member panel to plot the number of threads used by each Geode member. Pick GeodeArchive as panel's data source. Set the retention policy to autogen. In the FROM dropdown clause pick the VMStats:vmStats measurement.
	Adjust the time range to reflect the interval when the statistic is collected