Hazelcast Grafana App - padogrid/padogrid GitHub Wiki

◀️ Hazelcast Desktop App :link: Hazelcast jet_demo App ▶️

The grafana app provides a simple and quick way to integrate Hazelcast with Grafana by including several commands for accessing Grafana along with pre-configured Hazelcast dashboards. For example, you can import the included dashboards into Grafana with the import_grafana command and monitor the entire cluster in a single view.

1. Installing Grafana App

The Grafana app is part of the padogrid distribution. Run the create_app to install it in your workspace.

create_app -product hazelcast -app grafana

2. Enabling/Disabling Prometheus

Support for Prometheus is enabled by default for all Hazelcast clusters created by the create_cluster command. You can enable or disable it by setting the prometheus.enabled property in each cluster's etc/cluster.properties file as follows:

# etc/cluster.properties
# By default, Prometheus is enabled.
prometheus.enabled=true

3. Enabling Prometheus for Management Center

By default, the clusters created by create_cluster also enable Clustered JMX for Management Center. To enable Prometheus for Management Center, however, you must explicitly configure the Management Center as shown below.

# Change directory to the cluster's directory
cd_cluster
vi bash_sh/setenv.sh

Add the following in the bin_sh/setenv.sh file. Note that you also need to configure Prometheus to include the Management Center as one of its targets as described in the 5.2. Prometheus section.

# Port number can be arbitrary but the same port number must
# be included in the Prometheus configuration file.
MC_JAVA_OPTS="-Dhazelcast.mc.prometheusExporter.enabled=true \
              -Dhazelcast.mc.prometheusExporter.port=2222"

Restart the Management Center.

stop_mc
start_mc

4. JMX Exporter Agent

Grafana is supported via the JMX exporter provided by Prometheus. It is already included in the distribution and fully integrated with Hazelcast out of the box. You can learn more about the exporter from the following site:

URL: https://github.com/prometheus/jmx_exporter

5. Required Software

There are three (3) required software components that you must install before you can use the grafana app.

  • JQ - JSON Processor
  • Prometheus
  • Grafana
  • curl

5.1. JQ - JSON Processor

The PadoGrid grafana app relies on JQ to process JSON objects.

URL: https://stedolan.github.io/jq

Include it in your PATH:

# Assuming jq is placed in your home bin directory:
export PATH=~/bin:$PATH

5.2. Prometheus

5.2.1. Installing and Starting Prometheus

PadoGrid includes integrated support for Prometheus and Grafana, greatly simplifying the installation and management steps.

Install Prometheus using install_padogrid and update_padogrid:

install_padogrid -product prometheus
update_padogrid -product prometheus

Start Prometheus from the grafana app:

cd_app grafana/bin_sh
./start_prometheus

5.2.2. Monitoring Prometheus

You can monitor Prometheus from your browser:

URL: http://localhost:9090

To view a complete list of metrics:

curl -G http://localhost:9090/federate -d 'match[]={__name__!=""}'

5.3. Grafana

5.3.1. Installing and StartingGrafana

Install Grafana using install_padogird and update_padogrid:

# Install Grafana Enterprise
install_padogrid -product grafana-enterprise
update_padogrid -product grafana-enterprise

# Or install Grafana OSS
install_padogrid -product grafana-oss
update_padogrid -product grafana-oss

Start Grafana from the grafana app:

cd_app grafana/bin_sh
./start_grafana

5.3.2. Monitoring Grafana

Once Grafana is running, use your web browser to set the user account as follows:

URL: http://localhost:3000

User Name: admin
Password: admin

The grafana app has been preconfigured with the above user name and password. If you have a different account, then you can change them in setenv.sh. Note that the included commands require the user with administration privileges

5.4. Cygwin: curl

❗️ If you are running this app in the Windows environment then make sure to install curl from Cygwin. Other implementations may not work properly.

6. Importing Dashboards

The dashboards are organized by Grafana folders and they can be found in the following directory:

cd_app grafana
ls etc/dashboards

The following dashboard folders are included in this distribution.

  • Hazelcast-perf_test - A set of dashboards for monitoring the metrics specific to the perf_test app.
  • Hazelcast - A set of dashboards for monitoring a single Hazelcast cluster.
  • HazelcastDual - A set of dashboards for comparing two (2) Hazelcast clusters side-by-side.
  • HazelcastAll - A set of dashboards for federating multiple Hazelcast clusters.
  • Padogrid - A set of dashboards for PadoGrid specific dashboards.

If Hazelcast is running on Kubernetes, then you need to change the default label value to namespace or service by executing padogrid_update_cluster_templating as follows.

./padogrid_update_cluster_templating -label namespace

The above command changes the default label, job, to namespace so that the Hazelcast metrics can be filtered by Kubernetes namespace. For non-Kubernetes, padogrid_update_cluster_templating is not necessary since the dashboards are predefined with the default label, job.

To import all folders, run import_folder -all as follows.

cd bin_sh
./import_folder -all

To import the default folder, i.e., Hazelcast-perf_test, run import_folder as follows.

cd bin_sh
./import_folder

To import folders individually, specify the -folder option.

# To import the Hazelcast-perf_test folder in 'etc/dashboards':
./import_folder -folder Hazelcast-perf_test

# To import the Hazelcast folder in 'etc/dashboards':
./import_folder -folder Hazelcast

# To import the HazelcastDual folder in 'etc/dashboards':
./import_folder -folder HazelcastDual

# To import the HazelcastAll folder in 'etc/dashboards':
./import_folder -folder HazelcastAll

# To import the Padogrid folder in 'etc/dashboards':
./import_folder -folder Padogrid

# To import all folders in 'etc/dashboards':
./import_folder -all

6.1. App: perf_test

The Hazelcast-perf_test folder includes the perf_test app dashboards. To view data in these dashboards, you must run the perf_test ingestion and transaction scripts. The following command creates the default app, perf_test, in your workspace.

create_app

For perf_test details, see perf_test README.

6.2. Hazelcast Dashboards

The Hazelcast* folders contain the main dashboard with the prefix, 00Main. The main dashboard is the main console for navigating all the dashboards in the respective folder. It has the layout similar to the Management Center as shown in the screen shot in Section 10.2 Hazelcast Folder.

To quickly activate the dashboards with data, you can run perf_test with the group-workflow-*.properties files.

The Hazelcast dasboards support multiple clusters. If you are running Prometheus in PadoGrid, then to include multiple clusters, you must add them in the etc/prometheus.yml file. You can use the the included etc/prometheus-clusters.yml as an example. This file configures two (2) Hazelast clusters named, 'myhz' and 'myhz2'.

7. Exporting Dashboards

You can also export your dashboards to use them as backup or templates by executing the export_folder command.

# Export all folders found in Grafana. By default, the dashboards are 
# exported in the export/ directory. You can change it in setenv.sh.
./export_folder -all

8. Creating Dashboard Templates

You must convert the exported dashboards to templates by executing the export_to_template command before you can import them back to Grafana. This is due to the Grafana dependency of non-unique local IDs. The generated templates are portable and can be imported into any instance of Grafana using the import_folder command.

# Convert the exported folders to templates. The templates are placed in
# the templates/ directory. See the usage for details.
./export_to_template

8.1. Synchronizing Hazelcast* Folders

8. Synchronizing Folders

If you made changes to dashboards from the browser and want to save them in the local file system in the form of templates then execute the padogrid_sync_folders command. This command exports all of the dashboards in the Hazelcast* folders, creates templates, applies the required Grafana variables to the templates, and re-imports the templates to Grafana.

# Using 'job':
./padogrid_sync_folders -label job
# To make dashboards editable
./padogrid_sync_folders -label job -editable

# For Kubernetes:
./padogrid_sync_folders -label namespace
# To make dashboards editable
./padogrid_sync_folders -label namespace -editable

9. Other Commands

The bin_sh directory contains many useful commands for working with dashboards. You can display the usage of each command by specifying the -? option as shown below.

./create_folder -?
Usage:
   ./create_folder [-folder <folder-name>] [-?]

   Creates the specified Grafana folder.

Default: ./create_folder -folder Hazelcast-perf_test

10. Screenshots

10.1. padogrid_perf_test Folder

perf_test Folder

10.2. Hazelcast Folder

Hazelcast Folder

11. Teardown

cd_app grafana/bin_sh
./stop_grafana
./stop_prometheus

◀️ Hazelcast Desktop App :link: Hazelcast jet_demo App ▶️