keycloak-scim/testsuite/performance

Keycloak Performance Testsuite

Requirements:

• Bash 2.05+
• Maven 3.1.1+
• Keycloak server distribution installed in the local Maven repository. To do this run mvn install -Pdistribution from the root of the Keycloak project.

Docker Compose Provisioner

• Docker 1.13+
• Docker Compose 1.14+

Getting started for the impatient

Here's how to perform a simple tests run:

# Clone keycloak repository if you don't have it yet
# git clone https://github.com/keycloak/keycloak.git

# Build Keycloak distribution - needed to build docker image with latest Keycloak server
mvn clean install -DskipTests -Pdistribution

# Now build, provision and run the test
cd testsuite/performance
mvn clean install

# Make sure your Docker daemon is running THEN
mvn verify -Pprovision
mvn verify -Pgenerate-data -Ddataset=100u2c -DnumOfWorkers=10 -DhashIterations=100
mvn verify -Ptest -Ddataset=100u2c -DusersPerSec=2 -DrampUpPeriod=10 -DuserThinkTime=0 -DbadLoginAttempts=1 -DrefreshTokenCount=1 -DmeasurementPeriod=60 -DfilterResults=true

Now open the generated report in a browser - the link to .html file is displayed at the end of the test.

After the test run you may want to tear down the docker instances for the next run to be able to import data:

mvn verify -Pteardown

You can perform all phases in a single run:

mvn verify -Pprovision,generate-data,test,teardown -Ddataset=100u2c -DnumOfWorkers=10 -DhashIterations=100 -DusersPerSec=4 -DrampUpPeriod=10

Note: The order in which maven profiles are listed does not determine the order in which profile related plugins are executed. teardown profile always executes last.

Keep reading for more information.

Provisioning

Provision

Provisioners

Depending on the target environment different provisioners may be used. Provisioner can be selected via property -Dprovisioner=PROVISIONER.

Default value is docker-compose which is intended for testing on a local docker host. This is currently the only implemented option. See README.docker-compose.md for more details.

Deployment Types

Different types of deployment can be provisioned. The default deployment is singlenode with only a single instance of Keycloak server and a database. Additional options are cluster and crossdc which can be enabled with a profile (see below).

Usage

Usage: mvn verify -P provision[,DEPLOYMENT_PROFILE] [-Dprovisioning.properties=NAMED_PROPERTY_SET].

The properties are loaded from tests/parameters/provisioning/${provisioning.properties}.properties file. Individual parameters can be overriden from command line via -D params.

Default property set is docker-compose/4cpus/singlenode.

To load a custom properties file specify -Dprovisioning.properties.file=ABSOLUTE_PATH_TO_FILE instead of -Dprovisioning.properties. This file needs to contain all properties required by the specific combination of provisioner and deployment type. See examples in folder tests/parameters/provisioning/docker-compose/4cpus.

Available parameters are described in README.provisioning-parameters.md.

Examples:

• Provision a single-node deployment with docker-compose: mvn verify -P provision
• Provision a cluster deployment with docker-compose: mvn verify -P provision,cluster
• Provision a cluster deployment with docker-compose, overriding some properties: mvn verify -P provision,cluster -Dkeycloak.scale=2 -Dlb.worker.task-max-threads=32
• Provision a cross-DC deployment with docker-compose: mvn verify -P provision,crossdc
• Provision a cross-DC deployment with docker-compose using a custom properties file: mvn verify -P provision,crossdc -Dprovisioning.properties.file=/tmp/custom-crossdc.properties

Provisioned System

The provision operation will produce a provisioned-system.properties inside the tests/target directory with information about the provisioned system such as the type of deployment and URLs of Keycloak servers and load balancers. This information is then used by operations generate-data, import-dump, test, teardown.

Provisioning operation is idempotent for a specific combination of provisioner+deployment. When running multiple times the system will be simply updated based on the new parameters. However when switching between different provisioiners or deployment types it is always necessary to tear down the currently running system.

Note: When switching deployment type from singlenode or cluster to crossdc (or the other way around) it is necessary to update the generated Keycloak server configuration (inside keycloak/target directory) by adding a clean goal to the provisioning command like so: mvn clean verify -Pprovision …. It is not necessary to update this configuration when switching between singlenode and cluster deployments.

Collect Artifacts

Usage: mvn verify -Pcollect

Collects artifacts such as logs from the provisioned system and stores them in tests/target/collected-artifacts/${deployment}-TIMESTAMP/. When used in combination with teardown (see below) the artifacts are collected just before the system is torn down.

Teardown

Usage: mvn verify -Pteardown [-Dprovisioner=<PROVISIONER>]

Note: Unless the provisioned system has been properly torn down the maven build will not allow a cleanup of the tests/target directory because it contains the provisioned-system.properties with information about the still-running system.

Testing

Generate Test Data

Usage: mvn verify -P generate-data [-Ddataset=NAMED_PROPERTY_SET] [-DnumOfWorkers=N]. The default dataset is 2u2c. Workers default to 1.

The parameters are loaded from tests/parameters/datasets/${dataset}.properties file. Individual properties can be overriden from command line via -D params.

To use a custom properties file specify -Ddataset.properties.file=ABSOLUTE_PATH_TO_FILE instead of -Ddataset.

To generate data using a different version of Keycloak Admin Client set property -Dserver.version=SERVER_VERSION to match the version of the provisioned server.

Dataset Parameters

Property	Description	Value in the Default Dataset
numOfRealms	Number of realms to be created.	1
usersPerRealm	Number of users per realm.	2
clientsPerRealm	Number of clients per realm.	2
realmRoles	Number of realm-roles per realm.	2
realmRolesPerUser	Number of realm-roles assigned to a created user. Has to be less than or equal to realmRoles.	2
clientRolesPerUser	Number of client-roles assigned to a created user. Has to be less than or equal to clientsPerRealm * clientRolesPerClient.	2
clientRolesPerClient	Number of client-roles per created client.	2
hashIterations	Number of password hashing iterations.	27500

Examples:

• Generate the default dataset. mvn verify -P generate-data
• Generate the 100u2c dataset. mvn verify -P generate-data -Ddataset=100u2c
• Generate the 100u2c dataset but override some parameters. mvn verify -P generate-data -Ddataset=100u2c -DclientRolesPerUser=5 -DclientRolesPerClient=5

Export Database

To export the generated data to a data-dump file enable profile -P export-dump. This will create a ${DATASET}.sql.gz file next to the dataset properties file.

Example: mvn verify -P generate-data,export-dump -Ddataset=100u2c

Import Database

To import data from an existing data-dump file use profile -P import-dump.

Example: mvn verify -P import-dump -Ddataset=100u2c

If the dump file doesn't exist locally the script will attempt to download it from ${db.dump.download.site} which defaults to https://downloads.jboss.org/keycloak-qe/${server.version} with server.version defaulting to ${project.version} from pom.xml.

Warning: Don't override dataset parameters (with -Dparam=value) when running export/import because then the contents of dump file might not match the properties file.

Run Tests

Usage: mvn verify -P test [-Dtest.properties=NAMED_PROPERTY_SET]. Default property set is oidc-login-logout.

The parameters are loaded from tests/parameters/test/${test.properties}.properties file. Individual properties can be overriden from command line via -D params.

To use a custom properties file specify -Dtest.properties.file=ABSOLUTE_PATH_TO_FILE instead of -Dtest.properties.

Dataset

When running the tests it is necessary to define the dataset to be used.

Parameter	Description	Default Value
dataset	Name of the dataset to use. Individual parameters can be overriden from CLI. For details see the section above.	default
sequentialRealmsFrom	Use sequential realm iteration starting from specific index. Must be lower than numOfRealms parameter from dataset properties. Useful for user registration scenario.	-1 random iteration
sequentialUsersFrom	Use sequential user iteration starting from specific index. Must be lower than usersPerRealm parameter from dataset properties. Useful for user registration scenario.	-1 random iteration

Common Test Run Parameters

Parameter	Description	Default Value
gatling.simulationClass	Classname of the simulation to be run.	keycloak.OIDCLoginAndLogoutSimulation
usersPerSec	Arrival rate of new users per second. Can be a floating point number.	1.0 for OIDCLoginAndLogoutSimulation, 0.2 for AdminConsoleSimulation
rampUpPeriod	Period during which the users will be ramped up. (seconds)	15
warmUpPeriod	Period with steady number of users intended for the system under test to warm up. (seconds)	15
measurementPeriod	A measurement period after the system is warmed up. (seconds)	30
filterResults	Whether to filter out requests which are outside of the measurementPeriod.	false
userThinkTime	Pause between individual scenario steps.	5
refreshTokenPeriod	Period after which token should be refreshed.	10
logoutPct	Percentage of users who should log out at the end of scenario.	100

Test Assertion	Description	Default Value
maxFailedRequests	Maximum number of failed requests.	0
maxMeanReponseTime	Maximum mean response time of all requests.	300

Test Run Parameters specific to OIDCLoginAndLogoutSimulation

Parameter	Description	Default Value
badLoginAttempts		0
refreshTokenCount		0

Examples:

• Run test with default test and dataset parameters:

mvn verify -P test

• Run test specific test and dataset parameters:

mvn verify -P test -Dtest.properties=oidc-login-logout -Ddataset=100u2c

• Run test with specific test and dataset parameters, overriding some from command line:

mvn verify -P test -Dtest.properties=admin-console -Ddataset=100u2c -DrampUpPeriod=30 -DwarmUpPeriod=60 -DusersPerSec=0.3

Running OIDCRegisterAndLogoutSimulation

Running the user registration simulation requires a different approach to dataset and how it's iterated.

• It requires sequential iteration instead of the default random one.
• In case some users are already registered it requires starting the iteration from a specific index .

Example A:

1. Generate dataset with 0 users: mvn verify -P generate-data -DusersPerRealm=0
2. Run the registration test:

mvn verify -P test -D test.properties=oidc-register-logout -DsequentialUsersFrom=0 -DusersPerRealm=<MAX_EXPECTED_REGISTRATIONS>

Example B:

1. Generate or import dataset with 100 users: mvn verify -P generate-data -Ddataset=100u2c. This will create 1 realm and users 0-99.
2. Run the registration test starting from user 100:

mvn verify -P test -D test.properties=oidc-register-logout -DsequentialUsersFrom=100 -DusersPerRealm=<MAX_EXPECTED_REGISTRATIONS>

Testing with HTTPS

If the provisioned server is secured with HTTPS it is possible to set the truststore which contains the server certificate. The truststore is used in phases generate-data and test.

Usage: mvn verify -P generate-data,test -DtrustStore=<PATH_TO_TRUSTSTORE> -DtrustStorePassword=<TRUSTSTORE_PASSWORD>

To automatically generate the truststore file run a utility script tests/create-truststore.sh HOST:PORT [TRUSTSTORE_PASSWORD]. The script requires openssl and keytool (included in JDK).

Example: tests/create-truststore.sh localhost:8443 truststorepass

Monitoring

JMX

To enable access to JMX on the WildFly-backed services set properties management.user and management.user.password during the provisioning phase.

JVisualVM

• Set JBOSS_HOME variable to point to a valid WildFly 10+ installation.
• Start JVisualVM with jboss-client.jar on classpath: ./jvisualvm --cp:a $JBOSS_HOME/bin/client/jboss-client.jar.
• Add a local JMX connection: service:jmx:remote+http://localhost:9990. ^[*]
• Check "Use security credentials" and set admin:admin. (The default credentials can be overriden by providing env. variables DEBUG_USER and DEBUG_USER_PASSWORD to the container.)
• Open the added connection.

[*] For singlenode this points to the JMX console of the Keycloak server. To get the connection URLs for cluster or crossdc deployments see the JMX section in the generated provisioned-system.properties file.

• Property keycloak.frontend.servers.jmx contains JMX URLs of the Load Balancers.
• Property keycloak.backend.servers.jmx contains JMX URLs of the clustered Keycloak servers.
• Property infinispan.servers.jmx contains JMX URLs of the Infinispan servers, in Cross-DC deployment.

Docker Monitoring

There is a docker-based solution for monitoring CPU, memory and network usage per container. It uses CAdvisor service to export container metrics into InfluxDB time series database, and Grafana web app to query the DB and present results as graphs.

• To enable run: mvn verify -Pmonitoring
• To disable run: mvn verify -Pmonitoring-off[,delete-monitoring-data]. By default the monitoring history is preserved. If you wish to delete it enable the delete-monitoring-data profile when turning monitoring off.

To view monitoring dashboard open Grafana UI at: http://localhost:3000/dashboard/file/resource-usage-combined.json.

Sysstat metrics

To enable recording of sysstat metrics use -Psar. This will run the sar command during the test and process its binary output to produce textual and CSV files with CPU utilisation stats. To also enable creation of PNG charts use -Psar,gnuplot. For this to work Gnuplot needs to be installed on the machine. To compress the binary output with bzip add -Dbzip=true to the commandline.

Results will be stored in folder: tests/target/sar.

Developing tests in IntelliJ IDEA

Add scala support to IDEA

Install the correct Scala SDK

First you need to install Scala SDK. In Scala land it's very important that all libraries used are compatible with specific version of Scala. Gatling version that we use uses Scala version 2.11.7. In order to avoid conflicts between Scala used by IDEA, and Scala dependencies in pom.xml it's very important to use that same version of Scala SDK for development.

Thus, it's best to download and install this SDK version

Install IntelliJ's official Scala plugin

Open Preferences in IntelliJ. Type 'plugins' in the search box. In the right pane click on 'Install JetBrains plugin'. Type 'scala' in the search box, and click Install button of the Scala plugin.

Run OIDCLoginAndLogoutSimulation from IntelliJ

Make sure that performance maven profile is enabled for IDEA to treat performance directory as a project module.

You may also need to rebuild the module in IDEA for scala objects to become available.

Then find Engine object In ProjectExplorer (you can use ctrl-N / cmd-O). Right click on class name and select Run or Debug as if it was a JUnit tests.

You'll have to edit a test configuration, and set 'VM options' to a list of -Dkey=value pairs to override default configuration values in TestConfig class.

Make sure to set 'Use classpath of module' to 'performance-test'.

When tests are executed via maven, the Engine object is not used. It exists only for running tests in IDE.

If test startup fails due to not being able to find the test classes try reimporting the 'performance' module from pom.xml (right click on 'performance' directory, select 'Maven' at the bottom of context menu, then 'Reimport')

If you want to run a different simulation - not DefaultSimulation - you can edit Engine object source, or create another Engine object for a different simulation.

Troubleshoot

Verbose logging

You can find logback-test.xml file in tests/src/test/resources directory. This files contains logging information in log4j xml format. Root logger is by default set to WARN, but if you want to increase verbosity you can change it to DEBUG or INFO.