Quick Start
RediB is a Java-based end-to-end testing framework. So, you will need to write your test cases In Java, or languages that can use Java libraries like the ones that can run on JVM, e.g. Scala. RediB can be used alongside the popular testing frameworks in your programming language of choice e.g. JUnit in Java. Here, we use Java and JUnit . We also use Maven as the build system.
Adding dependencies
First, create a simple Maven application and add RediB’s dependency to your pom file.
<dependency> <groupId>io.redit</groupId> <artifactId>redit</artifactId> <version>0.1.0</version> </dependency>
Also add failsafe plugin to your pom file to be able to run integration tests.
<project>
[...]
<build>
<plugins>
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-failsafe-plugin</artifactId>
<version>3.0.0-M3</version>
<executions>
<execution>
<goals>
<goal>integration-test</goal>
<goal>verify</goal>
</goals>
</execution>
</executions>
</plugin>
</plugins>
</build>
[...]
</project>
Creating a Dockerfile
Next, you need to create a Dockerfile for your application and that Dockerfile should add any dependency that may be
needed by your application. In case you want to use the network partition capability
of RediB, you need to install iptables
package as well. Network delay and loss will also need the iproute
package to be installed. Here, we assume the application under test is written in Java.
So, we create a Dockerfile in the docker/Dockerfile
address with the following content:
FROM openjdk:8-stretch
RUN apt update && apt install -y iptables iproute
Important
In case you are using Docker Toolbox (and consequently boot2docker) on Windows or Mac, be aware that your currently
installed boot2docker image may be missing sched_netem
kernel module which is included in most of the
linux distributions and is needed for tc
command in the iproute
package to work. So, unless you upgrade your
boot2docker image (normally through running docker-machine upgrade [machine_name]
, you won’t be able to use the
network operation capabilities of RediB.
Adding a Test Case
Now, create a JUnit integration test case (ending with IT so failsafe picks it up) in the project’s test directory. Here, we provide an example for testing the situation of multithread. You can find the full code in the Redit project.
1 public class MultithreadTest {
2 public static final Logger logger = LoggerFactory.getLogger(MultithreadTest.class);
3
4 @Test
5 public void simpleDefinition() throws DeploymentVerificationException, RuntimeEngineException, TimeoutException, WorkspaceException {
6 Deployment deployment = Deployment.builder("sample-multithread")
7 // Service Definitions
8 .withServiceFromJvmClasspath("s1", "target/classes", "**commons-io*.jar")
9 .startCommand("java -cp ${REDIT_JVM_CLASSPATH} io.redit.samples.multithread.Main")
10 .dockerImageName("redit/sample-multithread")
11 .dockerFileAddress("../sample-multithread/docker/Dockerfile", true)
12 .logFile("/var/log/sample1")
13 .logDirectory("/var/log/samples")
14 .serviceType(ServiceType.JAVA).and()
15 // Node Definitions
16 .withNode("n1", "s1")
17 .stackTrace("e1", "io.redit.samples.multithread.Main.helloWorld1," +
18 "io.redit.samples.multithread.Main.hello")
19 .stackTrace("e2", "io.redit.samples.multithread.Main.helloWorld2," +
20 "io.redit.samples.multithread.Main.helloWorld")
21 .stackTrace("e3", "io.redit.samples.multithread.Main.helloWorld3," +
22 "io.redit.samples.multithread.Main.hello")
23 .stackTrace("e4", "org.apache.commons.io.FilenameUtils.normalize")
24 .blockBefore("bbe2", "e2")
25 .unblockBefore("ubbe2", "e2")
26 .garbageCollection("g1")
27 .and()
28 .withNode("n2", "s1").offOnStartup().and()
29 .withNode("n3", "s1").and()
30 .withNode("n4", "s1").and()
31 // Test Case Events
32 .testCaseEvents("x1", "x2")
33 // Run Sequence Definition
34 .runSequence("bbe2 * e1 * ubbe2 * x1 * e2 * e3 * x2 * e4")
35 .sharedDirectory("/redit")
36 .build();
37
38 ReditRunner runner = ReditRunner.run(deployment);
39 // Starting node n2
40 runner.runtime().enforceOrder("x1",10, () -> runner.runtime().startNode("n2"));
41 // Adding new nodes to the deployed environment
42 runner.addNode(Node.limitedBuilder("n5", "s1"));
43 // Imposing overlapping network partitions
44 NetPart netPart1 = NetPart.partitions("n1","n2").connect(1, NetPart.REST, false).build();
45 NetPart netPart2 = NetPart.partitions("n1","n2,n3").connect(1, NetPart.REST).build();
46 runner.runtime().networkPartition(netPart1);
47 runner.runtime().networkPartition(netPart2);
48 // Imposing 10 secs of clock drift in node n1
49 runner.runtime().clockDrift("n1", -10000);
50 // Applying network delay and loss on node n2 before restarting it
51 runner.runtime().networkOperation("n2", NetOp.delay(50).jitter(10), NetOp.loss(30));
52 // removing the first network partition and restarting node n2
53 runner.runtime().enforceOrder("x2", 10, () -> {
54 runner.runtime().removeNetworkPartition(netPart1);
55 runner.runtime().restartNode("n2", 10);
56 });
57 // removing the second network partition
58 runner.runtime().removeNetworkPartition(netPart2);
59 // Applying different kinds of network operations in different orders
60 runner.runtime().networkOperation("n1", NetOp.delay(100).jitter(10), NetOp.loss(30),
61 NetOp.removeDelay(), NetOp.delay(10).jitter(4), NetOp.removeLoss(),
62 NetOp.removeDelay(), NetOp.loss(20), NetOp.removeLoss());
63 // Waiting for the run sequence to be completed
64 runner.runtime().waitForRunSequenceCompletion(60, 20);
65 }
66}
Each RediB test case should start with defining a new Deployment
object. A deployment definition consists of a a set
of service and node definitions. A Service is a node template and defines the docker image for the node, the start bash
command, required environment variables, common paths, etc. for a specific type of node.
Line 8-14 defines service1
service. Line 9 defines the start command for the node, and in this case, it is using the start.sh
bash file and it feeding it with -conf /config.cfg
argument. This
config file will be provided separately through node definitions later. Line 14 concludes the service definition by marking it as a Java application.
If the programming language in use is listed in ServiceType
enum, make sure to mark your application with the right
ServiceType
.
Important
If your program runs on JVM and your programming language in use is not listed in the ServiceType
enum, just choose ServiceType.Java
as the service type.
Lines 16-30 defines four nodes named n1
, n2
, n3
and n4
from service1
service and is adding a separate local config file
to each of them which will be located at the same target address /config.cfg
. Most of the service configuration can be
overriden by nodes.
Line 38 starts the defined deployment and line 64 stops the deployment after all tests are executed.
Line 42 shows how to start node while running. In
this case, a clock dirft of 100ms will be applied to node n1
. Line 44-47 shows how a network partition can be defined
and imposed. Here, each of the nodes will be in a separate partition. Line 45 shows an example of imposing network delay and loss on all the interfaces of a specific node.
Here, a network delay from a uniform distribution with mean=100 and variance=10 will be applied on n1
and 30% of the
packets will be lost.
Logger Configuration
RediB uses SLF4J for logging. As such, you can configure your logging tool of choice. A sample configuration with Logback can be like this:
<?xml version="1.0" encoding="UTF-8"?>
<configuration>
<appender name="Console" class="ch.qos.logback.core.ConsoleAppender">
<layout class="ch.qos.logback.classic.PatternLayout">
<Pattern>%d{HH:mm:ss.SSS} [%thread] %-5level %logger{36} - %msg%n</Pattern>
</layout>
</appender>
<logger name="io.redit" level="DEBUG"/>
<root level="ERROR">
<appender-ref ref="Console" />
</root>
</configuration>
Running the Test Case
Finally, to run the test cases, run the following bash command:
$ mvn clean verify