Increment the resource of JVM and the number of threads in Travis instead of default

[asdf2014]

Increment the resource of JVM and the number of threads in Travis instead of default. Maybe, it could reduce the times that happen Timeout Problem even resolve the problem.

{
  "language": "java",
  "jdk": "oraclejdk8",
  "before_install": "sudo echo \"MAVEN_OPTS='-Xmx2048m -Xms1024m'\" > ~/.mavenrc",
  "script": [
    "mvn clean test -B -Pparallel-test -Dmaven.fork.count=2 -T 1C && mvn clean -Pstrict compile test-compile -B -T 1C"
  ],
  "sudo": "required",
  "cache": {
    "directories": [
      "$HOME/.m2"
    ]
  },
  "global_env": "MAVEN_OPTS=-Xmx2048m -Xms1024m",
  "group": "stable",
  "dist": "precise",
  "os": "linux"
}

[asdf2014]

It's weird! In my branch, it works. 😓
https://github.com/asdf2014/druid/commits/travis

[drcrallen]

Sudo required moves back onto the vm launch method instead of container. Why is that needed?

[asdf2014]

@drcrallen Because changing the .mavenrc file need the sudo permission.

before_install: sudo echo "MAVEN_OPTS='-Xmx2048m -Xms1024m'" > ~/.mavenrc

[drcrallen]

We (administrators) should be able to add environment variables to the build. Is the reason for the expanded heap so that tests can run in parallel?

[asdf2014]

Yeah, the default resource of Travis is few, i think it may be the cause of the task often Timeout.

[drcrallen]

Sudo on echo does not do what you are wanting. This needs thorough testing before acceptance.

[asdf2014]

I think so, maybe my assumption has some problem.

[jihoonson]

Hi @asdf2014, I think we need to figure out what the root cause is and devise an workaround if inevitable.
#4142 is to show the cause of failure when DruidCoordinatorTest fails, and just merged to master. Let's see what happens if this problem occurs again.

[asdf2014]

@jihoonson Wonderful! Look forward to.

[asdf2014]

@drcrallen @jihoonson Travis is good enough. Now close this PR. Thanks for your comments. 👍