timeout error comes up more and more frequently during project CI

[huangyuqi]

Now, the PR checks unsuccessfully more frequently in continuous-integration/travis-ci/pr, with the error log:

Your test run exceeded 120 minutes. 
One possible solution is to split up your test run.

The reason is obvious,

• the project of Heapster becomes bigger day by day. we need to run CI check go test parallelly.
• And in current version, the CI script run go test in all file path in the project, including vendor; we need to find out the file path which includes *_test.go, then execute go test.

HEAPSTER_ROOT=./
heapster::test-unit::find_dirs() {
  (
    cd ${HEAPSTER_ROOT}
    find . -not \( \
        \( \
          -path './_artifacts/*' \
          -o -path './docs/*' \
          -o -path './Godeps/*' \
          -o -path './deploy/*' \
          -o -path './version/*' \
          -o -path './vendor/*'\
        \) -prune \
      \) -name '*_test.go' -print0 | xargs -0n1 dirname | sed 's|^\./||' | sort -u
  )
}
set -- $(heapster::test-unit::find_dirs)

@piosz

[huangyuqi]

I will bring a fix PR with second way.

[piosz]

As a first step we can just get rid off test-unit-cov

[AlmogBaku]

@piosz @huangyuqi this can be solved by using glide #1314 (or specifically see my review comment in #1335 )

[huangyuqi]

@piosz I agree. there are duplicate contents between test-unit and test-unit-cov.