Fixed ballast test cases by using cross-platform lib

[owais]

This should make the test work on the following platforms:

• FreeBSD i386/amd64/arm
• Linux i386/amd64/arm
• Windows/amd64
• Darwin i386/amd64
• OpenBSD amd64

[codecov-io]

Codecov Report

Merging #61 into master will decrease coverage by 0.08%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master      #61      +/-   ##
==========================================
- Coverage   69.67%   69.58%   -0.09%     
==========================================
  Files          86       86              
  Lines        5606     5606              
==========================================
- Hits         3906     3901       -5     
- Misses       1481     1485       +4     
- Partials      219      220       +1

Impacted Files	Coverage Δ
cmd/occollector/app/collector/collector.go	79.26% <0%> (-3.05%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 2ad0e1d...7c5664b. Read the comment docs.

[tigrannajaryan]

Strange that it shows reduction in codecov. Also the build result does not seem to be reported from Travis to Github properly.

[owais]

@tigrannajaryan it had a typo linu/amd64 instead of linux so the test case never ran :)

[pjanotti]

Change itself looks good, need to check if the counter for Darwin is correct per offline conversation.

[owais]

Update

The test was working across platforms but not so reliably on Travis CI. It seems the testsuite and the Go compiler toolchain consuming very different amount of memory across different environments. We could get it to work but I don't think it would be worth the effort especially considering it would likely break in new environments and we would have to maintain it. I think it's better to have this as a standalone test that spins up a production collector instance and monitors it's memory usage. @pjanotti suggested we use the perf testbed for this and I think that makes a lot of sense. As a result this PR is just disabling the ballast test for now. We'll remove it once it is implemented as a perf test.

@tigrannajaryan @songy23

[owais]

I actually removed it completely to get rid of the dependency as there was no point in carrying it for nothing.

[tigrannajaryan]

I think it's better to have this as a standalone test that spins up a production collector instance and monitor's it's memory usage. @pjanotti suggested we use the perf testbed for this and I think that makes a lot of sense.

I agree. This makes a lot of sense as an E2E test.