http://en.wikipedia.org/wiki/Regression_testingThe term "regression testing" is usually in the context of software or system development, not system health monitoring.
In general, regression testing requires that you define a set of test cases for defined requirements and known already-solved problems. When you make a change to the system (e.g. you find and fix a new problem), you run the regression test suite to make sure you haven't accidentally re-introduced an older problem when fixing the new one, or violated one of the basic requirements.
Then you add a test for the problem you just fixed, so in the future the test suite will alert you if this same problem reappears (e.g. something has caused your fix to become ineffective).
Regression tests are, by their nature, very specific to their subject.
> test of cpu activity,memory activity,process activity..kernel logs .hardware logs.. disk space ...swap space
This sounds like system health monitoring, not regression testing.
There are various system health monitoring solutions available. Some of them are even open source/freeware. Why are you re-inventing the wheel? Have you examined the available solutions and found them somehow inadequate or not applicable? Why?
If you really are thinking about regression testing and not generic health monitoring, please define the unit you're primarily wishing to test: is it some application, the Linux OS kernel, or the system hardware?
Performance regression testing will also require using your own judgement: if the new version runs the test faster but uses more memory and/or disk space to do that, is that an acceptable trade-off or not? How much is too much?
MK
MK
