Prasad, as far as I can tell, it depends on how 'parallel' the vector is in relation to being 'perfectly' parallel. I ran a test with a 1 unit long cantilevered beam, and if the orientation vector is perfectly aligned (using <1.0, 0.0, 0.0 >), it gives the same warning message you saw, plus the following FATAL error:
*** USER FATAL MESSAGE 2026 (EMG)
ELEMENT 1 GEOMETRY YIELDS UNREASONABLE MATRIX.
I found that as long as there was a slightly non-parallel vector, it would only give the warning message... It only gave the warning only when I used <1.0, 1.0E-12, 0.0> and gave the warning plus the fatal with <1.0, 1.0E-16, 0.0>.
In all the warning only test cases, the results were identical. You would have to validate your own model before assuming the results are good.
I tested as far back as v2010 with the same results.
No matter what the results are, it's a poor practice to define orientations that are close to parallel with the element. I don't know why anyone would want to...it would make it very difficult for another user to figure out if the orientations are defined correctly.