Generally, For F test, it's upper tail test. It means, Reject H0 if F>F(a,df1,df2). and also you can use P-value to decide your decision. If P-value >a, fail to reject H0, otherwise, reject H0.
I am confused by the results you got b/c you got different decisions if you use F value and P-value methods...
Is there something wrong with your program or data....? don't know.