The assumption of normality of the sample is for a one sample t test. You have a two sample and the expectation is for normality of the populations. This means your sample does not have to be normally distributed (this works on the central limit theorem).

A nice write up of assumptions of t-tests can be found here.