Two different cases to clarify first:

If you actually have some data which does not know the exact value, but do know that it is greater than 3, then it is called censored data.

If those data greater than 3 are actually missed, i.e. you do not know the "sample size" and you are unaware of those data, then you are facing a truncated data.

In either cases you may use MLE or any other reasonable / specially devised method for weibull to estimate the parameters.