using lassoglm for logistic regression

I have 3 predictors that are numerical and 2 predictors that are categorical. The expected classes are simply yes or no, so I reckon glmfit will be a good function to use. Upon using glmfit, a warning message appeared:

Warning: X is ill conditioned, or the model is overparametrised, and some coefficients are not identifiable. You should use caution in making predictions.

This suggests that MAYBE some of the predictors might be correlated.

I am now trying to use glmfit with regularisation (lassoglm) and I want to ask the following:

1. I am given 100 lambdas. I understand these are geometric sequence of lambdas. (but of what values? and how do i choose which lambda?)
2. How do I know which predictors are correlated and must be ignored?