I am trying to test for interaction effect in my logistic model where outcome is dummy and the two interacting term is categorical. One of the independent term is dummy and other one has 4 categories. When I tried to run it using y x1#x2, gives me the error below. The model runs but it doesn't give odds ratio for 0 1, 0 2, and 1 4 combination. I have attached the image from the results. urbanrural is dummy where 0 means rural and 1 means urban and cat_white has 4 categories with <5%, 6%-25%, 26%-50%, and >50%. Also, it is not using 209 observations, I have 30513 obs but the model is using 30304 obs. Please suggest what is the best possible way to handle this, so that it gives me odds ratio for all possible combinations.

note: 0.urbanrural#1.cat_white != 0 predicts failure perfectly
0.urbanrural#1.cat_white dropped and 62 obs not used

note: 0.urbanrural#2.cat_white != 0 predicts failure perfectly
0.urbanrural#2.cat_white dropped and 147 obs not used

note: 1.urbanrural#4.cat_white omitted because of collinearity