There is a general problem with substructure matching with azoles (pyrazoles, triazoles, tetrazoles, etc.) bearing a hydrogen on one of nitrogens, ie. 1,2,3-triazole (picture attached).
When MOL is converted into smiles then you get c1cnn[nH]1. When you search this pattern against indexed smiles then you won`t obtain any structures bearing a substituent on the nitrogen. A workaround is to search for c1cnnn1 but this require intentional string substitution in smiles from [nH] to n by the script that performs the search.
This is not a bug. If you add a "H" on the nitrogen, you are indicating you do not want substitution at this position.