Knowledge Graphs are increasingly becoming popular for a variety of downstream tasks like Question Answering and Information Retrieval. However, the Knowledge Graphs are often incomplete, thus leading to poor performance. As a result, there has been a lot of interest in the task of Knowledge Base Completion. More recently, Graph Neural Networks have been used to capture structural information inherently stored in these Knowledge Graphs and have been shown to achieve SOTA performance across a variety of datasets. In this survey, we understand the various strengths and weaknesses of the proposed methodology and try to find new exciting research problems in this area that require further investigation.