Network meta-analysis (NMA) also known as mixed treatment or multiple treatment comparisons, is commonly used to incorporate direct and indirect evidence comparing treatments. This is an extension to meta-analysis which seeks to estimate the combined estimate of treatment comparisons from multiple studies. With recent advances in methods and software, Bayesian approaches to NMA have become quite popular. Current inconsistency detection in NMA is usually based on contrast-based (CB) models. We look at an arm-based (AB) random effects model, where we detect discrepancy of direct and indirect evidence for comparing treatments. We compare sources of inconsistency identified by our approach and existing loop-based CB methods using real and simulated datasets and demonstrate that our methods can offer powerful inconsistency detection. After detection of inconsistency is done we try to perform some diagnostics to network meta-analysis to see if the trials that are causing the inconsistencies are just outliers or influential.