We analyse developer coordination in Kubernetes, a large open source project comparable in scale to a major enterprise software division. Over 11 years, 25,953 contributors created 46,768 issues and 339,332 comments, providing a complete record of collaboration rarely visible in proprietary settings. We examine how contributors organise themselves across technical domains and how coordination affects development outcomes. Our results show that collaboration is highly modular, with contributors interacting 42.9 times more often within domains than across them. Issues related to multiple domains take 4.19 times longer to resolve (due to both coordination complexity and issue difficulty), and coordination within the modules depends on a small subset of contributors. These findings demonstrate how large-scale coordination dependencies can be observed and quantified in open source development, offering a transferable tool for analyzing enterprise-scale software projects.