Non-Functional Requirements (NFR) should be defined in the early stages of the software development process, driving developers to make important design decisions. Neglecting NFRs may lead developers to create systems that are difficult to maintain and do not meet user expectations. Despite its importance, the discussion of NFRs is often ad-hoc and scattered through multiple sources, limiting developers’ awareness of NFRs. In that scenario, Pull Request (PR) discussions can be used, thus providing a centralized platform for comprehensive NFR discussions. However, existing studies do not explore this important source of information in open-source software development, which developers widely use to discuss software requirements. In this study, we report an investigation of NFR discussions available in PR of repositories of the Spring ecosystem. We collected, manually curated, and analyzed PR discussions addressing four categories of NFR: maintainability, security, performance, and robustness. We found that more than 77% of the discussions related to NFRs are triggered in the PR title and/or description, indicating that developers are often provided with sufficient information straightway. We observed that discussions surrounding these PRs tend to address the introduction of a code change or explain some anomaly regarding a particular NFR. To gain more knowledge of NFR discussions, we investigated developers involved in discussions and fixing NFR issues. For that, we performed an in-depth analysis of \devsAnalyzed developers that stood out in collaborating with the mapped PRs. We discovered that the developers who discuss NFRs often open and review most PRs related to NFRs. To complement this analysis, we applied a survey with \devsSurvey to gather their perceptions on NFR discussions. By observing how developers approach NFRs and participate in discussions, we documented the best practices and strategies newcomers can use to address NFRs effectively. Through the characterization of NFR discussions, we aim to pave the way for developing automated tools to identify NFRs better. Finally, we provide a curated dataset of 1,533 PR discussions classified with NFR presence.