Performance of CMS Tracker Alignment during LHC Run 2 and Commissioning towards Run 3

The CMS experiment's innermost detector, the tracker, is designed to accurately measure the momentum of charged particles and reconstruct the primary vertices. As the operating conditions during data-taking change frequently, movements may occur in the substructures of the tracker, thereby necessitating regular updating of the detector geometry to describe the position, orientation, and curvature of the tracker modules. The process of determining the new geometry parameters is called tracker alignment. Tracker alignment is performed numerous times throughout the data-taking period using reconstructed tracks from collisions and cosmic rays data and further fine-tuned once the data collection is completed. The strategies for and the performance of the tracker alignment during Run 2 (2016-2018) will be presented, emphasising the ultimate accuracy achieved with the legacy reprocessing. The data-driven techniques used to derive the alignment parameters and the methods used to validate the alignment performance will be reviewed. Finally, the preparations of CMS towards tracker alignment during Run 3 (2022-2025)--in particular, the very first alignment performed after the LHC Long Shutdown 2, with cosmic ray muons and collision data at $\sqrt{s}$ = 900 GeV--will be discussed.