Inside Atlassian’s Merge Queues: How we ship faster with fewer incidents

This post details Atlassian's adoption of Bitbucket Merge Queues across over 70 large repositories, including Jira, Rovo, and Trello. It quantifies the impact of Merge Queues by showing a reduction in CI reliability failures due to semantic merge issues from 7-10% to nearly zero, and a decrease in internal incidents from 3-5 per week to rare edge cases. It also highlights an increase in developer satisfaction regarding build reliability from 70% to 82%. The post explains the three core functions of Merge Queues: queuing PRs, validating against expected future state, and merging or ejecting. It also provides specific configuration details from the Jira team, including setting build concurrency, merge strategy, and using parent-child pipelines for validation. Failure behaviors such as failed builds ejecting PRs and admin controls for reordering/hot-fixes are also described.