CONCEPT Cited by 1 source
CI reliability¶
CI reliability is the fraction of continuous-integration pipeline runs that succeed for reasons attributable to the code under test. A failing run is "unreliable" (as opposed to legitimately failing) when its cause is one of: flaky test, infrastructure flake, or a semantic merge conflict that didn't show up at branch-CI time — i.e. a failure mode the individual author could not predict or address at submit time.
Distinct from PR-level pass rate (does your branch pipeline pass
before you merge) — CI reliability is typically measured at the
target-branch level (main) because that's where the combined
effects of concurrent-merge interactions show up. Atlassian's Jira
repo disclosed CI failures from semantic merge issues at 7–10%
of all CI failures before merge queues, driven to near zero
afterwards.
(Source: sources/2026-04-29-atlassian-inside-atlassians-merge-queues)
Seen in¶
- sources/2026-04-29-atlassian-inside-atlassians-merge-queues — Jira-repo disclosure; semantic-merge-conflict-caused CI failures as the load-bearing unreliability driver in trunk-based development at scale.
Related¶
- concepts/build-reliability
- concepts/semantic-merge-conflict
- concepts/merge-queue
- concepts/developer-velocity
- concepts/trunk-based-development
- patterns/validate-against-future-state-of-main
- patterns/eject-failing-pr-keep-queue-running
- systems/bitbucket-merge-queues
- systems/bitbucket-pipelines
- companies/atlassian