[DL.CI.7] Validate the reproducibility of builds
Category: OPTIONAL
Every build for a specific version of source code should ideally be able to generate the same outputs from the same inputs. The implementation of reproducible builds primarily involves the creation of an immutable and consistently created build environment and controlling the inputs for each and every build.
Between each build, the environment should be destroyed and recreated so that it is immutable. Use infrastructure as code (IaC) and containerization to help with automating the creation of the environment in a repeatable and consistent way. Have controls in place to detect and prevent configuration drift that may alter the build environment post-creation. All dependencies and software components used to create the environment and perform the build should be version pinned and recorded.
Any manual intervention during the build can introduce variability. Every step in the build process needs to be automated. Factors that can render the build nondeterministic, such as unrestricted network access and the use of random generators or timestamps that modify the build artifact, must be limited.
Verify the reproducibility by establishing processes that regularly check the reproducibility of the builds. This can involve triggering builds from the same source code in different environments and comparing the results. Adopt mechanisms like binary diffing or checksum comparison to validate the reproducibility of the build. Set up alarms that raise alerts when discrepancies occur to provide fast feedback when there are inconsistencies.
Having reproducible builds is optional and not recommended for all organizations or workloads. While striving for reproducibility is encouraged, it may not be achievable in every context. For example, some builds may depend on specific environmental parameters or timing elements that make reproducibility difficult.
Related information:
