It looks like tests.r-b.o is intended to systematically vary the kernel
on amd64:

https://tests.reproducible-builds.org/debian/index_variations.html

  kernel version	     on amd64 systematically varied

But I have seen several builds that are definitely affected by kernel
variations come out reproducible, or embed the same kernel version,
e.g. the tests for dtrace:

https://tests.reproducible-builds.org/debian/rb-pkg/unstable/arm64/diffoscope-results/dtrace.html

  44 	codethink03-arm64                       44 	i-capture-the-hostname
  45 	6.12.88+deb13-cloud-arm64	            45 	6.12.88+deb13-cloud-arm64
  46 	#1·SMP·Debian·6.12.88-1·(2026-05-15)    46   #1·SMP·Debian·6.12.88-1·(2026-05-15)

While it is showing a different kernel on reproduce.debian.net for
arm64:

https://reproduce.debian.net/arm64/api/v1/builds/231311/artifacts/550442/diffoscope

  -6.12.88+deb13-arm64
  -#1 SMP Debian 6.12.88-1 (2026-05-15)
  +6.12.90+deb13-cloud-arm64
  +#1 SMP Debian 6.12.90-1 (2026-05-22)

This led the maintainer to assume it was an arm-specific issue in their
package since the kernel version was not different on other
architectures:

https://bugs.debian.org/1140167

I think it would be good to have tests.reproducible-builds.org
systematically varying the kernel again...

We will get intermitant reproduce.debian.net failure results on packages
just showing up periodically... which does not systematically vary the
kernel(I might suggest it would be worth doing so!), but due to security
updates for the kernel happening not in sync with the buildd.debian.org
machines, will result in episodic "regressions" when kernel updates
become available...

At least somewhere there should be systematic testing of this issue. :)

I might come up with a patch/merge request at some point, but for now
filing this just to avoid forgetting it.

live well,
  vagrant