On Thu, Apr 13, 2023 at 06:48:14PM -0400, Charlie Li wrote: > Shawn Webb wrote: > > Does the ZFS project have some sort of automated testing to catch > > data-gobbling, pool killing bugs? It seems like this would have been > > caught with some CI/CD stress testing automation scripts. > > > I can't speak about how the OpenZFS project does things, but this particular > corruption does not have any deterministic characteristics both pre- and > post-condition, so would be hard to automate testing.
My approach would be to have a policy by which any new feature scheduled to land in the main branch must also not show any regressions when running `poudriere bulk -ac`. Such a policy could be enforced via server-side git commit hook. One problem, though, is that implementing that policy isn't just a matter of code, but also infrastructure, so there's a tangible monetary cost. I should mention that I appreciate the selfless hard work of those involved in the FreeBSD and OpenZFS projects. I hope for continued incremental improvements. Thanks, -- Shawn Webb Cofounder / Security Engineer HardenedBSD https://git.hardenedbsd.org/hardenedbsd/pubkeys/-/raw/master/Shawn_Webb/03A4CBEBB82EA5A67D9F3853FF2E67A277F8E1FA.pub.asc
signature.asc
Description: PGP signature
