Please don't remove and don't abandon it. There must be an option for simplified deployment without any orchestration, ceph-deploy is that option. I understand that you want to force everyone use containers, but it's definitely unnecessary to fast track it. There are funny things with it, too. For example, official Ceph containers have old smartctl thus drive health collection doesn't work. Another example is that it's currently impossible to change grafana url with cephadm because the orchestration module always resets it to a hardcoded value. And I'm sure there are plenty more issues... >> That said, often there’s a genuine need for progress, eg. ceph-disk. >> I never had that many problems with it myself, though I recognize >> that it was … complex and messy so the move to ceph-volume made >> sense. > > And there, there was a clear transition by removing ceph-disk from the > code base and documenting the rationale for doing so. > >> So for ceph-deploy, I wouldn’t object to not updating it to work with >> future releases, but please don’t remove it from git or package >> repositories, as some people will need to use it indefinitely. > > The challenge I see here: as long as ceph-deploy is included in the code > base and packages (and docs), people assume it's still maintained. As > this does not seem to be the case, it should officially be deprecated > and removed eventually. Otherwise, we're just setting false expectations > and cause frustration. > > Lenz _______________________________________________ Dev mailing list -- dev@xxxxxxx To unsubscribe send an email to dev-leave@xxxxxxx
