Greetings, Ceph developers!
As of Octopus, users are expected to deploy Ceph containerized, either using
cephadm or k8s/Rook.
In a containerized deployment, only a small subset of the Ceph packages gets
installed in the host system. The daemon packages (ceph-osd, ceph-mon, ceph-mgr,
nfs-ganesha, etc.) invariably are *not* installed in the host system. Instead,
they are installed inside containers.
This raises the question "for daemons running in containers, how is the user
expected to access the respective manpages?" In pre-Octopus Ceph deployments,
the answer was "Run 'man <daemon>' on the node where the daemon package
installed." But from Octopus onward, that no longer works. Is the user expected
to attach to a running container (how?) and run "man <daemon>" inside it?
I haven't tried it, but somehow that doesn't seem "right" to me.
Show replies by date