From mboxrd@z Thu Jan 1 00:00:00 1970 From: hch@lst.de (Christoph Hellwig) Date: Fri, 25 Nov 2016 08:43:46 +0100 Subject: [PATCH 1/3] nvmetcli: Adding manpage/html generation In-Reply-To: <1479924073-6952-2-git-send-email-james_p_freyensee@linux.intel.com> References: <1479924073-6952-1-git-send-email-james_p_freyensee@linux.intel.com> <1479924073-6952-2-git-send-email-james_p_freyensee@linux.intel.com> Message-ID: <20161125074346.GA753@lst.de> Hi Jay, thanks a lot for writing this man page! I have a few comments below, but otherwise this looks really great. > +NAME > +---- > +nvmetcli - Configure NVMe Target for NVMe-over-Fabrics solution. I'd just say "Configure the NVMe-over-Fabrics Target". > +[verse] > +nvmetcli > +nvmetcli [cmd] [file] Should we just list the three actual command here? > +DESCRIPTION > +----------- > +*nvmetcli* is a python-based program used for viewing, editing, saving, I'd drop python here - that's an implementation detail not interesting to the actual user reading the man page. > It enables an administrator to assign local > +storage resources backed by either local NVMe devices, OS files, or > +system volumes and export them to remote systems via network fabrics > +based on the NVMe-over-Fabrics specification from http://www.nvmexpress.org. I'd something like "It allows to export a local block device (including, but not limited to a NVMe devices).." We can't really export files directly but need to turn them into a block device first using the loop device, and NVMe devices aren't really treated special, although they are of course worse mentioning. > +nvmetcli manages only one NVMe Target as there is only one Linux kernel > +NVMe Target per system. Can we just remove this sentence? As per the spec there is no such thing as a NVMe Target (unfortunately, bit out my teeth on that), but just a collection of subsystems and ports - with nvmetcli we can create sets of subsystems and ports that are entirelt disjoint. > +|================== > +| cd | Same as Linux, allows to move around the tree. > +| ls | Similar as Linux, lists contents of current tree node. Can we drop the "Same as" and "Similar as"? It's really traditional Unix shells that we are similar too, but that wording seems a bit clumsy. > +ADDITIONAL INFORMATION > +---------------------- > +nvmetcli has the ability to start and stop the NVMe Target configuration > +on boot and shutdown through the *systemctl* Linux utility via a .service file. It might be good to mention the actual service, e.g. nvmetcli ships with a nvmet systemd service file, which automatically restores the saved config from /etc/nvme/nvmet.json on system startup, and clears the configuration on shutdown. You can uses systemctl (and we probably want a man page reference to systemctl here, no idea how to do that in asciidoc) to enable and disable this functionality. > + > +AUTHORS > +------- > +This was written by mailto:james.p.freyensee at intel.com[Jay Freyensee] > +for mailto:hch at infradead.org[Christoph Hellwig]. If at all I'd word this as This man page was written by mailto:james.p.freyensee at intel.com[Jay Freyensee]. nvmetcli was originall written by mailto:hch at infradead.org[Christoph Hellwig]. But I'd be personallly fine with just dropping all this from the man page and instead have an AUTHORS file. Any suggestions from someone else? > + > +REPORTING BUGS & DEVELOPMENT > +----------------------------- > +Please send patches to linux-nvme at lists.infradead.org for review and acceptance. Might be worth to also mention bug reports here. > +doc: ${NAME} > + ${MAKE} -C ${DOCDIR} > + > +cleandoc: > + @rm -f ${DOCDIR}/${PKGNAME}.1 > + @rm -f ${DOCDIR}/${PKGNAME}.html > + @rm -f ${DOCDIR}/${PKGNAME}.xml Should this be ${MAKE} -C ${DOCDIR} clean? Also I suspect this should be a .8 page, not .1 given that we install into sbin.