On 08/10/2013 12:37 PM, ronnie sahlberg wrote: > On Fri, Aug 9, 2013 at 11:09 PM, Dan Mick <dan.mick@xxxxxxxxxxx> wrote: >> I can confirm that, with all instances of 'rdb' changed to 'rbd', and the >> obvious change to bs_rbd.c, this works for bs_rbd as well. If I'd been more >> intelligent I'd have committed your patch locally before hacking on it. >> It's late. :) > > Thanks. I have updated the patch. >> >> +1 on the idea and overall structure; it would be interesting in particular >> to hear others' opinions about the choice of /usr/lib/tgtd/backing-store, >> although it seems fine to me. > > I have no strong feelings about where to store the modules so if > anyone has a better suggestion for the directory, please let me know > and I'll change it. > In these cases the installing scripts can just create a symlink in /usr/lib/tgtd/backing-store to a module that resides with in it's own package, say rdb. This way you compile for the real place and need not have tgtd installed (/usr/lib/tgtd/backing-store exist) You will however need a 3rd package, say tgtd-rdb, dependent on both that ties everything in at installation time. My $0.017 Boaz > >> >> Ceph users would benefit greatly from the ability for the official released >> packages (assuming the package maintainers will do it) to contain the rbd >> backend without requiring compilation. > > Yes. It would also be really neat if at a later stage one could > unload/reload a bs module at runtime. This I think would greatly > improve the experience for people that are developing and testing > backends. > >> >> <> -- To unsubscribe from this list: send the line "unsubscribe stgt" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
