It seems as though there might be use for a flag to allow using more of the disk somehow, or to tar up existing files rather than copying them. I assume in most cases the user can make that decision themselves by manually removing older files or other things on the FS.
-Reid On Tue, Sep 24, 2013 at 10:57 AM, Ben Pfaff <b...@nicira.com> wrote: > On Tue, Sep 24, 2013 at 10:46:59AM -0700, Shih-Hao Li wrote: > > > It might be worth documenting how to avoid the cap if the user really > > > wants to (it looks like outputting to a fd instead of a file avoids > > > it?). > > > > How about adding a new parameter, such as the percentage of free disk > > space allowed to use? If it is 0, then no check. By default, always > > check and use 90 regardless output_fd or output_file cases. > > OK. > > > > > +def get_free_disk_space(path): > > > > + path = os.path.abspath(path) > > > > + while not os.path.ismount(path): > > > > + path = os.path.dirname(path) > > > > + s = os.statvfs(path) > > > > + return s.f_frsize * s.f_bfree > > > > > What is the reason for chaining up to a mount point? All the > > > documentation I see says that statvfs returns information about the > file > > > system containing the provided path and doesn't mention special > > > importance of mount points > > I actually borrowed the code from another NVP script. > > I think ismount may not do much here, but it does help tracking down to > a valid path > > because we haven't created the tar file at this time. > > OK, the 'dirname' makes sense then. I still don't think the ismount > chain is necessary. > _______________________________________________ > dev mailing list > dev@openvswitch.org > http://openvswitch.org/mailman/listinfo/dev >
_______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
