On 06/24/2016 06:49 PM, Frank Filz wrote: > I have pushed my latest version of patches that includes removing the > attributes from the fsal_obj_handle, and also copy them out on creates and > lookups. > > I am having issues with pynfs SATT15 test case. Something is causing the > change attribute to roll back, such that it doesn't appear to change. > > Ganesha is faking things up to make this test pass, but I wonder, we > currently base change attribute on the greater of mtim or ctim. My suspicion > is that it should only be based on ctim, but even doing that is not > sufficient to make SATT15 pass without the code inside the #if 0 in > mdcache_refresh_attrs that prevents rolling back change. > > I think we need to have a discussion about how to best maintain a change > attribute. > > Thanks >
I don't think ctime is appropriate, since that leaves out writes entirely. It seems from reading man pages that mtime & ctime is the correct thing. In general, we're probably going to need fixups, since we cannot guarantee that all FSALs will have super-fine-grained modification times. I suppose FSALs could update the change time directly, if they have some better analog than modification time... Daniel ------------------------------------------------------------------------------ Attend Shape: An AT&T Tech Expo July 15-16. Meet us at AT&T Park in San Francisco, CA to explore cutting-edge tech and listen to tech luminaries present their vision of the future. This family event has something for everyone, including kids. Get more information and register today. http://sdm.link/attshape _______________________________________________ Nfs-ganesha-devel mailing list Nfs-ganesha-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/nfs-ganesha-devel
