On Mon, Dec 02, 2013 at 11:46:32AM -0500, Chuck Lever wrote: > Hi- > > Questions about NFSD should go to Bruce Fields (copied), the Linux NFS server maintainer. > > > On Nov 29, 2013, at 6:21 AM, Charles 'Boyo <charlesboyo@xxxxxxxxx> wrote: > > > Hello all. > > > > Please guide me into to find out the impact a busy nfsd is having on > > the Linux page cache. Is it possible to determine which files have > > portions in the page cache? > > That is not possible even for local files, AFAIK. (/me prepares to eat his words). Right. Sounds like an interesting research project.... > > Secondly, files opened by nfs clients do not show up in lsof (or ftop) > > on the nfs server? Is there any way to list these open files and more > > importantly, observe the i/o to them? > > > NFSv3 does not have an OPEN operation. It would be difficult to tell, accurately, what an NFSv3 client was up to by looking for open files on an NFS server. > > NFSv4 does perform file OPENs on the wire, but our I/O metrics are per-NFS operation, not per-file. So you can see how many READs and WRITEs are done in aggregate, but not by which client, and not for a particular file. > > NFSv4 also has delegation, which allows clients to operate on files without contacting the NFS server. I/O activity that is cached by a client would not be visible to the storage server in any way. Yes. We could (and really should) have some statistics/debugging interfaces to expose information like which clients currently hold NFSv4 opens, but we don't currently. --b. -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
