Return-Path: Received: from elasmtp-junco.atl.sa.earthlink.net ([209.86.89.63]:41298 "EHLO elasmtp-junco.atl.sa.earthlink.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753148AbcK3Q44 (ORCPT ); Wed, 30 Nov 2016 11:56:56 -0500 From: "Frank Filz" To: "'J. Bruce Fields'" Cc: "'Tom Haynes'" , "'Linux NFS Mailing list'" References: <1480228001-64821-1-git-send-email-loghyr@primarydata.com> <20161128163321.GB30805@parsley.fieldses.org> <012c01d24a9a$92ff7580$b8fe6080$@mindspring.com> <20161130142412.GA18463@parsley.fieldses.org> In-Reply-To: <20161130142412.GA18463@parsley.fieldses.org> Subject: RE: [PATCH pynfs 00/12] Flex File support Date: Wed, 30 Nov 2016 08:56:38 -0800 Message-ID: <018401d24b2a$b8037670$280a6350$@mindspring.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Sender: linux-nfs-owner@vger.kernel.org List-ID: > > > -----Original Message----- > > > From: linux-nfs-owner@vger.kernel.org [mailto:linux-nfs- > > > owner@vger.kernel.org] On Behalf Of J. Bruce Fields > > > Sent: Monday, November 28, 2016 8:33 AM > > > To: Tom Haynes > > > Cc: Linux NFS Mailing list > > > Subject: Re: [PATCH pynfs 00/12] Flex File support > > > > > > On Sat, Nov 26, 2016 at 10:26:29PM -0800, Tom Haynes wrote: > > > > I wanted to add client support for the flex file layout. > > > > > > > > Note, I did not add pynfs as a flag because I didn't want to mess > > > > up with any existing uses of it. > > > > > > > > The other major change here is in closing all opened files and > > > > destroying all clientids. With all the tests which run against my > > > > server, there are no longer any open files. There are however 11 > > > > clientids remaining. > > > > > > > > I will track those down. > > > > > > > > fwiw - these are in my staging branch at > > > > git://linux-nfs.org/~loghyr/pynfs.git > > > > > > Thanks! I'll try these and take a look. > > > > > > How did you notice all the leftover state? I run with a pretty > > > short > > lease time > > > (to speed testing), so I guess the leftover state must expire too > > > quickly > > to > > > cause me problems. > > > > > > Anyway it's good to have that cleaned up. > > > > Not closing files and releasing clientids at least in the 4.0 tests > > has actually shaken out bugs in Ganesha... > > > > But maybe that kind of thing should be an explicit test, create lots > > of clientids/sessions, and then wait for them all to expire and see if > > the server can handle all the activity. > > Yes, probably so. > > I seem to recall this not being the first time this has caused problems for > people with test scripts that try to unmount immediately after running pynfs. Yea... One option would be for pynfs to delay for the lease period (or maybe twice the lease period) after the last test. One nice thing (and now that you mention it, I'm pretty sure we've had this conversation before) about the status quo is that the lease expiry for earlier tests happens in the middle of other tests. Maybe not so nice an deterministic for the test suite, but great at exposing issues... Frank --- This email has been checked for viruses by Avast antivirus software. https://www.avast.com/antivirus