Return-Path: linux-nfs-owner@vger.kernel.org Received: from fieldses.org ([174.143.236.118]:38311 "EHLO fieldses.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755394Ab3DWVO0 (ORCPT ); Tue, 23 Apr 2013 17:14:26 -0400 Date: Tue, 23 Apr 2013 17:14:25 -0400 To: "linux.pnfs" Cc: linux-nfs Subject: Re: pNFS layered over cluster FS, is this the status quo? Message-ID: <20130423211425.GK3676@fieldses.org> References: <201304211613254596282@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <201304211613254596282@gmail.com> From: "J. Bruce Fields" Sender: linux-nfs-owner@vger.kernel.org List-ID: On Sun, Apr 21, 2013 at 04:13:27PM +0800, linux.pnfs wrote: > pNFS layered over cluster FS(such as GFS2), DLM implement layoutget(a pnfs export operations) but it not support IOMODE_RW, that is write data must via meta data server, -->> is this the status quo? > > the MDS' layoutget will support IOMODE_RW and client will RW data via data server in the future? > or is there any support for: howto parellel R&W data via data server? This is up to the implementation. The linux gfs2 pnfs patches implement a very simple prototype. --b.