From: Mingming Cao Subject: Re: [PATCH v6 0/4] ext4: Coordinate data-only flush requests sent by fsync Date: Tue, 30 Nov 2010 16:14:39 -0800 Message-ID: <1291162479.2419.179.camel@mingming-laptop> References: <20101129220536.12401.16581.stgit@elm3b57.beaverton.ibm.com> <4CF43BC9.8040603@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Cc: "Darrick J. Wong" , Jens Axboe , "Theodore Ts'o" , Neil Brown , Andreas Dilger , Alasdair G Kergon , Jan Kara , Mike Snitzer , linux-kernel , linux-raid@vger.kernel.org, Keith Mannthey , dm-devel@redhat.com, Tejun Heo , linux-ext4@vger.kernel.org, Christoph Hellwig , Josef Bacik To: Ric Wheeler Return-path: In-Reply-To: <4CF43BC9.8040603@redhat.com> Sender: linux-raid-owner@vger.kernel.org List-Id: linux-ext4.vger.kernel.org On Mon, 2010-11-29 at 18:48 -0500, Ric Wheeler wrote: > On 11/29/2010 05:05 PM, Darrick J. Wong wrote: > > On certain types of hardware, issuing a write cache flush takes a considerable > > amount of time. Typically, these are simple storage systems with write cache > > enabled and no battery to save that cache after a power failure. When we > > encounter a system with many I/O threads that write data and then call fsync > > after more transactions accumulate, ext4_sync_file performs a data-only flush, > > the performance of which is suboptimal because each of those threads issues its > > own flush command to the drive instead of trying to coordinate the flush, > > thereby wasting execution time. > > > > Instead of each fsync call initiating its own flush, there's now a flag to > > indicate if (0) no flushes are ongoing, (1) we're delaying a short time to > > collect other fsync threads, or (2) we're actually in-progress on a flush. > > > > So, if someone calls ext4_sync_file and no flushes are in progress, the flag > > shifts from 0->1 and the thread delays for a short time to see if there are any > > other threads that are close behind in ext4_sync_file. After that wait, the > > state transitions to 2 and the flush is issued. Once that's done, the state > > goes back to 0 and a completion is signalled. > > > > Those close-behind threads see the flag is already 1, and go to sleep until the > > completion is signalled. Instead of issuing a flush themselves, they simply > > wait for that first thread to do it for them. If they see that the flag is 2, > > they wait for the current flush to finish, and start over. > > > > However, there are a couple of exceptions to this rule. First, there exist > > high-end storage arrays with battery-backed write caches for which flush > > commands take very little time (< 2ms); on these systems, performing the > > coordination actually lowers performance. Given the earlier patch to the block > > layer to report low-level device flush times, we can detect this situation and > > have all threads issue flushes without coordinating, as we did before. The > > second case is when there's a single thread issuing flushes, in which case it > > can skip the coordination. > > > > This author of this patch is aware that jbd2 has a similar flush coordination > > scheme for journal commits. An earlier version of this patch simply created a > > new empty journal transaction and committed it, but that approach was shown to > > increase the amount of write traffic heading towards the disk, which in turn > > lowered performance considerably, especially in the case where directio was in > > use. Therefore, this patch adds the coordination code directly to ext4. > > Hi Darrick, > > Just curious why we would need to have batching in both places? Doesn't your > patch set make the jbd2 transaction batching redundant? > We hoped JBD2 could take care of the batching too. But ftrace shows there are a fair amount of barriers (440 barriers/second, if I remember right) sending from ext4_sync_file(), but not coming from jbd2 side. :( > I noticed that the patches have a default delay and a mount option to override > that default. The jbd2 code today tries to measure the average time needed in a > transaction and automatically tune itself. Can't we do something similar with > your patch set? (I hate to see yet another mount option added!) > I don't like a new mount option too. Darrick's new mount option is for the threshold to turn on/off batching. Probably could make it a tunables instead of a mount option. Regards, Mingming > Regards, > > Ric > >