From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Martin K. Petersen" Subject: Re: [PATCH v2] block: disallow changing max_sectors_kb on a request stacking device Date: Tue, 08 Nov 2016 16:10:02 -0500 Message-ID: References: <20161028194549.14556-1-snitzer@redhat.com> <20161107164029.GA27464@redhat.com> <20161107192646.GA11572@redhat.com> <8d0f16fe-7293-3b1c-1c60-32a50c77fa89@kernel.dk> <20161107212732.GA28478@redhat.com> <20161108033457.GA30325@redhat.com> Mime-Version: 1.0 Content-Type: text/plain Return-path: In-Reply-To: <20161108033457.GA30325@redhat.com> (Mike Snitzer's message of "Mon, 7 Nov 2016 22:34:57 -0500") Sender: linux-block-owner@vger.kernel.org To: Mike Snitzer Cc: "Martin K. Petersen" , Jens Axboe , linux-block@vger.kernel.org, dm-devel@redhat.com List-Id: dm-devel.ids >>>>> "Mike" == Mike Snitzer writes: Mike, >> However, doesn't it make more sense to tweak limits on DM device >> instead of the underlying ones? It seems a bit counter-intuitive to >> me to change max_sectors_kb on a different device than the one where >> the filesystem whose max I/O size you want to change is located. Mike> As you know, the limits are stacked from the bottom-up. Yep, but since max_sectors_kb is a performance tunable and not something enforced by the hardware, maybe we should consider treating it differently? Mike> As the commit header from this thread mentioned, what I've arrived Mike> at is to have multipathd establish a desired max_sectors_kb (if Mike> configured to do so in multipath.conf) on the underlying paths Mike> _before_ the multipath device is created. Yet to check with Ben Mike> Marzinski or others but it seems like a reasonable feature (and Mike> really the cleaniset way to deal with this issue IMHO.. no need Mike> for lots of kernel code changes for something so niche). That's fine. Another option would be to use the max_dev_sectors queue limit to contain the minimum max_sectors from below. It was really only envisioned as a LLD limit but it may be useful in this case. queue_max_sectors_store() already enforces it. -- Martin K. Petersen Oracle Linux Engineering