From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mail-qt0-f180.google.com ([209.85.216.180]:53841 "EHLO mail-qt0-f180.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932484AbdJJRV3 (ORCPT ); Tue, 10 Oct 2017 13:21:29 -0400 Received: by mail-qt0-f180.google.com with SMTP id n61so25498390qte.10 for ; Tue, 10 Oct 2017 10:21:29 -0700 (PDT) From: Josef Bacik Subject: [PATCH 0/2] Add fio performance tracking to fstests Date: Tue, 10 Oct 2017 13:21:23 -0400 Message-Id: <1507656085-17101-1-git-send-email-josef@toxicpanda.com> Sender: fstests-owner@vger.kernel.org To: kernel-team@fb.com, fstests@vger.kernel.org, david@fromorbit.com, tytso@mit.edu List-ID: Everybody agrees that a new project for performance testing is not super useful, and that it would be better to integrate it into something we all already run. These two patches take what was fsperf and puts it into fstests proper. This provides better flexibility for test environments, and at least gets us started towards adding more of these sort of tests into our day to day activities. The only thing extra to run these tests is the PERF_CONFIGNAME option that must be specified in order for the perf tests to be run. We need this (or something like it) to make sure that we compare runs that have similar setups so that we don't end up failing because we did one run with NVME but the next run with SATA. I've only added one of the fio tests from fsperf because I want to make sure we're all happy with the generic support that I've added. Once these are merged I'll send the next tests along and rework how we do comparisons so we're comparing averages of previous runs instead of just whatever the last run was. Thanks, Josef