From mboxrd@z Thu Jan 1 00:00:00 1970 From: Loic Dachary Subject: Merging rados related backports Date: Fri, 5 Aug 2016 15:08:36 +0200 Message-ID: <57A48FD4.40401@dachary.org> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 8bit Return-path: Received: from relay3-d.mail.gandi.net ([217.70.183.195]:35984 "EHLO relay3-d.mail.gandi.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751799AbcHENIm (ORCPT ); Fri, 5 Aug 2016 09:08:42 -0400 Sender: ceph-devel-owner@vger.kernel.org List-ID: To: Samuel Just Cc: Ceph Development Hi Sam, When running the rados suite on a given backport targetting jewel (or hammer for that matter, but it's less of an issue) I figured the smallest subset would be enough to guarantee there is no trivial regression, this is why we're routinely using X/2000. I understand this is not enough coverage to release jewel and we have the nightlies running X/14 for that. When jewel is eventually released, all of the rados suite should have been run. If you think there are too many regressions merged in the jewel branch because X/2000 did not detect them, we will increase the size of the subset. This will slow down backporting when the lab is misbehaving, but that's less time consuming than chassing a regression long after the guilty commit has been merged. Cheers [1] http://tracker.ceph.com/projects/ceph-releases/wiki/Sepia/97 -- Loïc Dachary, Artisan Logiciel Libre