From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from localhost (localhost [127.0.0.1]) by turing.freelists.org (Avenir Technologies Mail Multiplex) with ESMTP id 24D4A21833 for ; Sun, 30 Jun 2019 15:57:16 -0400 (EDT) Received: from turing.freelists.org ([127.0.0.1]) by localhost (turing.freelists.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Aa9Lqs6ouZL1 for ; Sun, 30 Jun 2019 15:57:16 -0400 (EDT) Received: from smtp55.i.mail.ru (smtp55.i.mail.ru [217.69.128.35]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by turing.freelists.org (Avenir Technologies Mail Multiplex) with ESMTPS id 7873920145 for ; Sun, 30 Jun 2019 15:57:15 -0400 (EDT) Received: by smtp55.i.mail.ru with esmtpa (envelope-from ) id 1hhfwn-0001H8-1R for tarantool-patches@freelists.org; Sun, 30 Jun 2019 22:57:13 +0300 From: Vladislav Shpilevoy Subject: [tarantool-patches] [PATCH v2 0/2] SWIM log TTD, and TTL Date: Sun, 30 Jun 2019 21:58:09 +0200 Message-Id: MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Sender: tarantool-patches-bounce@freelists.org Errors-to: tarantool-patches-bounce@freelists.org Reply-To: tarantool-patches@freelists.org List-Help: List-Unsubscribe: List-software: Ecartis version 1.0.0 List-Id: tarantool-patches List-Subscribe: List-Owner: List-post: List-Archive: To: tarantool-patches@freelists.org It is a first patchset to make SWIM more usable in big clusters. Most of the problems arise from incorrect interpretation of the original SWIM paper, and extensions adaptation. The first commit makes SWIM conform with the original paper and disseminate each event for log(cluster_size) times instead of cluster_size. After it there appears a problem in too early deletion of dead and left members, when GC is on. Strictly speaking, it existed even before the patch, but now it is more clear. The second commit tries to solve it by keeping dead and left members in a limbo queue until the whole cluster is aware about their death. It solves the problem partially, but doesn't work when heartbeat_rate is too small. See the commit message for an explanation. However, there is another solution how to do not struggle with TTL and limbo. I've sent it in another thread. V1: https://www.freelists.org/post/tarantool-patches/PATCH-02-SWIM-big-cluster-improvements-part-1 Changes in V2: - Drop commit about preference of new events over old ones; - Split logarithmic TTD and TTL + limbo queue in separate commits; Branch: http://github.com/tarantool/tarantool/tree/gerold103/gh-4253-dissemination-speed Issue: https://github.com/tarantool/tarantool/issues/4253 Vladislav Shpilevoy (2): swim: disseminate event for log(cluster_size) steps [RAW] swim: send 'left' and 'dead' to limbo before deletion src/lib/swim/swim.c | 117 ++++++++++++++++++++++++++++++++++++++++-- test/unit/swim.c | 74 ++++++++++++++++++++++++-- test/unit/swim.result | 8 ++- 3 files changed, 190 insertions(+), 9 deletions(-) -- 2.20.1 (Apple Git-117)