From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Fri, 8 Jun 2018 09:39:15 +0300 From: Vladimir Davydov Subject: Re: [PATCH] memtx: don't delay deletion of temporary tuples during snapshot Message-ID: <20180608063915.6r6rzuutlr4cpob5@esperanza> References: <20180608035636.GG6866@chai> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20180608035636.GG6866@chai> To: Konstantin Osipov Cc: tarantool-patches@freelists.org List-ID: On Fri, Jun 08, 2018 at 06:56:36AM +0300, Konstantin Osipov wrote: > * Vladimir Davydov [18/06/04 23:48]: > > Since tuples stored in temporary spaces are never written to disk, we > > can always delete them immediately, even when a snapshot is in progress. > > > > Closes #3432 > > OK to push. Where to? 1.9 or 1.10? The ticket is on 1.10 milestone, but there's a comment that says it should be fixed in 1.9 ... > > https://github.com/tarantool/tarantool/issues/3432 > > https://github.com/tarantool/tarantool/commits/gh-3432-memtx-dont-delay-free-temp-tuples > > @@ -1309,3 +1311,97 @@ s:select() > > s:drop() > > --- > > ... > > +for i = 1, 100 do box.space.test:insert{i} end > > +--- > > +... > > +-- Create a temporary space. > > +count = 500 > > Please is there a chance this test writes less than 50MB of data > to the database? A couple of order of magnitudes would be really > nice. The problem is tarantool won't start if the memory limit is < 40 MB. So if we want to use less memory in this test, say 10 MB, we will have to set memory limit to 40 + 10 MB, but that would raise questions, as 10 + 10 MB should fit in 50 MB. So I decided to insert 50 MB of data and set limit 100 MB - this is clear, because 50 + 50 MB won't fit in 100 MB no matter how much memory the core needs to start. Anyway, the test runs with wal_mode=none so it should be fast.