linux-kernel - Re: [LKP] [net] 19f92a030c: apachebench.requests_per

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Date:   Wed, 13 Nov 2019 18:21:02 +0100
From:   Willy Tarreau <w@....eu>
To:     Eric Dumazet <edumazet@...gle.com>
Cc:     Feng Tang <feng.tang@...el.com>,
        kernel test robot <rong.a.chen@...el.com>,
        "David S. Miller" <davem@...emloft.net>, Yue Cao <ycao009@....edu>,
        LKML <linux-kernel@...r.kernel.org>,
        Linus Torvalds <torvalds@...ux-foundation.org>,
        lkp@...ts.01.org
Subject: Re: [LKP] [net] 19f92a030c: apachebench.requests_per_second -37.9%
 regression

On Wed, Nov 13, 2019 at 08:46:00AM -0800, Eric Dumazet wrote:
> We have been using 4096 at Google for about 5 years, and Google is still
> alive.

Same for most haproxy users, changing from the default 128 to several
thousands always fixes their initial performance problems. There really
must be something else in this test, or it depends on a strange corner
case that's triggered by the increase, but that sounds odd.

Regards,
Willy