[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CANn89i+c_N=-FPfRa5e=qZXbQFHSW1fhdZoqgBGq6VdAzdAJZA@mail.gmail.com>
Date: Tue, 9 May 2023 15:36:15 +0200
From: Eric Dumazet <edumazet@...gle.com>
To: Cathy Zhang <cathy.zhang@...el.com>
Cc: davem@...emloft.net, kuba@...nel.org, pabeni@...hat.com,
jesse.brandeburg@...el.com, suresh.srinivas@...el.com, tim.c.chen@...el.com,
lizhen.you@...el.com, eric.dumazet@...il.com, netdev@...r.kernel.org
Subject: Re: [PATCH net-next 2/2] net: Add sysctl_reclaim_threshold
On Mon, May 8, 2023 at 4:08 AM Cathy Zhang <cathy.zhang@...el.com> wrote:
>
> Add a new ABI /proc/sys/net/core/reclaim_threshold which allows to
> change the size of reserved memory from reclaiming in sk_mem_uncharge.
> It allows to keep sk->sk_forward_alloc as small as possible when
> system is under memory pressure, it also allows to change it larger to
> avoid memcg charge overhead and improve performance when system is not
> under memory pressure. The original reclaim threshold for reserved
> memory per-socket is 2MB, it's selected as the max value, while the
> default value is 64KB which is closer to the maximum size of sk_buff.
>
> Issue the following command as root to change the default value:
>
> echo 16384 > /proc/sys/net/core/reclaim_threshold
So, when the system is under pressure, who is going to reduce this sysctl ?
I am sorry, but if you really need good performance numbers for benchmarks,
maybe add a static key to completely disable memory accounting for UDP
and TCP...
Powered by blists - more mailing lists