[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20070704120652.262380eb.dada1@cosmosbay.com>
Date: Wed, 4 Jul 2007 12:06:52 +0200
From: Eric Dumazet <dada1@...mosbay.com>
To: "Robert Iakobashvili" <coroberti@...il.com>
Cc: "Evgeniy Polyakov" <johnpol@....mipt.ru>,
NetDev <netdev@...r.kernel.org>
Subject: Re: Beyond 64K TCP connections limit per IP-address
On Wed, 4 Jul 2007 11:40:48 +0200
"Robert Iakobashvili" <coroberti@...il.com> wrote:
> On 7/4/07, Evgeniy Polyakov <johnpol@....mipt.ru> wrote:
> > On Wed, Jul 04, 2007 at 09:50:31AM +0200, Robert Iakobashvili (coroberti@...il.com) wrote:
> > > If I am correct, a TCP server can make up to
> > > 64K accepts for a port at a single IP-address.
> >
> > No, it is essentially unlimited - linux uses local/remote addr/port
> > tuples for hash chains, so there is no per-addr limits.
> > If there is some kind of binds, then yes, only 64k ports per address.
>
> Thanks, it clarified me the issue.
> Probably, I am experiencing some local problem with
> the web-server I am using for tests.
If your setup is :
Server A with one IP address listening to port 80
'Client B' with one IP address, trying to open many sockets to A (port 80)
Then yes you have a 64k limit for this particular client B. Just add 15 more clients (or 16 IP addresses on B) if you really want to stress A ;)
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists