Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Sat, 13 Oct 2001 16:58:14 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Sat, 13 Oct 2001 16:58:04 -0400 Received: from pigpen.lucentctc.com ([199.93.237.4]:28686 "EHLO pigpen.lucentctc.com") by vger.kernel.org with ESMTP id ; Sat, 13 Oct 2001 16:57:56 -0400 Message-ID: From: "Kingsbury, Michael" To: "'David Schwartz'" , "'linux-kernel@vger.kernel.org'" Subject: RE: High Rate of Sockets -> No buffer space availible errors Date: Sat, 13 Oct 2001 16:58:17 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Yes, these are TCP sockets. I do see some errors with regard to too many open files, but not before the buffer error. -mike -----Original Message----- From: David Schwartz [mailto:davids@webmaster.com] Sent: Saturday, October 13, 2001 10:13 AM To: mkingsbury@avayactc.com; 'linux-kernel@vger.kernel.org' Subject: Re: High Rate of Sockets -> No buffer space availible errors On Sat, 13 Oct 2001 11:08:48 -0400, Kingsbury, Michael wrote: >I have a network testing application that is opening & closing sockets with >other machines at a high rate (multi-threaded, 1000 opens & closes a second >with ~20 machines.) There's a seperate thread per machine its connecting >to, and each thread opens a socket, transmits 8k, and closes. Are these TCP sockets? It can take around 2 minutes to close a TCP connection. So 1,000 opens/closes a second could potentially mean 120,000 connections sitting around. Are you sure you aren't running out of local ports or something else? >The problem lies with an error of 'No buffer space availible' within the >first couple of seconds. I've tried the SO_SNDBUF& SO_RVCBUF, but that >doesn't make sense in my head anyways. Anyone seen problems like this under >similar conditions & maybe any remedys? Which system call returns the error? socket? bind? send? receive? And why are you using so many threads? Are you under the misperception that you need lots of thread to do lots of work? Perhaps your architecture is a major part of the problem. DS - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/