From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nivedita Singhvi Subject: Re: Stupid netstat question Date: Mon, 21 Oct 2002 14:16:33 -0700 Sender: netdev-bounce@oss.sgi.com Message-ID: <3DB46EB1.69B820FE@us.ibm.com> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Cc: netdev@oss.sgi.com Return-path: To: Ian Nelson Errors-to: netdev-bounce@oss.sgi.com List-Id: netdev.vger.kernel.org Ian Nelson wrote: > > What exactly does the TCPSchedulerFailed metric mean? > I'm building a network appliance that is doing a lot of nmap and nessus > type stuff and I'm getting an extreme number of TCPSchedulerFailed and > TCPAbortOnData. I've been tracking what I think is a bug in an Ethernet > driver but I'm starting to think it's a software problem in application > land. Its the number of packets that are prequeued (partially completed tcp processing, waiting for a recvmsg to complete & send ack etc) when the delayed ack timer goes off. i.e I think, the thinking is this shouldnt happen, since the delayed ack timer really shouldnt go off - the receiver should have picked up the skb's and sent the acks. I'm probably off by a mile here..Dont know. It may be related to the AbortOnData counts. Are you closing apps with linger off or something? thanks, Nivedita