From owner-freebsd-current Mon Aug 21 10:29:04 1995 Return-Path: current-owner Received: (from majordom@localhost) by freefall.FreeBSD.org (8.6.11/8.6.6) id KAA01659 for current-outgoing; Mon, 21 Aug 1995 10:29:04 -0700 Received: from vax.cs.pitt.edu (vax.cs.pitt.edu [136.142.79.5]) by freefall.FreeBSD.org (8.6.11/8.6.6) with ESMTP id KAA01627 for ; Mon, 21 Aug 1995 10:28:52 -0700 Received: from w2xo.pgh.pa.us by vax.cs.pitt.edu (8.6.10/1.14) for ; id NAA23787; Mon, 21 Aug 1995 13:29:09 -0400 Received: (from durham@localhost) by w2xo.pgh.pa.us (8.6.11/8.6.9) id NAA00845; Mon, 21 Aug 1995 13:24:19 -0400 Date: Mon, 21 Aug 1995 13:24:15 -0400 (EDT) From: Jim Durham To: current@freebsd.org Subject: buffers full on 2.0.5R Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: current-owner@freebsd.org Precedence: bulk I have: 1. An internal 28.8 modem running iijPPP to my unreliable internet provider. 2. 2 polled serial ports used by an application with select(). The application is a ham radio AX25 server that was originally written for DOS and is a giant polling loop. Yeah..I need to fix it..but no time right now.. I have noticed some strange behavior with this set-up. I *think* this is what happens... The modem PPP stuff goes nuts due to some problem at the server end and won't pass IP packets. I think this is *his* problem. iijPPP still indicates that the PPP connection is working. If you telnet to it's daemon, the PPP is still upper case. However, when this happens, I start getting thousands of overflows on the serial port reported and, if you try to ping someone out the SLIP connection, you get "buffers full" errors. The application polling the serial ports goes comatose also. It won't respond to keyboard input and it's children become zombies. It sure looks like something is scribbling on some kernel structures when the IP buffers fill up. Note that as long as the system is running normally, this doesn't happen. It's only when the network "jams up". Another thing, that happened. I was playing with the routing tables to see if that was causing a problem, as someone suggested.I managed to get a default route entry in the routing table with localhost as a gateway when there was already an entry for my slip provider's system as a default. I had two default entries. I couldn't get it to go away except by rebooting. Netstat kept telling me it was there, but route delete said it wasn't. Route flush didn't do anything at all, everything remained. It sounds like the routing tables were also lunched at the time (pointer pointing nowhere?) -Jim (If it can be broken, I'll break it) Durham