From owner-freebsd-questions Thu Mar 14 12:04:11 1996 Return-Path: owner-questions Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id MAA28160 for questions-outgoing; Thu, 14 Mar 1996 12:04:11 -0800 (PST) Received: from bsd.tseinc.com (bsd.tseinc.com [199.217.191.65]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id MAA28154 for ; Thu, 14 Mar 1996 12:04:08 -0800 (PST) Received: from ws2.tseinc.com (ws2.tseinc.com [199.217.203.22]) by bsd.tseinc.com (8.6.11/8.6.9) with SMTP id JAA02128 for ; Thu, 14 Mar 1996 09:46:56 GMT Date: Thu, 14 Mar 1996 09:46:56 GMT Message-Id: <199603140946.JAA02128@bsd.tseinc.com> X-Sender: jlwest@bsd.tseinc.com X-Mailer: Windows Eudora Light Version 1.5.2 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: freebsd-questions@freebsd.org From: "Jay L. West" Subject: Help!! Severe PPP problems! Sender: owner-questions@freebsd.org X-Loop: FreeBSD.org Precedence: bulk We are a small start-up ISP. We started with a 28.8k connection to our upstream provider and about 7 users. We ran that way for a few months and all went very well. Then we switched our upstream connection to 64k ISDN and all h*** broke lose. Here's what happens: After a fresh boot everything runs great for from 30 minutes to 2 hours roughly. Then I notice via systat -pigs and ps -aux that the ppp daemon running the connection to our provider starts using 124% of the cpu. The process shows as 'ppp -auto providername'. It is using 500k virtual memory and 440k real memory. It is runnable (flags are Rs). When I try a ping to our provider I get either no response (the ping just hangs) or sometimes I get 'no buffers available'. A tcpdump -i tun0 -q shows no traffic in or out. A netstat -i shows no errors in or out. A netstat -s shows no checksum problems. Sometimes systat will say segmentation fault. The ppp link via ISDN to our provider is running at 155200k baud on a 16550 UART. We're running 2.0.5 on a 486dx4/100 with 16mb of RAM. When I first noticed a ping generated the message 'no buffer space available' I set NMBCLUSTERS to 1500 and then later set it to 2048. I have spent two days trying to trouble shoot the problem and customers are getting upset (there's a euphamism ): ). Any help would be very very very much appreciated. We may try to upgrade with a fresh install to 2.1 if that would help. Does anyone have any ideas? Please? Jay West The Software Exchange (1-800-669-8203)