Date: Mon, 18 Mar 1996 08:22:00 -0600 From: "Jay L. West" <jlwest@tseinc.com> To: freebsd-questions@freebsd.org Subject: Help!!! Severe PPP problems Message-ID: <199603181422.IAA00291@bsd.tseinc.com>
next in thread | raw e-mail | index | archive | help
Thanks a MILLION for all who have responded thus far. I missed some of the
mail replies though. We upgraded to 2.1 this weekend and the major problem
of the 'ppp -auto provider' grabbing the whole cpu seems to have
disappeared. There is one other problem that remains and I have done a lot
of work to isolate exactly what happens. Hopefully this will give
particulars so others here may help.
The last problem we're having is with iijppp and routes. FYI- we're running
2.1 on a 486dx4/100 with 16mb RAM. The machine has an ethernet card (ed0)
for our local network of workstations (199.217.203). There is an ISDN iijppp
connection to our upstream provider (199.217.191.65 -> 199.217.191.9). When
an iijppp user dials in, the ips are assigned based on the tty device
(199.217.203.40 (us) to 199.217.203.xx (them, range 41-48)). Gateway
behavior (forwarding) is turned on. Here's my sysconfig, ppplogin script,
ppp.conf, and ppp.linkup. The link to my provider is brought up via 'ppp
-auto provider' and 'ping -c1 199.217.191.9' in /etc/netstart.
Here's my /etc/sysconfig lines:
network_interfaces="lo0 ed0 tun0"
ifconfig_lo0="inet localhost"
ifconfig_ed0="inet 199.217.203.1 netmask 0xffffff00"
ifconfig_tun0="inet 199.217.191.65 199.217.191.9 netmask 0xffffff00"
----------------------------------------------------------------------
Here's my iijppp login script (dial up users have this as their login shell):
/usr/sbin/ppp -direct `tty | cut -d/ -f3`
----------------------------------------------------------------------
Here's my /etc/ppp/ppp.conf file (phone number, id and password removed):
default:
disable lqr
deny lqr
set timeout 900
provider:
set device /dev/cuaa1
set speed 115200
set dial "ABORT BUSY ABORT NO\\sCARRIER TIMEOUT 5 \"\" ATE1Q0
OK-AT-OK \\dATDT\\T TIMEOUT 40 CONNECT"
set phone xxxxxxx
set login "TIMEOUT 10 gin:-BREAK-gin: yyyyyy word: zzzzzzz"
set timeout 0
set ifaddr 199.217.191.65 199.217.191.9
ttyd2:
set ifaddr 199.217.203.40 199.217.203.41 255.255.255.0
ttyd3:
set ifaddr 199.217.203.40 199.217.203.42 255.255.255.0
ttyd4:
set ifaddr 199.217.203.40 199.217.203.43 255.255.255.0
ttyd5:
set ifaddr 199.217.203.40 199.217.203.44 255.255.255.0
ttyd6:
set ifaddr 199.217.203.40 199.217.203.45 255.255.255.0
ttyd7:
set ifaddr 199.217.203.40 199.217.203.46 255.255.255.0
ttyd8:
set ifaddr 199.217.203.40 199.217.203.47 255.255.255.0
ttyd9:
set ifaddr 199.217.203.40 199.217.203.48 255.255.255.0
----------------------------------------------------------------------
Here's my /etc/ppp/ppp.linkup file:
provider:
add 0 0 HISADDR
ttyd2:
ttyd3:
ttyd4:
ttyd5:
ttyd6:
ttyd7:
ttyd8:
ttyd9:
----------------------------------------------------------------------
Ok, with that out of the way, here's the problem... Before a dial in user
connects (ie after a fresh boot) here's what my routing table looks like:
bsd# netstat -rn
Routing tables
Internet:
Destination Gateway Flags Refs Use Netif Expire
default 199.217.191.9 UGc 20 210 tun0
127.0.0.1 127.0.0.1 UH 2 142 lo0
199.217.191.9 199.217.191.65 UH 21 1 tun0
199.217.191.65 127.0.0.1 UGHS 0 0 lo0
199.217.203 link#1 UC 0 0
199.217.203.22 0:c0:a8:28:63:8 UHLW 2 920 ed0 330
224 199.217.191.65 US 0 0 tun0
Note that 203.22 is just a workstation off the ed0. Now, when a user dials
in here's the new routing table set up:
bsd# netstat -rn
Routing tables
Internet:
Destination Gateway Flags Refs Use Netif Expire
default 199.217.191.9 UGc 20 210 tun0
127.0.0.1 127.0.0.1 UH 2 142 lo0
199.217.191.9 199.217.191.65 UH 21 1 tun0
199.217.191.65 127.0.0.1 UGHS 0 0 lo0
199.217.203 link#1 UC 0 0
199.217.203.22 0:c0:a8:28:63:8 UHLW 2 920 ed0 330
199.217.203.41 199.217.203.40 UH 1 24 tun1
224 199.217.191.65 US 0 0 tun0
This looks great. If the user disconnects normally, the .41 to .40 route
disappears and all is well. However, if the user disconnects abnormally (ie,
his system crashes or he just turns off his modem, or he disconnects while
receiving data) the phone line hangs up but look at the resulting routing table:
bsd# netstat -rn
Routing tables
Internet:
Destination Gateway Flags Refs Use Netif Expire
default 199.217.191.9 UGc 20 210 tun0
127.0.0.1 127.0.0.1 UH 2 142 lo0
199.217.191.9 199.217.191.65 UH 21 1 tun0
199.217.191.65 127.0.0.1 UGHS 0 0 lo0
199.217.203 link#1 UC 0 0
199.217.203.22 0:c0:a8:28:63:8 UHLW 2 920 ed0 330
199.217.203.41 link#1 UHLW 1 24
224 199.217.191.65 US 0 0 tun0
Note the 203.40 has become link#1, the flags have changed from UH to UHLW,
and the tunx device is gone. If a user dials back in on this port, the route
stays as above and he can log in but not get out tun0 to the Inet. If I do a
'route delete 199.217.203.41' the route will disappear but will then
suddenly re-appear about 30 seconds later (with no user dialing in at all).
The only way I have found to correct this is to do a 'route 199.217.203.41
delete' exactly just before they dial in. Then the route gets added
correctly and if they exit correctly it will disappear and stay gone. If
users exit normally the above configuration works great. Incidentally, if I
do a 'ping 199.217.203.99' (there's no such system on my class C) a link#1
route like above appears, but it disappears in a minute or so and stays
gone. I'm not sure if this is related. Users who connect and disconnect
normally work just great. I copied the above routing information 'after the
fact' so the refs and use fields may not be exact but the rest is. In case
it is germane, this system is running name services (primary for the
tseinc.com domain, primary for the 199.217.203 in-addr.arpa net) and no
routing protocols are being run.
I hope this gives enough information for someone to help us out. Any input
is MOST appreciated. THANKS!
Jay L. West
The Software Exchange, Inc. 1-800-669-8203
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199603181422.IAA00291>