Help me optimize my ppp server.

Help me optimize my ppp server.

Post by Kyler Jone » Wed, 10 Feb 1999 04:00:00

Hi!  I'm trying to rework my serial/ppp setup because my ppp, dial
in server just isn't working the way I'd like.  Mostly, there
are problems with slow access (seems to slow down the longer the
user is online) and getting disconnected a lot.

Also, compression never works (as far as I know - maybe there is
compression being used besides what the kernel reports as disabled?).

Now, I'm not saying that my server is unusable, just that I know for
sure that  it isn't quite right.  I have tested a friends machine
dialing into another ISP and it's definitely faster and more

The main usage of my server is as an ip masquerading host to provide
internet for some friends, etc..

Here is my setup:

 Redhat 5.2
 kernel 2.2.1
 seterial 2.15 (latest)
 ppp 2.3.5 (latest)
 glibc (of course)
 getty_ps 2.0.7j (latest)
 USR Sportster 33.6 Internal (16550A UART)

I only just recently move to Redhat, but I had all the same issues
with Slackware.

I've upgraded all the packages in the /Documentation/Changes file
that came with the 2.2.1 kernel.

Here's my config files (everything I can think of):

1.  The pertinent part of /etc/inittab:

 s1:345:respawn:/sbin/uugetty ttyS1 38400 vt100

2.  The pertinent part of /etc/ppp/options:

 asyncmap 0

3. My /etc/default/uugetty.ttyS1 file:

 INIT="" ATS0=1\r OK
 #CONNECT="" ATs0=1\r CONNECT \s\A

4. Here is the script executed by dialin users to start ppp:

 # ppplogin - script to fire up pppd on login
 mesg n
 stty -echo
 exec /usr/sbin/pppd -detach silent modem crtscts

5. Here is some stuff I have in my /etc/rc.d/rc.local:

 /sbin/modprobe ppp
 /sbin/modprobe bsd_comp
 /sbin/modprobe ppp_deflate
 /bin/setserial /dev/ttyS1 spd_hi

6. I'm using the stock /etc/gettydefs that came with redhat (seems to be

   the same as Slackware, has entries for all speeds, etc.).

7. (Finally) Here is the syslog info from /var/log/messages:

 pppd[1122]: pppd 2.3.5 started by mo, uid 1007
 Feb  8 20:21:40 Mirkwood pppd[1122]: Using interface ppp0
 Feb  8 20:21:40 Mirkwood pppd[1122]: Connect: ppp0 <--> /dev/ttyS1
 Feb  8 20:21:45 Mirkwood pppd[1122]: CCP terminated by peer
 Feb  8 20:21:45 Mirkwood pppd[1122]: Compression disabled by peer.
 Feb  8 20:21:48 Mirkwood pppd[1122]: Cannot determine ethernet address
 for proxy ARP
 Feb  8 20:21:48 Mirkwood pppd[1122]: local  IP address
 Feb  8 20:21:48 Mirkwood pppd[1122]: remote IP address
 Feb  8 20:29:19 Mirkwood pppd[1122]: Hangup (SIGHUP)
 Feb  8 20:29:19 Mirkwood pppd[1122]: Modem hangup
 Feb  8 20:29:19 Mirkwood pppd[1122]: Connection terminated.
 Feb  8 20:29:19 Mirkwood pppd[1122]: Exit.
 Feb  8 20:29:22 Mirkwood kernel: Use of setserial/setrocket to set
 SPD_* flags is deprecated
 Feb  8 20:29:30 Mirkwood last message repeated 2 times

First of all, I know about the rpoxyarp crap, why it isn't working and
I really don't need it.  I'm just too lazy to change it.

Second, I've played around quite a bit with slower speeds in inittab and

with the speed flags in setserial, nothing seems to change.

The compression modules are being loaded in my rc.local because I don't
believe the kernel was finding them.  T kept getting "can't find module
"compression" or something.  I can't remember now.  It only started when

I upgraded tp kernel 2.2.1.

Also, the last 2 lines in messages are new with 2.2.1.  Is this a
Is the spd_hi flag being discarded?  The Documentation/Changes file said

to upgrade to the latetst setserial , and I did.

One last thing:  Is there a way to determine (from my end) what connect
speed was actually established?  On the other end, it seems to be
random, and funny speeds too (like 26,800, stuff like that) although
sometimes it's normal.  It varies from time to time and from computer
to computer.

Anyways, thanks in advance for any help you can give.  It's definitely

In your debt,

Kyler Jones