PPTP Tunnel

I am trying to figure out a problem between 2 Debian 3.0 boxes being used to create a PPTP tunnel between two offices. There are two tunnels set up; one going over the primary 5Mb RF connection and the second over a 128Kb DSL connection. The problem is intermittent but is more frequent over the RF connection. The kernel (2.4.20) has been patched for mppe. I am not sure what causes the problem but the RF connection is not stable. Bandwidth comes and goes bringing down the primary tunnel and the tunnel sometimes does not come back up when the connection returns. I read something in a list that said it may be cause by header compression so I removed that with the 'noccp' option and that seemed to fix the problem for a few days but it's happening again. Here is some of the stuff I see in my syslog:

Apr 21 14:25:22 host1 pptpd[4560]: CTRL: PTY read or GRE write failed (pty,gre)=(-1,-1)
Apr 21 14:25:22 host1 pptpd[4560]: CTRL: Client 172.16.0.1 control co
nnection finished
Apr 21 14:25:22 host1 pptpd[4557]: MGR: No free connection slots or I
Ps - no more clients can connect!
Apr 21 14:25:22 host1 pptpd[4565]: CTRL: Client 172.16.0.1 control co
nnection started
Apr 21 14:25:22 host1 pptpd[4565]: CTRL: EOF or bad error reading ctr
l packet length.
Apr 21 14:25:22 host1 pptpd[4565]: CTRL: couldn't read packet header
(exit)
Apr 21 14:25:22 host1 pptpd[4565]: CTRL: CTRL read failed

The tunnel is set to persist so it will go down and come back up by itself...go figure. I am no Linux guru but I would be obliged to anyone who could help me. I am good with Cisco stuff if anyone would like a favor in return.

Seamus