08-19-2003 05:21 PM - edited 03-02-2019 09:43 AM
Hi all:
We have ADSL service and have some weird problems. Some of our customers PPPoE/VPDN sessions would stay connected with NAS(LNS) routers for hours even they are idle:
Interface User Mode Idle Peer Address
Vi189 johndo1@company. PPPoVPDN 1d4h:02:05 xxx.xxx.xxx.xxx
Vi191 johndo2@company. PPPoVPDN 1d1h:01:11 xxx.xxx.xxx.xxx
Vi193 johndo3@company. PPPoVPDN 23:02:33 xxx.xxx.xxx.xxx
Vi195 johndo4@company. PPPoVPDN 18:02:14 xxx.xxx.xxx.xxx
The real problem is that those customers will call in and complain that they cant connect to Internet. We have to clear those virtual interfaces up by command clear interface virtual-access # all the time. Now we are creating a script doing this. But it wont resolve the actual problem.
Seem to me, this magnitude of the problem relates the configuration command virtual-template 1 pre-clone xxx; the larger the # xxx is, the severer the problem is. When I drop the number down to 50, fewer customers complain. But the routers CPU utilization starts getting higher than normal (one of routers CPU actually reached 91% when over 2000 customers terminated on that router).
I also noticed that the most of those customers with such problem configure their home ADSL modems with so-called nailed-in connection instead of automatically timeout after a certain idle time. Personally I like this option since it seems to refresh the PPP sessions but would not destroy the VPDN tunnels; unfortunately, most of those customers dont like this idea.
Could anyone here explains on this issue in details or help me find some information (such as white papers, URLs. I am very interested in the theory behind this)? The most important thing question is: how could I fix this from my site?
Thanks
08-25-2003 12:56 PM
Here are a few documents on VPDN. There is also a support page for the same.
http://www.cisco.com/en/US/tech/tk801/tk703/technologies_tech_note09186a0080094586.shtml
08-25-2003 04:30 PM
thanks.
the problem is that the "dead" l2tp virtual access interfaces don't get closed as they should. i duplicated the problem by turning off adsl modem (test account); according the cisco reading materials, the cooresponding virtual access interface should be closed too. No, it was not. it was there for very long time even the remote adsl modem was shut down; because of this, the test account could not be used it until i cleared the virtual access interface.
here is the config:
vpdn-group ISP-LNS
accept-dialin
protocol l2tp
virtual-template 1
local name isp
l2tp tunnel timeout no-session 10
...
...
interface Virtual-Template1
ip unnumbered Loopback1
ip tcp adjust-mss 1452
no logging event link-status
peer default ip address pool sun-adsl
no keepalive
ppp mtu adaptive
ppp lcp predictive
ppp authentication chap pap
ppp ipcp predictive
ppp timeout idle 600
i seriously suspect that there are some kind of bugs for this problem
08-27-2003 09:18 AM
Hey had a similar problem.
What does your config look like ?
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide