Hi Folks,
I have been working with a couple of clients who have ADSL WANs (over a private IP provider, not the public Internet) who are experimenting with QoS and LFI.
The first client runs c837-k9o3sy6-mz.123-2.T.bin on a bunch of
837s, whilst the second runs c1700-k9o3sy7-mz.122-4.T7.bin on 1751s with ADSL WICs.I was able to add QoS and LFI to the 837s without much drama but I've had problems with the 1751s. When I add LFI to the 1751s the central site cannot ping PCs on the remote LAN with packets larger than 600 bytes. Interestingly it can ping the remote router itself with large packets. Switching ppp multilink off restores correct operation.
The 1751 does seem to get into multilink LFI mode as there is the usual pause when I enable multilink and then a "sh int di1" starts to show the virtual-access interfaces and "interleaves" appear in the stats display.
I assume that 600 bytes corresponds to 10 msec of the 512 Kbps bandwidth as I'm using "ppp multilink fragment delay 10".
The routers are not entirely identical as they have had quite different histories, so I'm assuming that the problem 1751 has a setting that is preventing fragment reassembly to the LAN interface???
The clients are't keen on me posting full configs, but the following may be enough to allow somone to spot the obvious error ;-)
The (working) 837 router looks like:
version 12.3 ... ip cef ... policy-map QoS-XXX ... policy-map DiffServ-Marking-XXX ... interface Ethernet0 ip address x.x.x.x 255.255.255.0 service-policy input DiffServ-Marking-XXX ip route-cache flow no ip mroute-cache hold-queue 100 out ... interface ATM0 no ip address no ip mroute-cache no atm ilmi-keepalive pvc 8/35 encapsulation aal5snap protocol ip inarp protocol ppp dialer dialer pool-member 1 ! dsl operating-mode auto dsl power-cutback 0 bridge-group 1 hold-queue 224 in ! interface ATM0.1 point-to-point no ip mroute-cache ... interface Dialer1 bandwidth 512 ip address negotiated service-policy output QoS-XXX encapsulation ppp ip tcp header-compression no ip mroute-cache dialer pool 1 dialer idle-timeout 0 dialer persistent delay initial 30 dialer persistent delay 30 dialer persistent max-attempts 3 dialer-group 1 keepalive 10 3 ppp authentication chap callin ppp chap hostname snipped-for-privacy@xxxxx.xxxxx.xxxxx ppp chap password 7 xxxxxxxxxxxxxxxxx ppp multilink ppp multilink fragment delay 10 ppp multilink interleave hold-queue 224 in
The (broken) 1751 router looks like:
version 12.2 ... ip cef ... policy-map QoS-XXX ... policy-map DiffServ-Marking-XXX ... interface ATM0/0 no ip address no ip mroute-cache atm vc-per-vp 256 no atm ilmi-keepalive pvc 8/35 encapsulation aal5snap protocol ip inarp protocol ppp dialer dialer pool-member 1 ! dsl operating-mode auto no fair-queue bridge-group 1 hold-queue 224 in ! interface ATM0/0.1 point-to-point no ip mroute-cache ... interface FastEthernet0/0 ip address x.x.x.x 255.255.255.0 service-policy input DiffServ-Marking-XXX no ip mroute-cache speed auto hold-queue 100 out ... interface Dialer1 bandwidth 512 ip address negotiated service-policy output QoS-XXX encapsulation ppp ip tcp header-compression no ip mroute-cache dialer pool 1 dialer idle-timeout 0 dialer persistent delay initial 30 dialer persistent delay 30 dialer persistent max-attempts 3 dialer-group 1 ppp authentication chap callin ppp chap hostname snipped-for-privacy@xxxxx.xxxxx.xxxxx ppp chap password 7 xxxxxxxxxxxxxxxxxxxxxxx ppp multilink ppp multilink fragment delay 10 ppp multilink interleave
Can anyone spot what's wrong? I suppose it could be the router model, the IOS version, or perhaps the ISP's core router config that are at fault, but I think I'm doing something that prevents reassembly.
Any thoughts greatly appreciated.
Cheers, Dieter.