Using a UBR10K and NSG 9000 40G. I can't get the Depi tunnel to actually come up. I using what I thought was a very simple config just to get an understanding of how things work. I'm stumped. Have them connected directly together. Should be simple but I'm missing something obvious I'm sure.
Relevant config
l2tp-class DEPI_CLASS
hello 5
retransmit retries 5
retransmit timeout max 1
depi-class DEPI_CP
mode mpt
depi-tunnel DEPI_1600
dest-ip 10.30.4.101
tos 100
l2tp-class DEPI_CLASS
depi-class DEPI_CP
controller Modular-Cable 6/0/0
rf-channel 0 cable downstream channel-id 1
rf-channel 0 frequency 111000000 annex B modulation 256qam interleave 32
rf-channel 0 network-delay auto
rf-channel 0 depi-tunnel DEPI_1600 tsid 1
rf-channel 0 rf-power 43.0
no rf-channel 0 rf-shutdown
interface Cable6/0/0
downstream Modular-Cable 6/0/0 rf-channel 0
no cable packet-cache
cable bundle 1
cable upstream max-ports 4
cable upstream 0 connector 0
cable upstream 0 frequency 21400000
cable upstream 0 channel-width 6400000 6400000
cable upstream 0 threshold snr-profiles 28 25
cable upstream 0 threshold hysteresis 4
cable upstream 0 docsis-mode atdma
cable upstream 0 minislot-size 1
cable upstream 0 power-adjust continue 6
cable upstream 0 range-backoff 3 6
cable upstream 0 modulation-profile 226 224
cable upstream 0 equalization-coefficient
no cable upstream 0 shutdown
interface GigabitEthernet6/0/0
ip address 10.30.4.1 255.255.255.0
negotiation auto
interface Modular-Cable6/0/0:0
cable bundle 1
cable rf-bandwidth-percent 10 remaining ratio 100
cable fiber-node 10
downstream Modular-Cable 6/0/0 rf-channel 0
upstream Cable 6/0 connector 0
The tunnel comes up and drops immediately.
#sh depi tunnel
LocTunID RemTunID Remote Name State Remote Address Sessn L2TP Class
Count
1515959812 1515959812 NSG-1 nosess 10.30.4.101 0 DEPI_CLASS
#sh depi session configured
Session Name State Reason
Modular-Cable6/0/0:0 IDLE recv CDN
debug output:
DEPI L2TP: uid:7[10.30.4.101/1] Event , state Waiting for timer to connect -> Connecting
DEPI L2TP: uid:7[10.30.4.101/1] L2TUN socket session connect requested
DEPI L2TP: Received L2TUN socket message
DEPI L2TP: uid:7[10.30.4.101/1] Event , state Connecting -> Connecting
DEPI L2TP: uid:7[10.30.4.101/1] SHUTDOWN Reason: "recv CDN"
DEPI L2TP: uid:7[10.30.4.101/1] Event , state Connecting -> Waiting for timer to connect
I have the modules set to 36 qams, 32-4 interleave, m-cmts dynamic, freq range starting at 108Mhz. I have use DLM checked on the M-CMTS page.
Strangely, sometimes the interleavers on channels 1-6 (group 1) change themselves to 128-4. !?!
In the cmts I can't set the rf power to 44? I was using that because that's the max for 36 active ports in the qam. It accepts it like normal but show run will have no setting for rf power. It will take 45 or 43? I think changing the power levels was what triggered the interleave change but I'm not sure and can't get it to repeat at the moment.
Anyway if anyone can point out what I've missed that would be great.
CMTS looks OK.
Not reason for you problem, but
I recommend putting the DEPI gigabit ethernet ports into a VRF. No need for these subnets to accessible by cable modems / customers / internet.
Create the VRF :
vrf definition VRF-DEPI
description ** DEPI traffic to EQAM
rd your_BGP_AS_numbers_here:1000 <--- both these numbers are arbitrary, you can use whatever you want
!
address-family ipv4
exit-address-family
!
And then set your ports to be inside the VRF like this :
interface GigabitEthernet5/0/0
description ** some_switch_name:ge-0/0/0
vrf forwarding VRF-DEPI
ip address 10.xx.xx.1 255.255.255.252
load-interval 30
negotiation auto
I use this L2TP stanza, but I dont think it would fix your problem :
l2tp-class L2TP-CLASS-EQAM
retransmit retries 5
retransmit timeout max 2
In your case, the tunnel are coming up, but then edgeqam is shutting it due to one or more of the negotiated CMTS parameters being not acceptable
Your edgeqam config sounds generally OK. But maybe there is a step you have missed along the way. When we setup our NSG9000s I think we followed this rough sequence :
* Applications -> M-CMTS -> Settings -> DLM : ticked
* Platform -> General -> DTI1/2
- Use DTI clock : ticked
* Platform -> General -> Chassis -> GbeX
- General : Enable port
- IP address : 10.30.4.101, 255.255.255.0 <--- this is your config, and is probably OK for a single 1Gbps uplink. But if you are going to have many 1G links from CMTS to switch, and then 10G uplink from switch to edgeqam I would put /30 on the EQ 10G interface, and then add additional /30 virtual IPs for each additional 1G uplink that is being aggregated into this same 10G port.
* Platform -> General -> Chassis -> Global RF & QAM :
- Operation mode : 6B-6B-12B-12B, set all modules
- RF Template : frequency, set all modules
- QAM manager : M-CMTS (**NOT M-CMTS DYNAMIC**), set all modules
* Platform -> General -> Module 1 -> Ports -> Port 1
- RF enable : Tick
- Freq range : 108 (needs to be 3MHz below your lowest center freq)
- QAM per port : Pick however many QAMs you are planning to use eg, 24 or 32 or 36 etc. Probably best to not go higher than the number of channels you are planning to use, as more channels will lower the allowed power range window.
- Power per QAM : leave at default
Repeat above step for each other port and card
Now the cards are configured, activate dynamic channel config via CMTS :
* Platform -> General -> Chassis -> Global RF & QAM :
- QAM manager : M-CMTS dynamic, set all modules
Now the EQ will learn all the DS parameters from the CMTS via the DEPI tunnel
* channel freq
* channel power
* shut/no shut
etc
On the CMTS you have to keep these in mind :
* configured channel power must be within the range the EQ accepts
See the range on Platform -> General -> Module X -> ports.
eg 24 qam per port means Allowed DS PWR is 38-46 dBmV
* Nominated TSID on CMTS tells the EQ which channel you are configuring.
Eg TSID 1-36 is EQ port 1/1 CH 1-36. TSID 37-71 is EQ port 1/2 CH 1-36
---
I wrote up some NSG instructions once before
Here is the thread
https://docsis.org/forums/docsis-chat/ubr10k-nsg9k-40g
Maybe read that one too, see if there is anything I have missed
Hope that helps!
I had hit the global "Mute all RFs" button when I first reset the unit. Works as advertised. When I ticked the "enable rf" on port 1 I didn't realize the global shutdown remained. Unmuting from global let everything come up.
I had actually found that previous thread and used it during my setup. That was extremely helpful and would have worked perfectly if I hadn't flubbed the mute button.
I will heed your advice concerning the vrf and smaller subnets. This was just my attempt to get things working with the most basic setup.
Once again, thanks for the help.
With the IP addressing, after some trial and error, I was forced to put a bit of thought into how to make it easy to modify / expand later without a major reconfig on CMTS and EQ.
For example, we initially had 2020 cards with SPA24.
Later the 2020 cards were replaced with 3G60, sometimes with or without 1G SPA (SPA24).
Later we will probably want use 3G SPA (SPA72) or 6G SPA (SPA144).
So I decided on this addressing/numbering system :
10.aa.bb.cc
aa = a code for each site
bb = the interface eg :
.10 = modular cable 1/0/x
.11 = modular cable 1/1/x
.12 = modular cable 1/2/x
.50 = modular cable 5/0/x
.51 = modular cable 5/1/x
cc = the subslot eg:
.1 = x/x/0 (eg 1/2/0)
.5 = x/x/1 (eg 5/0/1)
.9 = x/x/2 (eg 8/1/2)
The edgeqam is CMTS IP + 1
eg
for controller MC 1/2/0, the CMTS IP would be 10.aa.12.1/30, and the EQ IP would be 10.aa.12.2/30
for controller MC 5/0/1, the CMTS IP would be 10.aa.50.5/30, and the EQ IP would be 10.aa.50.6/30
for controller MC 8/1/2, the CMTS IP would be 10.aa.81.9/30, and the EQ IP would be 10.aa.81.10/30
Each uplink port from the CMTS is 1Gbps.
For a small site with just a couple of SPA24 you can plug these direct to the EQ.
For a larger site with many ports (eg 3+ 3G60 cards), you feed these 1Gbps links into a switch which then has 10G uplinks to the EQ.
On the EQ, you can put up to 10 of these 1Gbps subnets on a single 10G EQ port. Maybe only do up to 9 to give yourself a little bit of clearance.
For each 10G port on EQ, put one subnet as the first IP eg 10.aa.12.2/30, and then go to the virtual IPs and add 10.aa.50.6/30 and 10.aa.81.10/30.
Above would result in up to 3Gbps of traffic being directed to this 10G port.
Hope that makes sense!
I can see this post saving me more future headaches than I can possibly imagine. Thanks again, that's super helpful.