Hi All.. This is my first post to the forum and i am very happy to contact you.
I came here because i have problems about put a Linux Border Router to improve qos in my docsis implementation.
I Have over 500 CM's on a CMTS Cisco uBR7223 in my company connected to a dedicated point to point Internet connection (by optic fiber to a cisco router in the ISP side at 100 Mbps)... The traffic it is around the 50 Mbps DownStream and 25 Mbps UpStream in peak time. My Linux Router is based on Debian Linux and actually only make routing jobs, nothing about qos, but really do not work and i dont know why...
In Smaller conditions the Linux make a very good job, but when i put it between the CMTS and the Internet the problems begin. The speed decrease very much, many lost packets, collisions, and problems like that... The router rules in the Linux are well, as well in the CMTS...
The Server hardware is an Intel Dual Core processor with 4 GB of RAM and 2 Integrated NIC's Intel 10/100/1000... In the best conditions i have a maximum of 10 Mbps of DownStream and 5 Mbps for UpStream and the Internet goes very slow, is this problem caused by not enough hardware or incompatibility? or simply it is not so simply to get run?
This is that i am trying to do: CM's <-> CMTS <-> Linux Router <-> Internet
And this is the normal implementation: CM's <-> CMTS <-> Internet
It is possible to get work something like this? Anyone here made this before?
I know that Linux Routers can do a great job, but maybe here is a compatibility problem between Cisco and Linux. I was reading about
BGP but im not sure that this is important in this case.
I really appreciate your help.
Thanks...
Jag.
BGP Probably is not the case, what you are describing sounds like duplex mismatch, i.e. your linux box or cmts is only linking at half duplex. or possibly that one of them is linking at 10 megs instead of 100 megs. On the cmts you can run the command
sh interface fastEthernet 0/0
or whatever your ethernet interface is and look to see what duplex/speed setting you are connected to.
on debian you need to run ethtool you will need to read man ethtool to find the cli flags that need to be set
Hi
Maybe we would like to see some outputs from your linux box like top? ifconfig eth0 and for eth1 for CRC errors? is connected directly to the cmts? the linux box makes NAT?
Bye
Tomorrow i will send you some other information...
Regards...
jag.
Hi
Well I would like if you can erase the counters to see those 5000 aprox errors...
I think is fine your card from the cmts. Now we need some reports from your linux.
See you.
Here's an alternative idea. Use M0n0wall. It's awesome.
emkowale
This is the sh interface output from my CMTS... This interface is directly connected to Internet.
I will put the Linux outputs tomorrow...
FastEthernet1/0 is up, line protocol is up
Hardware is DEC21140, address is 00d0.5822.d81c (bia 00d0.5822.d81c)
Description: Internet Link
Internet address is 192.168.31.118/30
MTU 1500 bytes, BW 100000 Kbit, DLY 100 usec,
reliability 255/255, txload 46/255, rxload 110/255
Encapsulation ARPA, loopback not set
Keepalive set (10 sec)
Full-duplex, 100Mb/s, 100BaseTX/FX
ARP type: ARPA, ARP Timeout 04:00:00
Last input 00:00:00, output 00:00:00, output hang never
Last clearing of "show interface" counters never
Input queue: 0/75/395/265 (size/max/drops/flushes); Total output drops: 0
Queueing strategy: fifo
Output queue: 0/40 (size/max)
5 minute input rate 43368000 bits/sec, 7957 packets/sec
5 minute output rate 18365000 bits/sec, 7024 packets/sec
624714957 packets input, 1057694323 bytes
Received 1459 broadcasts, 0 runts, 0 giants, 12 throttles
5576 input errors, 0 CRC, 0 frame, 0 overrun, 5553 ignored
0 watchdog
0 input packets with dribble condition detected
553967074 packets output, 2352682127 bytes, 0 underruns
0 output errors, 0 collisions, 1 interface resets
0 babbles, 0 late collision, 0 deferred
0 lost carrier, 0 no carrier
0 output buffer failures, 0 output buffers swapped out
What do you think about this?
Tomorrow i will post other info...
jag.
Hola que tal?
Esas pruebas las haces desde tu linux? o desde la red hfc?
Verificaste que tu proveedor te este respetando lo contratado?
Viste el syslog que no te este acusando algun problema?
Yo en mi caso, uso linux y tambien en otros lugares openbsd.. y esta andando barbaro con 300mb de downstream y unos 7000 usuarios.
Saludos.
nixsis.lr@gmail.com