Cisco uBR10-MC5X20S Manual de usuario descargar pdf (Pagina 12)

message at “hellotime” interval in milliseconds to check the sanity of the Working chassis. If there is no

Hello-Acknowledgement for more than a time period equal to "holdtime", then it is declared the Working

has failed and initiates a switchover. The “holdtime” has to be at least 3 times larger than the “hellotime”.

The default is 5000 for hello and 15000 ms for hold. The max is 25000 ms. Since the 10K solution is

entirely intra-HCCP, don’t change this timer.

Tracking

By default, an HCCP interface tracks itself. So when Keepalive is enabled and it detects no incoming

upstream packets, it will failover. The “Track” command can also be used to track an uplink interface.

For example, if Working has a dedicated uplink (eg. GE) path and Protect has its own, these uplink

interfaces can be tracked. When one fails, the cable interface will failover to the standby. Since in the

10K solution, it seems Working and Protect may share the same uplink, this “Track” command is not

needed for this scenario.

To switch an entire linecard, 5 MAC domains must switch when using the 5x20 cards. One way may be

to configure virtual interfaces. Another way would be to use the CLI; power off or hw-module reset

commands.

KeepAlive

The purpose of this feature is to cover bad cabling between the CMTS and RF Switch. The way we detect

an HFC failure is to count incoming packets on all upstreams.

If within 3 keepalive periods there is no incoming packets (range requests/response, station maintenance,

data, etc.) on all of USs belonging to one DS, the line protocol will be down and HCCP assumes

something is wrong in that channel and will switchover. Remember, if there is a real HFC problem, the

switchover will occur, but won't do any good because it's still on the same bad HFC plant. This feature is

meant to cover failures in components that are not common between the Protect and Working interfaces

such as upconverters and certain cabling.

Keepalive is off by default on cable interfaces with older IOS, but is defaulted to a value of 10 seconds in

the newer code. Set keepalive as low as possible, which is 1 second.

It may be advantageous to set no keepalive on the Protect interfaces so it doesn’t fail back to the Working

interface if all the modems go offline.

Tip: If routine maintenance will be taking place in the cable plant (balancing amplifiers, etc) and signal

loss is eminent that will affect all the US ports of a MAC domain, "lockout" that interface, and its

ASIC companion interface, until the work is done.

Failover Times

DOCSIS 1.0 specifies 600 ms as DS sync loss, but it doesn't specify what the CM should do after sync

loss. Most CMs don’t re-register immediately after sync loss. Refer to CSCdv71490 UBR10K:

Interoperability of N+1 feature with third party modems.

Station Maintenance for modems is 1 second per modem until you get to 20 modems, then it’s every 20

seconds when there are 20 or more modems in the MAC domain. This used to be set for every 25

seconds. When HCCP is configured, the top number is 15 seconds for a higher probability of a successful

failover. This is because of the T4 timer in modems that is set at 30 seconds. If a modem were to

1 2 ... 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Sin comentarios

Cisco uBR10-MC5X20S Manual de usuario Pagina 12