
uBR10012 with MC5x20 Cards
These bullet points indicate the situations that are tracked for failover initiation. These are considered the
most prevalent issues that could allow modems to drop offline.
• Shut down the active cable interface (works, but not supported)
• On-line insertion removal (OIR) of active line card
• Software CLI based commands – hccp “g” switch “m”
• Software crash of the active line card
• DS cabling failure via “KeepAlive” feature
• Resetting the line card - Hw-module subslot x/y reset
•
Power outage on Working linecard – cable power off x/y
A DS failure could be from a bad internal upconverter or cable between the 10K and RF Switch. The
Keepalive feature tracks all communication on all the US ports of a particular MAC domain. When there
is no communication what-so-ever, a failover will initiate based on some user-configurable thresholds and
timers.
Since the 5x20 card is really 5, 1x4 MAC domains, you can make switch groups based on MAC domains.
A MAC domain is 1 DS and all its associated USs. These MAC domains are user-configurable with the
virtual interface feature released in 15BC1 IOS, but are statically set as 1x4 MAC domains by default.
Note: If an upstream connector is assigned for virtual interface configurations, the entire linecard will
failover when a fault occurs instead of single MAC domains or ASIC companions.
The 5x20 card is labeled with USs 0-19 and DSs 0-4. DS 0 is associated with US 0-3, DS 1 is tied to USs
4-7, etc. The configuration in the CMTS is still considered to be USs 0-3 regardless of which MAC
domain. If you want to configure DS 4, US 17 would actually be considered US 1 since DS 4 uses USs
16-19 and 16 refers to US 0, etc.
If you shut down the Working interface, the protocol will initiate a failover via the configuration file. A
failover is not initiated by US ports being “shut”. Pulling an upstream cable from ONE port on a line
card is not generally considered a valid "event" to cause an N+1 line card failover. It is essentially
impossible to distinguish this from a disconnected attenuator in a fiber node or amplifier (used for
operational maintenance). Pulling the linecard out of the chassis, disconnecting the downstream cable
between the linecard & RF Switch, or some other software or hardware type fault on the card itself are all
considered valid "N+1" failover events.
On the uBR10K, you could even use the linecard power down command, which cuts power to the line
card, and thus causes a failure.
10K#cable power off ?
<slot>/<card> slot [5-8] card [0/1]
One interface will be designated as a “Protect” interface and all the commands will be configured in that
interface to backup all the interfaces in its Group. If a linecard is removed, one or more MAC domains
will be removed and a “Protect” card will be initiated to back it up. The configuration in the 10K will
make the appropriate RF Switch relays to switchover.
Kommentare zu diesen Handbüchern