Service Manual

If you use the “no prepend” option, the Local-AS does not prepend to the updates received from the eBGP peer. If you do not select
“no prepend” (the default), the Local-AS is added to the rst AS segment in the AS-PATH. If an inbound route-map is used to
prepend the as-path to the update from the peer, the Local-AS is added rst. For example, consider the topology described in the
previous illustration. If Router B has an inbound route-map applied on Router C to prepend "65001 65002" to the as-path, the
following events take place on Router B:
1. Receive and validate the update.
2. Prepend local-as 200 to as-path.
3. Prepend "65001 65002" to as-path.
Local-AS is prepended before the route-map to give an impression that update passed through a router in AS 200 before it reached
Router B.
BGP4 Management Information Base (MIB)
The FORCE10-BGP4-V2-MIB enhances Dell Networking OS BGP management information base (MIB) support with many new
simple network management protocol (SNMP) objects and notications (traps) dened in draft-ietf-idr-bgp4-mibv2-05. To see
these enhancements, download the MIB from the Dell website.
NOTE: For the Force10-BGP4-V2-MIB and other MIB documentation, refer to the Dell iSupport web page.
Important Points to Remember
• Because eBGP packets are not controlled by the ACL, packets from BGP neighbors cannot be blocked using the deny ip
command.
• The f10BgpM2AsPathTableEntry table, f10BgpM2AsPathSegmentIndex, and f10BgpM2AsPathElementIndex are used to retrieve
a particular ASN from the AS path. These indices are assigned to the AS segments and individual ASN in each segment starting
from 0. For example, an AS path list of {200 300 400} 500 consists of two segments: {200 300 400} with segment index 0 and
500 with segment index 1. ASN 200, 300, and 400 are assigned 0, 1, and 2 element indices in that order.
• Unknown optional transitive attributes within a given path attribute (PA) are assigned indices in order. These indices correspond
to the f10BgpM2PathAttrUnknownIndex eld in the f10BgpM2PathAttrUnknownEntry table.
• Negotiation of multiple instances of the same capability is not supported. F10BgpM2PeerCapAnnouncedIndex and
f10BgpM2PeerCapReceivedIndex are ignored in the peer capability lookup.
• Congure inbound BGP soft-reconguration on a peer for f10BgpM2PrexInPrexesRejected to display the number of prexes
ltered due to a policy. If you do enable BGP soft-reconfig, the denied prexes are not accounted for.
• F10BgpM2AdjRibsOutRoute stores the pointer to the NLRI in the peer's Adj-Rib-Out.
• PA Index (f10BgpM2PathAttrIndex eld in various tables) is used to retrieve specic attributes from the PA table. The Next-Hop,
RR Cluster-list, and Originator ID attributes are not stored in the PA Table and cannot be retrieved using the index passed
in
command. These elds are not populated in f10BgpM2PathAttrEntry, f10BgpM2PathAttrClusterEntry, and
f10BgpM2PathAttrOriginatorIdEntry.
• F10BgpM2PathAttrUnknownEntry contains the optional-transitive attribute details.
• Query for f10BgpM2LinkLocalNextHopEntry returns the default value for Link-local Next-hop.
• RFC 2545 and the f10BgpM2Rfc2545Group are not supported.
• An SNMP query displays up to 89 AS paths. A query for a larger AS path count displays as "…" at the end of the output.
• SNMP set for BGP is not supported. For all peer conguration tables (f10BgpM2PeerCongurationGroup,
f10BgpM2PeerRouteReectorCfgGroup, and f10BgpM2PeerAsConfederationCfgGroup), an SNMP set operation returns an
error. Only SNMP queries are supported. In addition, the f10BgpM2CfgPeerError, f10BgpM2CfgPeerBgpPeerEntry, and
f10BgpM2CfgPeerRowEntryStatus elds are to hold the SNMP set status and are ignored in SNMP query.
• The AFI/SAFI is not used as an index to the f10BgpM2PeerCountersEntry table. The BGP peer’s AFI/ SAFI (IPv4 Unicast or IPv6
Multicast) is used for various outbound counters. Counters corresponding to IPv4 Multicast cannot be queried.
• The f10BgpM2[Cfg]PeerReectorClient eld is populated based on the assumption that route-reector clients are not in a full
mesh if you enable BGP client-2-client reflection and that the BGP speaker acting as reector advertises routes
learned from one client to another client. If disabled, it is assumed that clients are in a full mesh and there is no need to advertise
prexes to the other clients.
• High CPU utilization may be observed during an SNMP walk of a large BGP Loc-RIB.
Border Gateway Protocol IPv4 (BGPv4)
177