H3C Technologies H3C S6800 Series Switches - Manual (Page 34)

21
a. Marks the FEC-label mappings learned from the session as stale.
b. Starts the Reconnect timer received from the GR restarter.
3. After LDP completes restart, the GR restarter re-establishes an LDP session with the GR helper.
{
If the LDP session is not set up before the Reconnect timer expires, the GR helper deletes the stale
FEC-label mappings and the corresponding MPLS forwarding entries.
{
If the LDP session is successfully set up before the Reconnect timer expires, the GR restarter sends
the remaining time of the MPLS Forwarding State Holding timer to the GR helper.
The remaining time is sent as the LDP Recovery time.
4. After the LDP session is re-established, the GR helper starts the LDP Recovery timer.
5. The GR restarter and the GR helper exchange label mappings and update their MPLS forwarding
tables.
The GR restarter compares each received label mapping against stale MPLS forwarding entries. If
a match is found, the restarter deletes the stale mark for the matching entry. Otherwise, it adds a
new entry for the label mapping.
The GR helper compares each received label mapping against stale FEC-label mappings. If a
match is found, the helper deletes the stale mark for the matching mapping. Otherwise, it adds the
received FEC-label mapping and a new MPLS forwarding entry for the mapping.
6. When the MPLS Forwarding State Holding timer expires, the GR restarter deletes all stale MPLS
forwarding entries.
7. When the LDP Recovery timer expires, the GR helper deletes all stale FEC-label mappings.
LDP NSR
LDP nonstop routing (NSR) backs up protocol states and data (including LDP session and LSP information)
from the active process to the standby process. When the LDP primary process fails, the backup process
seamlessly takes over primary processing. The LDP peers are not notified of the LDP interruption. The LDP
peers keep the LDP session in Operational state, and the forwarding is not interrupted.
The LDP primary process fails when one of the following situations occurs:
• The primary process restarts.
• The IRF member device where the primary process resides fails.
• The IRF member device where the primary process resides performs an ISSU.
• The LDP process' position determined by the process placement function is different from the
position where the LDP process is operating.
Choose either LDP NSR or LDP GR to ensure continuous traffic forwarding:
• Device requirements
{
LDP NSR must run on an IRF fabric formed by more than two IRF member devices, and the
primary and backup processes for LDP reside on different devices.
{
LDP GR can run on a device that does not form an IRF fabric with other devices.
• LDP peer requirements
{
With LDP NSR, LDP peers of the local device are not notified of any switchover event on the local
device. The local device does not require help from a peer to restore the MPLS forwarding
information.