FRRouting Release 10.6.0 [Download]

FRRouting Release 10.6.0 Available for Download

We are pleased to announce FRR release 10.6.0.

FRR 10.6.0 brings a long list of enhancements and fixes with 1585 commits from 86 developers.

Thanks to all contributors!

Debian Packages - https://deb.frrouting.org

RPM Packages - https://rpm.frrouting.org

Snaps - https://snapcraft.io/frr

Docker - quay.io/frrouting/frr

Release Overview

New Features Highlight

• LTTng traces for BFD
  • Added support for LTTng tracing in BFD, enabling improved observability and debugging capabilities for BFD sessions.
• Extended weights for the next-hop
  • Support for 16-bit next-hop weights has been added, aligning with capabilities introduced in Linux kernel v6.12. FRRouting now supports extended weight values for improved ECMP distribution accuracy.
• BGP ECMP by using Underlay Weights
  • BGP now provides a configuration option to use underlay-computed weights for ECMP path selection. Traditionally, BGP derives ECMP weights from extended community attributes. With this enhancement, operators may configure BGP to instead use underlay weights when calculating traffic distribution.
• BGP ECMP by using Next-next nodes (Next-hop dependent characteristic) count
  • An additional ECMP weighting mode has been introduced based on next-hop dependent characteristics (e.g., next-next-hop node count).
  • When bgp bestpath bandwidth ignore is enabled and next-hop characteristics are present, BGP uses the number of underlying next-next-hop paths to derive ECMP weights.
• BGP software version capability encoding changes
  • Encoding has been updated to align with version 15 of draft-abraitis-bgp-version-capability. Starting with this revision: The TLV length field is no longer explicitly encoded. The capability length field alone determines the encoding size.
  • A new command has been introduced bgp default software-version-capability latest-encoding. This enables use of the updated encoding format.
• BGP SRv6 unicast at the global (default) VRF
  • Support has been added for global (default VRF) unicast routing over an SRv6 core:
    • Global IPv4 over SRv6 Core (RFC 9252)
    • Global IPv6 over SRv6 Core (RFC 9252)
  • This enables SRv6-based transport for IPv4 and IPv6 services in the default routing table.
• BGP IPv6 VTEP support
  • Comprehensive IPv6 VTEP support has been added throughout the EVPN codebase. This enables EVPN deployments using IPv6 tunnel endpoints while maintaining full backward compatibility with IPv4 VTEPs.
• BGP multiple labels support for labeled unicast
  • FRRouting now supports multiple MPLS labels in BGP Labeled Unicast.
    • Previously: If multiple labels were received, only the bottom label was used.
    • Now: Up to 10 labels are supported per prefix.
  • This aligns with the maximum label stack depth defined in RFC 3017 and RFC 8277.
• BGP graceful restart for EVPN
  • Graceful Restart capability has been extended to EVPN address families. This aligns EVPN behavior with existing graceful restart support for IPv4 and IPv6 unicast.
• Implement forwarding-address-self command for OSPFv2
  • A new forwarding-address-self command has been implemented for OSPFv2. By default, when redistributing routes into OSPF, the forwarding address field is set based on the original next-hop. When forwarding-address-self is configured, the router sets the forwarding address field to its own address.
  • This is particularly useful in multi-homed (ECMP) ASBR deployments where traffic must be directed explicitly to the redistributing router.
• PIM/PIMv6 filter support for IGMP/MLD joins
  • Added support for filtering IGMP (IPv4) and MLD (IPv6) joins in PIM and PIMv6.
• libyang3 changes
  • Binary packages are now built against libyang3 by default. Users requiring libyang2 may still recompile FRRouting with libyang2 support.

Full Changelog: https://github.com/FRRouting/frr/compare/frr-10.6.0-dev...frr-10.6.0