Zebra

zebra is an IP routing manager. It provides kernel routing table updates, interface lookups, and redistribution of routes between different routing protocols.

Invoking zebra

Besides the common invocation options (Common Invocation Options), the zebra specific invocation options are listed below.

-b, --batch

Runs in batch mode. zebra parses configuration file and terminates immediately.

-k, --keep_kernel

When zebra starts up, don’t delete old self inserted routes.

-r, --retain

When program terminates, do not flush routes installed by zebra from the kernel.

-e X, --ecmp X

Run zebra with a limited ecmp ability compared to what it is compiled to. If you are running zebra on hardware limited functionality you can force zebra to limit the maximum ecmp allowed to X. This number is bounded by what you compiled FRR with as the maximum number.

-n, --vrfwnetns

When Zebra starts with this option, the VRF backend is based on Linux network namespaces. That implies that all network namespaces discovered by ZEBRA will create an associated VRF. The other daemons will operate on the VRF VRF defined by Zebra, as usual.

-o, --vrfdefaultname

When Zebra starts with this option, the default VRF name is changed to the parameter.

--v6-rr-semantics

The linux kernel is receiving the ability to use the same route replacement semantics for v6 that v4 uses. If you are using a kernel that supports this functionality then run Zebra with this option and we will use Route Replace Semantics instead of delete than add.

Configuration Addresses behaviour

At startup, Zebra will first discover the underlying networking objects from the operating system. This includes interfaces, addresses of interfaces, static routes, etc. Then, it will read the configuration file, including its own interface addresses, static routes, etc. All this information comprises the operational context from Zebra. But configuration context from Zebra will remain the same as the one from zebra.conf config file. As an example, executing the following show running-config will reflect what was in zebra.conf. In a similar way, networking objects that are configured outside of the Zebra like iproute2 will not impact the configuration context from Zebra. This behaviour permits you to continue saving your own config file, and decide what is really to be pushed on the config file, and what is dependent on the underlying system. Note that inversely, from Zebra, you will not be able to delete networking objects that were previously configured outside of Zebra.

Interface Commands

Standard Commands

interface IFNAME
interface IFNAME vrf VRF
shutdown
no shutdown

Up or down the current interface.

ip address ADDRESS/PREFIX
ipv6 address ADDRESS/PREFIX
no ip address ADDRESS/PREFIX
no ipv6 address ADDRESS/PREFIX

Set the IPv4 or IPv6 address/prefix for the interface.

ip address LOCAL-ADDR peer PEER-ADDR/PREFIX
no ip address LOCAL-ADDR peer PEER-ADDR/PREFIX

Configure an IPv4 Point-to-Point address on the interface. (The concept of PtP addressing does not exist for IPv6.)

local-addr has no subnet mask since the local side in PtP addressing is always a single (/32) address. peer-addr/prefix can be an arbitrary subnet behind the other end of the link (or even on the link in Point-to-Multipoint setups), though generally /32s are used.

ip address ADDRESS/PREFIX secondary
no ip address ADDRESS/PREFIX secondary

Set the secondary flag for this address. This causes ospfd to not treat the address as a distinct subnet.

description DESCRIPTION ...

Set description for the interface.

multicast
no multicast

Enable or disables multicast flag for the interface.

bandwidth (1-10000000)
no bandwidth (1-10000000)

Set bandwidth value of the interface in kilobits/sec. This is for calculating OSPF cost. This command does not affect the actual device configuration.

Enable/disable link-detect on platforms which support this. Currently only Linux and Solaris, and only where network interface drivers support reporting link-state via the IFF_RUNNING flag.

Virtual Routing and Forwarding

FRR supports VRF. VRF is a way to separate networking contexts on the same machine. Those networking contexts are associated with separate interfaces, thus making it possible to associate one interface with a specific VRF.

VRF can be used, for example, when instantiating per enterprise networking services, without having to instantiate the physical host machine or the routing management daemons for each enterprise. As a result, interfaces are separate for each set of VRF, and routing daemons can have their own context for each VRF.

This conceptual view introduces the Default VRF case. If the user does not configure any specific VRF, then by default, FRR uses the Default VRF.

Configuring VRF networking contexts can be done in various ways on FRR. The VRF interfaces can be configured by entering in interface configuration mode interface IFNAME vrf VRF.

A VRF backend mode is chosen when running Zebra.

If no option is chosen, then the Linux VRF implementation as references in https://www.kernel.org/doc/Documentation/networking/vrf.txt will be mapped over the Zebra VRF. The routing table associated to that VRF is a Linux table identifier located in the same Linux network namespace where Zebra started.

If the -n option is chosen, then the Linux network namespace will be mapped over the Zebra VRF. That implies that Zebra is able to configure several Linux network namespaces. The routing table associated to that VRF is the whole routing tables located in that namespace. For instance, this mode matches OpenStack Network Namespaces. It matches also OpenFastPath. The default behavior remains Linux VRF which is supported by the Linux kernel community, see https://www.kernel.org/doc/Documentation/networking/vrf.txt.

Because of that difference, there are some subtle differences when running some commands in relationship to VRF. Here is an extract of some of those commands:

vrf VRF

This command is available on configuration mode. By default, above command permits accessing the VRF configuration mode. This mode is available for both VRFs. It is to be noted that Zebra does not create Linux VRF. The network administrator can however decide to provision this command in configuration file to provide more clarity about the intended configuration.

netns NAMESPACE

This command is based on VRF configuration mode. This command is available when Zebra is run in -n mode. This command reflects which Linux network namespace is to be mapped with Zebra VRF. It is to be noted that Zebra creates and detects added/suppressed VRFs from the Linux environment (in fact, those managed with iproute2). The network administrator can however decide to provision this command in configuration file to provide more clarity about the intended configuration.

show ip route vrf VRF

The show command permits dumping the routing table associated to the VRF. If Zebra is launched with default settings, this will be the TABLENO of the VRF configured on the kernel, thanks to information provided in https://www.kernel.org/doc/Documentation/networking/vrf.txt. If Zebra is launched with -n option, this will be the default routing table of the Linux network namespace VRF.

show ip route vrf VRF table TABLENO

The show command is only available with -n option. This command will dump the routing table TABLENO of the Linux network namespace VRF.

By using the -n option, the Linux network namespace will be mapped over the Zebra VRF. One nice feature that is possible by handling Linux network namespace is the ability to name default VRF. At startup, Zebra discovers the available Linux network namespace by parsing folder /var/run/netns. Each file stands for a Linux network namespace, but not all Linux network namespaces are available under that folder. This is the case for default VRF. It is possible to name the default VRF, by creating a file, by executing following commands.

touch /var/run/netns/vrf0
mount --bind /proc/self/ns/net /var/run/netns/vrf0

Above command illustrates what happens when the default VRF is visible under var/run/netns/. Here, the default VRF file is vrf0. At startup, FRR detects the presence of that file. It detects that the file statistics information matches the same file statistics information as /proc/self/ns/net ( through stat() function). As statistics information matches, then vrf0 stands for the new default namespace name. Consequently, the VRF naming Default will be overriden by the new discovered namespace name vrf0.

For those who don’t use VRF backend with Linux network namespace, it is possible to statically configure and recompile FRR. It is possible to choose an alternate name for default VRF. Then, the default VRF naming will automatically be updated with the new name. To illustrate, if you want to recompile with global value, use the following command:

./configure --with-defaultvrfname=global

MPLS Commands

You can configure static mpls entries in zebra. Basically, handling MPLS consists of popping, swapping or pushing labels to IP packets.

MPLS Acronyms

LSR
Networking devices handling labels used to forward traffic between and through them.
LER
A Labeled edge router is located at the edge of an MPLS network, generally between an IP network and an MPLS network.

MPLS Push Action

The push action is generally used for LER devices, which want to encapsulate all traffic for a wished destination into an MPLS label. This action is stored in routing entry, and can be configured like a route:

[no] ip route NETWORK MASK GATEWAY|INTERFACE label LABEL

NETWORK ans MASK stand for the IP prefix entry to be added as static route entry. GATEWAY is the gateway IP address to reach, in order to reach the prefix. INTERFACE is the interface behind which the prefix is located. LABEL is the MPLS label to use to reach the prefix abovementioned.

You can check that the static entry is stored in the zebra RIB database, by looking at the presence of the entry.

zebra(configure)# ip route 1.1.1.1/32 10.0.1.1 label 777
zebra# show ip route
Codes: K - kernel route, C - connected, S - static, R - RIP,
O - OSPF, I - IS-IS, B - BGP, E - EIGRP, N - NHRP,
T - Table, v - VNC, V - VNC-Direct, A - Babel, D - SHARP,
F - PBR,
> - selected route, * - FIB route

S>* 1.1.1.1/32 [1/0] via 10.0.1.1, r2-eth0, label 777, 00:39:42

MPLS Swap and Pop Action

The swap action is generally used for LSR devices, which swap a packet with a label, with an other label. The Pop action is used on LER devices, at the termination of the MPLS traffic; this is used to remove MPLS header.

[no] mpls lsp INCOMING_LABEL GATEWAY OUTGOING_LABEL|explicit-null|implicit-null

INCOMING_LABEL and OUTGOING_LABEL are MPLS labels with values ranging from 16 to 1048575. GATEWAY is the gateway IP address where to send MPLS packet. The outgoing label can either be a value or have an explicit-null label header. This specific header can be read by IP devices. The incoming label can also be removed; in that case the implicit-null keyword is used, and the outgoing packet emitted is an IP packet without MPLS header.

You can check that the MPLS actions are stored in the zebra MPLS table, by looking at the presence of the entry.

show mpls table
zebra(configure)# mpls lsp 18 10.125.0.2 implicit-null
zebra(configure)# mpls lsp 19 10.125.0.2 20
zebra(configure)# mpls lsp 21 10.125.0.2 explicit-null
zebra# show mpls table
Inbound                            Outbound
Label     Type          Nexthop     Label
--------  -------  ---------------  --------
18     Static       10.125.0.2  implicit-null
19     Static       10.125.0.2  20
21     Static       10.125.0.2  IPv4 Explicit Null

Multicast RIB Commands

The Multicast RIB provides a separate table of unicast destinations which is used for Multicast Reverse Path Forwarding decisions. It is used with a multicast source’s IP address, hence contains not multicast group addresses but unicast addresses.

This table is fully separate from the default unicast table. However, RPF lookup can include the unicast table.

WARNING: RPF lookup results are non-responsive in this version of FRR, i.e. multicast routing does not actively react to changes in underlying unicast topology!

ip multicast rpf-lookup-mode MODE
no ip multicast rpf-lookup-mode [MODE]

MODE sets the method used to perform RPF lookups. Supported modes:

urib-only
Performs the lookup on the Unicast RIB. The Multicast RIB is never used.
mrib-only
Performs the lookup on the Multicast RIB. The Unicast RIB is never used.
mrib-then-urib
Tries to perform the lookup on the Multicast RIB. If any route is found, that route is used. Otherwise, the Unicast RIB is tried.
lower-distance
Performs a lookup on the Multicast RIB and Unicast RIB each. The result with the lower administrative distance is used; if they’re equal, the Multicast RIB takes precedence.
longer-prefix

Performs a lookup on the Multicast RIB and Unicast RIB each. The result with the longer prefix length is used; if they’re equal, the Multicast RIB takes precedence.

The mrib-then-urib setting is the default behavior if nothing is configured. If this is the desired behavior, it should be explicitly configured to make the configuration immune against possible changes in what the default behavior is.

Warning

Unreachable routes do not receive special treatment and do not cause fallback to a second lookup.

show ip rpf ADDR

Performs a Multicast RPF lookup, as configured with ip multicast rpf-lookup-mode MODE. ADDR specifies the multicast source address to look up.

> show ip rpf 192.0.2.1
Routing entry for 192.0.2.0/24 using Unicast RIB

Known via "kernel", distance 0, metric 0, best
* 198.51.100.1, via eth0

Indicates that a multicast source lookup for 192.0.2.1 would use an Unicast RIB entry for 192.0.2.0/24 with a gateway of 198.51.100.1.

show ip rpf

Prints the entire Multicast RIB. Note that this is independent of the configured RPF lookup mode, the Multicast RIB may be printed yet not used at all.

ip mroute PREFIX NEXTHOP [DISTANCE]
no ip mroute PREFIX NEXTHOP [DISTANCE]

Adds a static route entry to the Multicast RIB. This performs exactly as the ip route command, except that it inserts the route in the Multicast RIB instead of the Unicast RIB.

zebra Route Filtering

Zebra supports prefix-list s and Route Maps s to match routes received from other FRR components. The permit/deny facilities provided by these commands can be used to filter which routes zebra will install in the kernel.

ip protocol PROTOCOL route-map ROUTEMAP

Apply a route-map filter to routes for the specified protocol. PROTOCOL can be any or one of

  • system,
  • kernel,
  • connected,
  • static,
  • rip,
  • ripng,
  • ospf,
  • ospf6,
  • isis,
  • bgp,
  • hsls.
set src ADDRESS

Within a route-map, set the preferred source address for matching routes when installing in the kernel.

The following creates a prefix-list that matches all addresses, a route-map that sets the preferred source address, and applies the route-map to all rip routes.

ip prefix-list ANY permit 0.0.0.0/0 le 32
route-map RM1 permit 10
     match ip address prefix-list ANY
     set src 10.0.0.1

ip protocol rip route-map RM1

zebra FIB push interface

Zebra supports a ‘FIB push’ interface that allows an external component to learn the forwarding information computed by the FRR routing suite. This is a loadable module that needs to be enabled at startup as described in Loadable Module Support.

In FRR, the Routing Information Base (RIB) resides inside zebra. Routing protocols communicate their best routes to zebra, and zebra computes the best route across protocols for each prefix. This latter information makes up the Forwarding Information Base (FIB). Zebra feeds the FIB to the kernel, which allows the IP stack in the kernel to forward packets according to the routes computed by FRR. The kernel FIB is updated in an OS-specific way. For example, the Netlink interface is used on Linux, and route sockets are used on FreeBSD.

The FIB push interface aims to provide a cross-platform mechanism to support scenarios where the router has a forwarding path that is distinct from the kernel, commonly a hardware-based fast path. In these cases, the FIB needs to be maintained reliably in the fast path as well. We refer to the component that programs the forwarding plane (directly or indirectly) as the Forwarding Plane Manager or FPM.

The FIB push interface comprises of a TCP connection between zebra and the FPM. The connection is initiated by zebra – that is, the FPM acts as the TCP server.

The relevant zebra code kicks in when zebra is configured with the --enable-fpm flag. Zebra periodically attempts to connect to the well-known FPM port. Once the connection is up, zebra starts sending messages containing routes over the socket to the FPM. Zebra sends a complete copy of the forwarding table to the FPM, including routes that it may have picked up from the kernel. The existing interaction of zebra with the kernel remains unchanged – that is, the kernel continues to receive FIB updates as before.

The encapsulation header for the messages exchanged with the FPM is defined by the file fpm/fpm.h in the frr tree. The routes themselves are encoded in Netlink or protobuf format, with Netlink being the default.

Protobuf is one of a number of new serialization formats wherein the message schema is expressed in a purpose-built language. Code for encoding/decoding to/from the wire format is generated from the schema. Protobuf messages can be extended easily while maintaining backward-compatibility with older code. Protobuf has the following advantages over Netlink:

  • Code for serialization/deserialization is generated automatically. This reduces the likelihood of bugs, allows third-party programs to be integrated quickly, and makes it easy to add fields.
  • The message format is not tied to an OS (Linux), and can be evolved independently.

As mentioned before, zebra encodes routes sent to the FPM in Netlink format by default. The format can be controlled via the FPM module’s load-time option to zebra, which currently takes the values Netlink and protobuf.

The zebra FPM interface uses replace semantics. That is, if a ‘route add’ message for a prefix is followed by another ‘route add’ message, the information in the second message is complete by itself, and replaces the information sent in the first message.

If the connection to the FPM goes down for some reason, zebra sends the FPM a complete copy of the forwarding table(s) when it reconnects.

zebra Terminal Mode Commands

show ip route

Display current routes which zebra holds in its database.

Router# show ip route
Codes: K - kernel route, C - connected, S - static, R - RIP,
 B - BGP * - FIB route.

K* 0.0.0.0/0        203.181.89.241
S  0.0.0.0/0        203.181.89.1
C* 127.0.0.0/8      lo
C* 203.181.89.240/28      eth0
show ipv6 route
show interface
show ip prefix-list [NAME]
show route-map [NAME]
show ip protocol
show ipforward

Display whether the host’s IP forwarding function is enabled or not. Almost any UNIX kernel can be configured with IP forwarding disabled. If so, the box can’t work as a router.

show ipv6forward

Display whether the host’s IP v6 forwarding is enabled or not.

show zebra

Display various statistics related to the installation and deletion of routes, neighbor updates, and LSP’s into the kernel.

show zebra fpm stats

Display statistics related to the zebra code that interacts with the optional Forwarding Plane Manager (FPM) component.

clear zebra fpm stats

Reset statistics related to the zebra code that interacts with the optional Forwarding Plane Manager (FPM) component.