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 TIME, --graceful_restart TIME

If this option is specified, the graceful restart time is TIME seconds. Zebra, when started, will read in routes. Those routes that Zebra identifies that it was the originator of will be swept in TIME seconds. If no time is specified then we will sweep those routes immediately.

-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.

See also

Virtual Routing and Forwarding

-o, --vrfdefaultname

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

See also

Virtual Routing and Forwarding

-z <path_to_socket>, --socket <path_to_socket>

If this option is supplied on the cli, the path to the zebra control socket(zapi), is used. This option overrides a -N <namespace> option if handed to it on the cli.

--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.

--asic-offload [notify_on_offload|notify_on_ack]

The linux kernel has the ability to use asic-offload ( see switchdev development ). When the operator knows that FRR will be working in this way, allow them to specify this with FRR. At this point this code only supports asynchronous notification of the offload state. In other words the initial ACK received for linux kernel installation does not give zebra any data about what the state of the offload is. This option takes the optional paramegers notify_on_offload or notify_on_ack. This signals to zebra to notify upper level protocols about route installation/update on ack received from the linux kernel or from offload notification.

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

Up or down the current interface.

ip address ADDRESS/PREFIX
ipv6 address ADDRESS/PREFIX

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

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.

description DESCRIPTION ...

Set description for the interface.

multicast

Enable or disables multicast flag for the interface.

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 only where network interface drivers support reporting link-state via the IFF_RUNNING flag.

In FRR, link-detect is on by default.

Nexthop Tracking

Nexthop tracking doesn't resolve nexthops via the default route by default. Allowing this might be useful when e.g. you want to allow BGP to peer across the default route.

ip nht resolve-via-default

Allow IPv4 nexthop tracking to resolve via the default route. This parameter is configured per-VRF, so the command is also available in the VRF subnode.

ipv6 nht resolve-via-default

Allow IPv6 nexthop tracking to resolve via the default route. This parameter is configured per-VRF, so the command is also available in the VRF subnode.

Administrative Distance

Administrative distance allows FRR to make decisions about what routes should be installed in the rib based upon the originating protocol. The lowest Admin Distance is the route selected. This is purely a subjective decision about ordering and care has been taken to choose the same distances that other routing suites have choosen.

Protocol Distance
System 0
Kernel 0
Connect 0
Static 1
NHRP 10
EBGP 20
EIGRP 90
BABEL 100
OSPF 110
ISIS 115
OPENFABRIC 115
RIP 120
Table 150
SHARP 150
IBGP 200
PBR 200

An admin distance of 255 indicates to Zebra that the route should not be installed into the Data Plane. Additionally routes with an admin distance of 255 will not be redistributed.

Zebra does treat Kernel routes as special case for the purposes of Admin Distance. Upon learning about a route that is not originated by FRR we read the metric value as a uint32_t. The top byte of the value is interpreted as the Administrative Distance and the low three bytes are read in as the metric. This special case is to facilitate VRF default routes.

Route Replace Semantics

When using the Linux Kernel as a forwarding plane, routes are installed with a metric of 20 to the kernel. Please note that the kernel's metric value bears no resemblence to FRR's RIB metric or admin distance. It merely is a way for the Linux Kernel to decide which route to use if it has multiple routes for the same prefix from multiple sources. An example here would be if someone else was running another routing suite besides FRR at the same time, the kernel must choose what route to use to forward on. FRR choose the value of 20 because of two reasons. FRR wanted a value small enough to be choosen but large enough that the operator could allow route prioritization by the kernel when multiple routing suites are being run and FRR wanted to take advantage of Route Replace semantics that the linux kernel offers. In order for Route Replacement semantics to work FRR must use the same metric when issuing the replace command. Currently FRR only supports Route Replace semantics using the Linux Kernel.

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. Please note when using the Linux VRF routing table it is expected that a default Kernel route will be installed that has a metric as outlined in the www.kernel.org doc above. The Linux Kernel does table lookup via a combination of rule application of the rule table and then route lookup of the specified table. If no route match is found then the next applicable rule is applied to find the next route table to use to look for a route match. As such if your VRF table does not have a default blackhole route with a high metric VRF route lookup will leave the table specified by the VRF, which is undesirable.

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.

show ip route vrf VRF tables

This command will dump the routing tables within the vrf scope. If vrf all is executed, all routing tables will be dumped.

show <ip|ipv6> route summary [vrf VRF] [table TABLENO] [prefix]

This command will dump a summary output of the specified VRF and TABLENO combination. If neither VRF or TABLENO is specified FRR defaults to the default vrf and default table. If prefix is specified dump the number of prefix routes.

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 overridden 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

ECMP

FRR supports ECMP as part of normal operations and is generally compiled with a limit of 64 way ECMP. This of course can be modified via configure options on compilation if the end operator desires to do so. Individual protocols each have their own way of dictating ECMP policy and their respective documentation should be read.

ECMP can be inspected in zebra by doing a show ip route X command.

eva# show ip route 4.4.4.4/32
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, f - OpenFabric,
       > - selected route, * - FIB route, q - queued, r - rejected, b - backup
       t - trapped, o - offload failure

D>* 4.4.4.4/32 [150/0] via 192.168.161.1, enp39s0, weight 1, 00:00:02
  *                    via 192.168.161.2, enp39s0, weight 1, 00:00:02
  *                    via 192.168.161.3, enp39s0, weight 1, 00:00:02
  *                    via 192.168.161.4, enp39s0, weight 1, 00:00:02
  *                    via 192.168.161.5, enp39s0, weight 1, 00:00:02
  *                    via 192.168.161.6, enp39s0, weight 1, 00:00:02
  *                    via 192.168.161.7, enp39s0, weight 1, 00:00:02
  *                    via 192.168.161.8, enp39s0, weight 1, 00:00:02
  *                    via 192.168.161.9, enp39s0, weight 1, 00:00:02
  *                    via 192.168.161.10, enp39s0, weight 1, 00:00:02
  *                    via 192.168.161.11, enp39s0, weight 1, 00:00:02
  *                    via 192.168.161.12, enp39s0, weight 1, 00:00:02
  *                    via 192.168.161.13, enp39s0, weight 1, 00:00:02
  *                    via 192.168.161.14, enp39s0, weight 1, 00:00:02
  *                    via 192.168.161.15, enp39s0, weight 1, 00:00:02
  *                    via 192.168.161.16, enp39s0, weight 1, 00:00:02

In this example we have 16 way ecmp for the 4.4.4.4/32 route. The * character tells us that the route is installed in the Data Plane, or FIB.

If you are using the Linux kernel as a Data Plane, this can be inspected via a ip route show X command:

sharpd@eva ~/f/doc(ecmp_doc_change)> ip route show 4.4.4.4/32
4.4.4.4 nhid 185483868 proto sharp metric 20
   nexthop via 192.168.161.1 dev enp39s0 weight 1
   nexthop via 192.168.161.10 dev enp39s0 weight 1
   nexthop via 192.168.161.11 dev enp39s0 weight 1
   nexthop via 192.168.161.12 dev enp39s0 weight 1
   nexthop via 192.168.161.13 dev enp39s0 weight 1
   nexthop via 192.168.161.14 dev enp39s0 weight 1
   nexthop via 192.168.161.15 dev enp39s0 weight 1
   nexthop via 192.168.161.16 dev enp39s0 weight 1
   nexthop via 192.168.161.2 dev enp39s0 weight 1
   nexthop via 192.168.161.3 dev enp39s0 weight 1
   nexthop via 192.168.161.4 dev enp39s0 weight 1
   nexthop via 192.168.161.5 dev enp39s0 weight 1
   nexthop via 192.168.161.6 dev enp39s0 weight 1
   nexthop via 192.168.161.7 dev enp39s0 weight 1
   nexthop via 192.168.161.8 dev enp39s0 weight 1
   nexthop via 192.168.161.9 dev enp39s0 weight 1

Once installed into the FIB, FRR currently has little control over what nexthops are choosen to forward packets on. Currently the Linux kernel has a fib_multipath_hash_policy sysctl which dictates how the hashing algorithm is used to forward packets.

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:

ip route NETWORK MASK GATEWAY|INTERFACE label LABEL

NETWORK and 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.

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

Segment-Routing IPv6

Segment-Routing is source routing paradigm that allows network operator to encode network intent into the packets. SRv6 is an implementation of Segment-Routing with application of IPv6 and segment-routing-header.

All routing daemon can use the Segment-Routing base framework implemented on zebra to use SRv6 routing mechanism. In that case, user must configure initial srv6 setting on FRR's cli or frr.conf or zebra.conf. This section shows how to configure SRv6 on FRR. Of course SRv6 can be used as standalone, and this section also helps that case.

show segment-routing srv6 locator [json]

This command dump SRv6-locator configured on zebra. SRv6-locator is used to route to the node before performing the SRv6-function. and that works as aggregation of SRv6-function's IDs. Following console log shows two SRv6-locators loc1 and loc2. All locators are identified by unique IPv6 prefix. User can get that information as JSON string when json key word at the end of cli is presented.

router# sh segment-routing srv6 locator
Locator:
Name                 ID      Prefix                   Status
-------------------- ------- ------------------------ -------
loc1                       1 2001:db8:1:1::/64        Up
loc2                       2 2001:db8:2:2::/64        Up
show segment-routing srv6 locator NAME detail [json]

As shown in the example, by specifying the name of the locator, you can see the detailed information for each locator. Locator can be represented by a single IPv6 prefix, but SRv6 is designed to share this Locator among multiple Routing Protocols. For this purpose, zebra divides the IPv6 prefix block that makes the Locator unique into multiple chunks, and manages the ownership of each chunk.

For example, loc1 has system as its owner. For example, loc1 is owned by system, which means that it is not yet proprietary to any routing protocol. For example, loc2 has sharp as its owner. This means that the shaprd for function development holds the owner of the chunk of this locator, and no other routing protocol will use this area.

router# show segment-routing srv6 locator loc1 detail
Name: loc1
Prefix: 2001:db8:1:1::/64
Chunks:
- prefix: 2001:db8:1:1::/64, owner: system

router# show segment-routing srv6 locator loc2 detail
Name: loc2
Prefix: 2001:db8:2:2::/64
Chunks:
- prefix: 2001:db8:2:2::/64, owner: sharp
segment-routing

Move from configure mode to segment-routing node.

srv6

Move from segment-routing node to srv6 node.

locators

Move from srv6 node to locator node. In this locator node, user can configure detailed settings such as the actual srv6 locator.

locator NAME

Create a new locator. If the name of an existing locator is specified, move to specified locator's configuration node to change the settings it.

prefix X:X::X:X/M [function-bits-length 32]

Set the ipv6 prefix block of the locator. SRv6 locator is defined by RFC8986. The actual routing protocol specifies the locator and allocates a SID to be used by each routing protocol. This SID is included in the locator as an IPv6 prefix.

Following example console log shows the typical configuration of SRv6 data-plane. After a new SRv6 locator, named loc1, is created, loc1's prefix is configured as 2001:db8:1:1::/64. If user or some routing daemon allocates new SID on this locator, new SID will allocated in range of this prefix. For example, if some routing daemon creates new SID on locator (2001:db8:1:1::/64), Then new SID will be 2001:db8:1:1:7::/80, 2001:db8:1:1:8::/80, and so on. Each locator has default SID that is SRv6 local function "End". Usually default SID is allocated as PREFIX:1::. (PREFIX is locator's prefix) For example, if user configure the locator's prefix as 2001:db8:1:1::/64, then default SID will be 2001:db8:1:1:1::)

The function bits range is 16bits by default. If operator want to change function bits range, they can configure with function-bits-length option.

router# configure terminal
router(config)# segment-routinig
router(config-sr)# srv6
router(config-srv6)# locators
router(config-srv6-locs)# locator loc1
router(config-srv6-loc)# prefix 2001:db8:1:1::/64

router(config-srv6-loc)# show run
...
segment-routing
 srv6
  locators
   locator loc1
    prefix 2001:db8:1:1::/64
   !
...

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

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]

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,
  • babel,
  • bgp,
  • connected,
  • eigrp,
  • isis,
  • kernel,
  • nhrp,
  • openfabric,
  • ospf,
  • ospf6,
  • rip,
  • sharp,
  • static,
  • ripng,
  • table,
  • vnc.

If you choose any as the option that will cause all protocols that are sending routes to zebra. You can specify a ip protocol PROTOCOL route-map ROUTEMAP on a per vrf basis, by entering this command under vrf mode for the vrf you want to apply the route-map against.

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

IPv6 example for OSPFv3.

ipv6 prefix-list ANY seq 10 permit any
route-map RM6 permit 10
    match ipv6 address prefix-list ANY
    set src 2001:db8:425:1000::3

ipv6 protocol ospf6 route-map RM6

Note

For both IPv4 and IPv6, the IP address has to exist on some interface when the route is getting installed into the system. Otherwise, kernel rejects the route. To solve the problem of disappearing IPv6 addresses when the interface goes down, use net.ipv6.conf.all.keep_addr_on_down sysctl option.

zebra route-map delay-timer (0-600)

Set the delay before any route-maps are processed in zebra. The default time for this is 5 seconds.

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 relevant zebra code kicks in when zebra is configured with the --enable-fpm flag and started with the module (-M fpm or -M dplane_fpm_nl).

Note

The fpm implementation attempts to connect to 127.0.0.1 port 2620 by default without configurations. The dplane_fpm_nl only attempts to connect to a server if configured.

Zebra periodically attempts to connect to the well-known FPM port (2620). 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 default FPM message format is netlink, however it can be controlled with the module load-time option. The modules accept the following options:

  • fpm: netlink and protobuf.
  • dplane_fpm_nl: none, it only implements netlink.

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.

For more details on the implementation, please read the developer's manual FPM section.

FPM Commands

fpm implementation

fpm connection ip A.B.C.D port (1-65535)

Configure zebra to connect to a different FPM server than the default of 127.0.0.1:2060

show zebra fpm stats

Shows the FPM statistics.

Sample output:

Counter                                       Total     Last 10 secs

connect_calls                                     3                2
connect_no_sock                                   0                0
read_cb_calls                                     2                2
write_cb_calls                                    2                0
write_calls                                       1                0
partial_writes                                    0                0
max_writes_hit                                    0                0
t_write_yields                                    0                0
nop_deletes_skipped                               6                0
route_adds                                        5                0
route_dels                                        0                0
updates_triggered                                11                0
redundant_triggers                                0                0
dests_del_after_update                            0                0
t_conn_down_starts                                0                0
t_conn_down_dests_processed                       0                0
t_conn_down_yields                                0                0
t_conn_down_finishes                              0                0
t_conn_up_starts                                  1                0
t_conn_up_dests_processed                        11                0
t_conn_up_yields                                  0                0
t_conn_up_aborts                                  0                0
t_conn_up_finishes                                1                0
clear zebra fpm stats

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

dplane_fpm_nl implementation

fpm address <A.B.C.D|X:X::X:X> [port (1-65535)]

Configures the FPM server address. Once configured zebra will attempt to connect to it immediately.

The no form disables FPM entirely. zebra will close any current connections and will not attempt to connect to it anymore.

fpm use-next-hop-groups

Use the new netlink messages RTM_NEWNEXTHOP / RTM_DELNEXTHOP to group repeated route next hop information.

The no form uses the old known FPM behavior of including next hop information in the route (e.g. RTM_NEWROUTE) messages.

show fpm counters [json]

Show the FPM statistics (plain text or JSON formatted).

Sample output:

                 FPM counters
                 ============
                Input bytes: 0
               Output bytes: 308
 Output buffer current size: 0
    Output buffer peak size: 308
          Connection closes: 0
          Connection errors: 0
 Data plane items processed: 0
  Data plane items enqueued: 0
Data plane items queue peak: 0
           Buffer full hits: 0
    User FPM configurations: 1
  User FPM disable requests: 0
clear fpm counters

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

Dataplane Commands

The zebra dataplane subsystem provides a framework for FIB programming. Zebra uses the dataplane to program the local kernel as it makes changes to objects such as IP routes, MPLS LSPs, and interface IP addresses. The dataplane runs in its own pthread, in order to off-load work from the main zebra pthread.

show zebra dplane [detailed]

Display statistics about the updates and events passing through the dataplane subsystem.

show zebra dplane providers

Display information about the running dataplane plugins that are providing updates to a FIB. By default, the local kernel plugin is present.

zebra dplane limit [NUMBER]

Configure the limit on the number of pending updates that are waiting to be processed by the dataplane pthread.

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 [ip|ipv6] route [PREFIX] [nexthop-group]

Display detailed information about a route. If [nexthop-group] is included, it will display the nexthop group ID the route is using as well.

show interface [NAME] [{vrf VRF|brief}] [json]
show interface [NAME] [{vrf all|brief}] [json]
show interface [NAME] [{vrf VRF|brief}] [nexthop-group]
show interface [NAME] [{vrf all|brief}] [nexthop-group]

Display interface information. If no extra information is added, it will dump information on all interfaces. If [NAME] is specified, it will display detailed information about that single interface. If [nexthop-group] is specified, it will display nexthop groups pointing out that interface.

If the json option is specified, output is displayed in JSON format.

show ip prefix-list [NAME]
show route-map [NAME]
show ip protocol
show ip forward

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 ipv6 forward

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 client [summary]

Display statistics about clients that are connected to zebra. This is useful for debugging and seeing how much data is being passed between zebra and it's clients. If the summary form of the command is choosen a table is displayed with shortened information.

show zebra router table summary

Display summarized data about tables created, their afi/safi/tableid and how many routes each table contains. Please note this is the total number of route nodes in the table. Which will be higher than the actual number of routes that are held.

show nexthop-group rib [ID] [vrf NAME] [singleton [ip|ip6]] [type]

Display nexthop groups created by zebra. The [vrf NAME] option is only meaningful if you have started zebra with the --vrfwnetns option as that nexthop groups are per namespace in linux. If you specify singleton you would like to see the singleton nexthop groups that do have an afi. [type] allows you to filter those only coming from a specific NHG type (protocol).

show <ip|ipv6> zebra route dump [<vrf> VRFNAME]

It dumps all the routes from RIB with detailed information including internal flags, status etc. This is defined as a hidden command.

Router-id

Many routing protocols require a router-id to be configured. To have a consistent router-id across all daemons, the following commands are available to configure and display the router-id:

[ip] router-id A.B.C.D

Allow entering of the router-id. This command also works under the vrf subnode, to allow router-id's per vrf.

[ip] router-id A.B.C.D vrf NAME

Configure the router-id of this router from the configure NODE. A show run of this command will display the router-id command under the vrf sub node. This command is deprecated and will be removed at some point in time in the future.

show [ip] router-id [vrf NAME]

Display the user configured router-id.

For protocols requiring an IPv6 router-id, the following commands are available:

ipv6 router-id X:X::X:X

Configure the IPv6 router-id of this router. Like its IPv4 counterpart, this command works under the vrf subnode, to allow router-id's per vrf.

show ipv6 router-id [vrf NAME]

Display the user configured IPv6 router-id.

sysctl settings

The linux kernel has a variety of sysctl's that affect it's operation as a router. This section is meant to act as a starting point for those sysctl's that must be used in order to provide FRR with smooth operation as a router. This section is not meant as the full documentation for sysctl's. The operator must use the sysctl documentation with the linux kernel for that. The following link has helpful references to many relevant sysctl values: https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt

Expected sysctl settings

net.ipv4.ip_forward = 1

This global option allows the linux kernel to forward (route) ipv4 packets incoming from one interface to an outgoing interface. If this is set to 0, the system will not route transit ipv4 packets, i.e. packets that are not sent to/from a process running on the local system.

net.ipv4.conf.{all,default,<interface>}.forwarding = 1

The linux kernel can selectively enable forwarding (routing) of ipv4 packets on a per interface basis. The forwarding check in the kernel dataplane occurs against the ingress Layer 3 interface, i.e. if the ingress L3 interface has forwarding set to 0, packets will not be routed.

net.ipv6.conf.{all,default,<interface>}.forwarding = 1

This per interface option allows the linux kernel to forward (route) transit ipv6 packets i.e. incoming from one Layer 3 interface to an outgoing Layer 3 interface. The forwarding check in the kernel dataplane occurs against the ingress Layer 3 interface, i.e. if the ingress L3 interface has forwarding set to 0, packets will not be routed.

net.ipv6.conf.all.keep_addr_on_down = 1

When an interface is taken down, do not remove the v6 addresses associated with the interface. This option is recommended because this is the default behavior for v4 as well.

net.ipv6.route.skip_notify_on_dev_down = 1

When an interface is taken down, the linux kernel will not notify, via netlink, about routes that used that interface being removed from the FIB. This option is recommended because this is the default behavior for v4 as well.

Optional sysctl settings

net.ipv4.conf.{all,default,<interface>}.bc_forwarding = 0

This per interface option allows the linux kernel to optionally allow Directed Broadcast (i.e. Routed Broadcast or Subnet Broadcast) packets to be routed onto the connected network segment where the subnet exists. If the local router receives a routed packet destined for a broadcast address of a connected subnet, setting bc_forwarding to 1 on the interface with the target subnet assigned to it will allow non locally-generated packets to be routed via the broadcast route. If bc_forwarding is set to 0, routed packets destined for a broadcast route will be dropped. e.g. Host1 (SIP:192.0.2.10, DIP:10.0.0.255) -> (eth0:192.0.2.1/24) Router1 (eth1:10.0.0.1/24) -> BC If net.ipv4.conf.{all,default,<interface>}.bc_forwarding=1, then Router1 will forward each packet destined to 10.0.0.255 onto the eth1 interface with a broadcast DMAC (ff:ff:ff:ff:ff:ff).

net.ipv4.conf.{all,default,<interface>}.arp_accept = 1

This per interface option allows the linux kernel to optionally skip the creation of ARP entries upon the receipt of a Gratuitous ARP (GARP) frame carrying an IP that is not already present in the ARP cache. Setting arp_accept to 0 on an interface will ensure NEW ARP entries are not created due to the arrival of a GARP frame. Note: This does not impact how the kernel reacts to GARP frames that carry a "known" IP (that is already in the ARP cache) -- an existing ARP entry will always be updated when a GARP for that IP is received.

net.ipv4.conf.{all,default,<interface>}.arp_ignore = 0

This per interface option allows the linux kernel to control what conditions must be met in order for an ARP reply to be sent in response to an ARP request targeting a local IP address. When arp_ignore is set to 0, the kernel will send ARP replies in response to any ARP Request with a Target-IP matching a local address. When arp_ignore is set to 1, the kernel will send ARP replies if the Target-IP in the ARP Request matches an IP address on the interface the Request arrived at. When arp_ignore is set to 2, the kernel will send ARP replies only if the Target-IP matches an IP address on the interface where the Request arrived AND the Sender-IP falls within the subnet assigned to the local IP/interface.

net.ipv4.conf.{all,default,<interface>}.arp_notify = 1

This per interface option allows the linux kernel to decide whether to send a Gratuitious ARP (GARP) frame when the Layer 3 interface comes UP. When arp_notify is set to 0, no GARP is sent. When arp_notify is set to 1, a GARP is sent when the interface comes UP.

net.ipv6.conf.{all,default,<interface>}.ndisc_notify = 1

This per interface option allows the linux kernel to decide whether to send an Unsolicited Neighbor Advertisement (U-NA) frame when the Layer 3 interface comes UP. When ndisc_notify is set to 0, no U-NA is sent. When ndisc_notify is set to 1, a U-NA is sent when the interface comes UP.

Debugging

debug zebra mpls [detailed]

MPLS-related events and information.

debug zebra events

Zebra events

debug zebra nht [detailed]

Nexthop-tracking / reachability information

debug zebra vxlan

VxLAN (EVPN) events

debug zebra pseudowires

Pseudowire events.

debug zebra packet [<recv|send>] [detail]

ZAPI message and packet details

debug zebra kernel

Kernel / OS events.

debug zebra kernel msgdump [<recv|send>]

Raw OS (netlink) message details.

debug zebra rib [detailed]

RIB events.

debug zebra fpm

FPM (forwarding-plane manager) events.

debug zebra dplane [detailed]

Dataplane / FIB events.

debug zebra pbr

PBR (policy-based routing) events.

debug zebra mlag

MLAG events.

debug zebra evpn mh <es|mac|neigh|nh>

EVPN multi-hop events.

debug zebra nexthop [detail]

Nexthop and nexthop-group events.