You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: content/cumulus-linux-517/Layer-2/Link-Layer-Discovery-Protocol.md
+8-1Lines changed: 8 additions & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -463,7 +463,13 @@ A profile overrides system defaults for an interface.
463
463
464
464
### BGP Unreachable Prefix TLV
465
465
466
-
The BGP unreachable prefix TLV allows LLDP to send prefix information learned from {{<linkurl="FRRouting/"text="FRRouting">}} to connected hosts. This is required in certain use cases such as EVPN unreachability signaling in disjoined planes. The following example enables the BGP unreachable prefix TLV globally:
466
+
The BGP unreachable prefix TLV allows LLDP to send prefix information learned from {{<linkurl="FRRouting/"text="FRRouting">}} to connected hosts. This is required in certain use cases such as EVPN unreachability signaling in disjoined planes.
467
+
468
+
{{%notice note%}}
469
+
BGP unreachable prefix TLV is a Beta feature.
470
+
{{%/notice%}}
471
+
472
+
The following example enables the BGP unreachable prefix TLV globally:
467
473
468
474
```
469
475
cumulus@leaf01:mgmt:~$ nv set system lldp tlv egress-policy unreachable-prefix state enabled
@@ -478,6 +484,7 @@ cumulus@leaf01:mgmt:~$ nv set system lldp tlv profile unreachable-prefix-only eg
478
484
cumulus@leaf01:mgmt:~$ nv set interface swp1-3 lldp tlv profile unreachable-prefix-only
479
485
cumulus@leaf01:mgmt:~$ nv config apply
480
486
```
487
+
481
488
{{%notice note%}}
482
489
You can only enable the unreachable prefix TLV in the `egress-policy` of a TLV profile; it is not supported on ingress.
To enable platform information statistics, such as the time of last reboot, the last reboot reason, or firmware version:
498
+
To enable platform information statistics, such as the time of last reboot, the last reboot reason, or hardware details such as the chasis version, model name, part number, serial number, and type:
499
499
500
500
```
501
501
cumulus@switch:~$ nv set system telemetry platform-stats class platform-info state enabled
@@ -3298,6 +3298,7 @@ CPU statistics include the CPU core number and operation mode (user, system, idl
3298
3298
| `nvswitch_platform_info_last_reboot_time` | Time of last reboot in ns since epoch.|
3299
3299
| `nvswitch_platform_info_last_reboot_reason` | Information about the last reboot reason of a component.|
3300
3300
| `nvswitch_platform_info_firmware_version` | Information about the firmware version of a component.|
3301
+
| `nvswitch_platform_info_hw_details` | *Component hardware details such as the version, model name, part number, serial number, and type.|
|`/components/component[name]/storage/state/counters/rotation-rate-rpm`| Disk rotation rate in RPMs (supported only on SATA disks). |
676
679
|`/components/component[name]/storage/state/counters/write-cache`| Indicates whether the disk has a write cache (supported only on SATA disks). |
677
680
|`/components/component[name]/storage/state/counters/write-cache-enabled`| Indicates whether the disk write cache is enabled. (supported only on SATA disks) |
Copy file name to clipboardExpand all lines: content/cumulus-linux-517/Network-Virtualization/Ethernet-Virtual-Private-Network-EVPN/EVPN-Enhancements.md
+44-4Lines changed: 44 additions & 4 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -232,19 +232,23 @@ address-family l2vpn evpn
232
232
233
233
## EVPN Unreachability in Disjoined Planes
234
234
235
-
In EVPN disjoined multi-plane topologies, each GPU in a cluster connects to multiple independent network planes. Leaf switches perform route aggregation per tenant VRF for scalability, but this suppresses visibility of individual host link failures. EVPN unreachability signaling in a tenant VRF sends a route to advertise host unreachability after a link failure, enabling leaf switches to send LLDP TLVs informing connected NICs to avoid unreachable paths.
235
+
In EVPN disjoined multi-plane topologies, each GPU in a cluster connects to multiple independent network planes. Leaf switches perform route aggregation per tenant VRF for scalability but this suppresses visibility of individual host link failures. EVPN unreachability signaling in a tenant VRF sends a route to advertise host unreachability after a link failure, enabling leaf switches to send LLDP TLVs informing connected NICs to avoid unreachable paths.
236
+
237
+
{{%notice note%}}
238
+
EVPN unreachability in disjoined planes is a Beta feature.
239
+
{{%/notice%}}
236
240
237
241
To enable EVPN unreachability signaling for disjoined multi-plane topologies on each leaf switch tenant VRF:
238
242
239
-
1. Configurate route aggregation in the tenant VRFs to summarize relevant network IPv4 and/or IPv6 prefixes
243
+
1. Configurate route aggregation in the tenant VRFs to summarize relevant network IPv4 and, or IPv6 prefixes:
2. Configure `bgp advertise-unreach interfaces-match <prefix>` under each IPv4 and/or IPv6 unreachability address-family to match interface prefixes attached to GPU NICs:
251
+
2. Configure `bgp advertise-unreach interfaces-match <prefix>` under each IPv4 and, or IPv6 unreachability address-family to match interface prefixes attached to GPU NICs:
5. Enable the LLDP {{<linkurl="Link-Layer-Discovery-Protocol/#bgp-unreachable-prefix-tlv"text="BGP unreachable prefix TLV">}} to distribute unreachable prefix information to connected hosts.
272
276
277
+
### Show EVPN Unreachability Information
278
+
279
+
To show EVPN unreachability signaling information, run the `nv show vrf <vrf-id> router bgp address-family l2vpn-evpn route` command:
280
+
281
+
```
282
+
cumulus@leaf01:mgmt:~$ nv show vrf TENANT1 router bgp address-family l2vpn-evpn route
283
+
```
284
+
285
+
To show EVPN unreachability signaling for a specific IPv4 route, run the `nv show vrf <vrf-id> router bgp address-family ipv4-unreachability <route>` command:
286
+
287
+
```
288
+
cumulus@leaf01:mgmt:~$ nv show vrf TENANT1 router bgp address-family ipv4-unreachability 10.1.0.0/16
289
+
```
290
+
291
+
To show EVPN unreachability signaling for a specific IPv6 route, run the `nv show vrf <vrf-id> router bgp address-family ipv6-unreachability <route>` command:
292
+
293
+
```
294
+
cumulus@leaf01:mgmt:~$ nv show vrf TENANT1 router bgp address-family ipv6-unreachability 2001:db8::/64
295
+
```
296
+
297
+
### Known Limitations
298
+
299
+
Be aware of the following limitations when configuring EVPN unreachability in disjoined planes.
300
+
301
+
#### switchd Restart
302
+
303
+
Restarting `switchd` results in network churn. FRR sees interfaces going down, which triggers BGP to start advertising EVPN unreachable routes and brings down peering. BGP withdraws the EVPN unreachable routes and re-establishes BGP sessions after the interfaces are back up operationally.
304
+
305
+
#### Link Down with Dynamic VRF Assignment on 802.1X Interfaces
306
+
307
+
When a link with a configured IP address goes down operationally on the leaf switch on only one plane, BGP on the leaf advertises the EVPN unreachable route type for the IP prefix of the interface that is operationally down. However, because the interface is unassigned from the VRF, BGP withdraws the EVPN unreachable route type from the remote leaf switch. Because the remote node still has the aggregate summary route, the leaf switch where the link is down attracts traffic and blocks it instead of rerouting traffic through other healthy planes.
308
+
309
+
#### Redistributed Route Overlapping with the Interface Match Command
310
+
311
+
If an interface address redistributed by BGP falls within the IP address range configured in the `bgp advertise-unreach interfaces-match` command and the interface goes down operationally, BGP withdraws the type-5 route and advertise the EVPN unreachable route type for the interface IP address.
312
+
273
313
## Enable EVPN in an iBGP Environment with an OSPF Underlay
274
314
275
315
You can use EVPN with an {{<linkurl="Open-Shortest-Path-First-OSPF"text="OSPF">}} or static route underlay. This is a more complex configuration than using <spanclass="a-tooltip">[eBGP](##"external BGP")</span>. In this case, <spanclass="a-tooltip">[iBGP](##"internal BGP")</span> advertises EVPN routes directly between <spanclass="a-tooltip">[VTEPs](##"Virtual Tunnel End Points")</span> and the spines are unaware of EVPN or BGP.
- {{<linkurl="Installing-a-New-Cumulus-Linux-Image-with-ONIE/#show-secure-boot-details"text="NVUE command to show secure boot status and details">}}
33
33
- {{<linkurl="RDMA-over-Converged-Ethernet-RoCE/#verify-roce-configuration"text="nv show interface <interface-id> qos roce counters supports multiple interfaces, including ranges">}}
34
34
- {{<linkurl="NVUE-CLI/#configuration-commands"text="NVUE prevents configuration changes during long background operations">}}
35
35
- {{<linkurl="NVUE-CLI/#verify-configuration-before-applying"text="NVUE command">}} and {{<linkurl="NVUE-API/#verify-configuration-before-applying"text="NVUE API option">}} to verify configuration before applying
36
36
- {{<linkurl="System-Power-and-Switch-Reboot/#show-reboot"text="NVUE command to show warm boot status ">}}
37
37
- Telemetry
38
-
- New OTEL metrics: {{<linkurl="Open-Telemetry-Export/#interface-statistic-format"text="link debounce">}}, {{<linkurl="Open-Telemetry-Export/#interface-statistic-format"text="PHY link down">}}, and {{<linkurl="Open-Telemetry-Export/#control-plane-statistic-format"text="control plane">}}
39
-
- New gNMI metrics: {{<linkurl="gNMI-Streaming/#metrics"text="PHY link down, link debounce, control plane, and systemd service">}}
38
+
- New OTEL metrics: {{<linkurl="Open-Telemetry-Export/#interface-statistic-format"text="link debounce">}}, {{<linkurl="Open-Telemetry-Export/#interface-statistic-format"text="PHY link down">}}, {{<linkurl="Open-Telemetry-Export/#control-plane-statistic-format"text="control plane">}}, and {{<linkurl="Open-Telemetry-Export/#platform-statistic-format"text="platform information">}}
39
+
- New gNMI metrics: {{<linkurl="gNMI-Streaming/#metrics"text="PHY link down, link debounce, control plane, platform information, and systemd service">}}
0 commit comments