-
Notifications
You must be signed in to change notification settings - Fork 1
Open
Description
In an a4x2 environment, the emulated Arista router is ignoring our LLDP packets.
LLDP is enabled on sidecar 0, which connects to Arista switch 1 on Ethernet port 2. We have LLDP-receive enabled:
cr1#show lldp ethernet 2
[...]
Port Tx Enabled Rx Enabled
---------- ---------------- ----------
Et2 Yes Yes
With tcpdump, we can see them arriving at the switch:
21:51:29.776896 LLDP, length 98
Chassis ID TLV (1), length 3
Subtype Chassis component (1): g0
Port ID TLV (2), length 8
Subtype Interface Name (5): qsfp0/0
Time to Live TLV (3), length 2: TTL 120s
System Name TLV (5), length 2: g0
System Description TLV (6), length 65
Oxide sled model: i86pc, Sidecar revision: softnpu_front_2_rear_4
System Capabilities TLV (7), length 4
System Capabilities [Router] (0x0010)
Enabled Capabilities [Router] (0x0010)
End TLV (0), length 0
[...]
For some reason, these packets aren't being accepted/processed by Eos. They don't even appear to get far enough to register as errors:
cr1#show lldp counters
[...]
Port Tx Frames Tx Length Exceeded
---------- --------------- ------------------
Et1 110 0
Et2 110 0
Et3 110 0
Port Rx Frames Rx Errors Rx Discard TLVs Discard TLVs Unknown
------- ------------ ------------- -------------- ---------------- ------------
Et1 0 0 0 0 0
Et2 0 0 0 0 0
Et3 0 0 0 0 0
Weirdly, a physical Arista switch is happy with LLDP packets from a real sidecar:
eveningstar#show lldp neighbors ethernet 26/1
Last table change time : 8:34:13 ago
Number of table inserts : 9
Number of table deletes : 6
Number of table drops : 0
Number of table age-outs : 3
Port Neighbor Device ID Neighbor Port ID TTL
------------ ------------------------ ---------------------- ---
Et26/1 BRM44220001 qsfp0/0 120
eveningstar#show lldp neighbors ethernet 26/1 detail
Interface Ethernet26/1 detected 1 LLDP neighbors:
Neighbor "BRM44220001"/"qsfp0/0", age 7 seconds
Discovered 8:34:21 ago; Last changed 8:34:21 ago
- Chassis ID type: Chassis component (1)
Chassis ID : "BRM44220001"
- Port ID type: Interface name(5)
A simple Venn diagram approach would suggest a problem in sidecar-lite and/or softnpu, but the tcpdump log shows that the packets are making it out of our emulated switch, which makes that less likely.
Metadata
Metadata
Assignees
Labels
No labels