Hi everyone,
I am working on vrf setup and have vlan interfaces configured
on eth1 and eth2. My goal is to direct specific vrf traffic through
these interfaces.
Currently i have two default routes on vrf id 120.(table-id)
0.0.0.0/0
unicast-ip4-chain
[@0]: dpo-load-balance: [proto:ip4 index:14 buckets:2 uRPF:140
to:[4:172]]
[0] [@5]: ipv4 via 192.168.22.1 eth1.20: mtu:1500 next:7
flags:[features ] 6805ca4ef4054cedfb917c98810000140800
[1] [@5]: ipv4 via 192.168.24.1 eth2.20: mtu:1500 next:8
flags:[features ] 54a0508870b34cedfb917c99810000140800
The issue I’m facing is that traffic is being load‑balanced across
sessions. Even though I assign a tx_sw_index and forward packets to
ip4-lookup, the lookup stage still performs load balancing, which
results in packets being sent out through different interfaces. For
example, if I select *eth1.10* in my custom node, the packet may still
be transmitted via *eth2.20*, causing asymmetric routing.
I also tried using priority‑based routing, but in that case the system
performs interface failover per session and always chooses the
highest‑priority path, which doesn’t solve the problem.
My question is: how can I avoid this load balancing behavior? I have
another custom node after ip4-lookup and before interface-output, So can
i modify adjacency and rewrite ethernet header in that custom node? Is
so, where I could modify the adjacency and explicitly assign the correct
interface for packet transmission? If so, how can I implement this?
Thanks in advance.
Thanks,
Dinesh
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#26789): https://lists.fd.io/g/vpp-dev/message/26789
Mute This Topic: https://lists.fd.io/mt/117654417/21656
Group Owner: [email protected]
Unsubscribe: https://lists.fd.io/g/vpp-dev/leave/14379924/21656/631435203/xyzzy
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-
