ROCE_POLICY is the ingress classification policy. In the referenced Cisco Nexus example, it classifies packets matching CoS 3 and DSCP 26, then assigns them to QoS group 3:
policy-map type qos ROCE_POLICY
class ROCE_CLASS
set qos-group 3
A type-QoS policy assigns traffic to a QoS group, which subsequently selects the corresponding network-QoS and queuing classes.
ROCE_QUEUING_OUT is the egress scheduling policy:
policy-map type queuing ROCE_QUEUING_OUT
class type queuing c-out-q3
bandwidth remaining percent 70
class type queuing c-out-q-default
bandwidth remaining percent 30
That type-queuing policies control scheduling and buffering for QoS groups and are applied to egress interfaces. bandwidth remaining percent assigns a percentage of the remaining interface bandwidth, not necessarily 70% of total link capacity.
They are attached to the interface as follows:
service-policy type qos input ROCE_POLICY
service-policy type queuing output ROCE_QUEUING_OUT
qos input is a classification mode and queuing output as egress queuing mode.
These policies alone do not enable the lossless class. The system network-QoS policy must map the same QoS group to the required PFC CoS. The pause pfc-cos value should match the corresponding QoS-group value; PFC operates hop-by-hop and per CoS.
Therefore, if vSAN RDMA uses priority/CoS 3, validate the complete path: ingress classification into QoS group 3, PFC for CoS 3 in the network-QoS policy, the appropriate egress queue, and PFC on every relevant interface.
The exact queue names and supported QoS configuration depend on the Nexus model and NX-OS release, so the example should not be copied without checking the applicable Cisco configuration guide.
Original Message:
Sent: Sep 13, 2026 04:04 AM
From: Bab
Subject: Enabling IEEE mode for vSAN RDMA
What is ROCE_POLICY and ROCE_QUEUING_OUT here ?
Original Message:
Sent: Mar 13, 2024 03:35 PM
From: leo_lns
Subject: Enabling IEEE mode for vSAN RDMA
Your log output has: Unsupported team policy = 8 status = Success
It means you are trying to use teaming, which is not supported with RDMA. Inside DV switch vSan Distributed Port Group - teaming and failover, set one nic to active and one to standby. Set load balancing as "use explicit failover order"
If it's still "mode: 0 - Unknown" you should check your switch support for DCBX. We had this problem on a Cisco Nexus 9200. It only supports DCBX with nxos version 7.0.3.I7.10 or above.
It has worked for us with this configuration:
NXOS: version 7.0(3)I7(10)
interface Ethernet1/35
lldp dcbx version ieee
switchport mode trunk
switchport trunk allowed vlan 30
priority-flow-control mode on
priority-flow-control watch-dog-interval on
mtu 9216
service-policy type qos input ROCE_POLICY
service-policy type queuing output ROCE_QUEUING_OUT
no shutdown